TBPN Live - Crémieux & Kian on Nucleus Genomics, Michael Kratsios on Genesis, Nvidia Responds to TPU Progress | Kian Sadeghi, Joe Weisenthal, Sebastian Siemiatkowski, David Faugno, Keller Rinaudo, Royce Branning

Starting point is 00:00:00 You're watching TBPN. Today is Tuesday, November 25th, 2025. We are live from the TBPN Ultramm, the Temple of Technology, the Fortress of Finance, the Capital of Capital. We had a fan, yesterday was Anthropic, Claude 4.5 day. We had a lot of fun talking to Shulte about that. You should go check it out.

Starting point is 00:00:22 We wrote a little write-up. I collaborated with Brandon and Tyler to kind of give our thoughts on the state of the AI race with regard to Open AI and Anthropic and what makes Anthropic special. The things that stuck out to me, I mean, the thing that went viral was just the fact that apparently Dario goes around Slack and writes essays every single day. And everyone was like, give me the essays, turn it into a book, paging striped press. We got to get striped press to turn into a book. Yeah, I mean, well, that, but I was also thinking of potentially a risk factor for them

Starting point is 00:00:54 of being like the the Dariofiles a disgruntled employee that saves them all and then leaks them all I imagine because he even when he's on Mike he's known to say some things that he at times

Starting point is 00:01:10 sort of people people I feel like people take him out of out of context a lot like he will say he's kind of the final boss if this doesn't yeah if this doesn't go well we could lose 50% of white collar work or entry level white color work and and people will be like anthropic state admission is to

Starting point is 00:01:30 destroy jobs take your father's job yeah yeah it's uh it's rough but uh ramp time is money save both easy use corporate cards bill payments accounting and a whole lot more all in one place um timeline was in turmoil over the weekend and yesterday we covered a little bit about the nucleus dust up on the timeline. Cremew will be coming on the show at 1145. And then... Fast followed. Keon, the CEO of Nucleus,

Starting point is 00:02:02 will be coming on the show at noon. So we will kind of have both sides. Then we have Joe Wisenthal, joining from Bloomberg. And then we have... Who else do we have today? Cratios. Krasios is coming on

Starting point is 00:02:15 to break down Project Genesis, which we're very excited about. Anyway, let's run through what other news stories were at the top of the timeline while we pulled those up. Let me tell you about Restream, one live stream, 30 plus destinations. If you want to multistream, go to

Starting point is 00:02:31 Restream.com. Oh, yes, the biggest news in tech in AI is that the Ilius Sutskiver, Dwarkesh Patel podcast has dropped. Do we have... Hit the timeline. Do we have the opening clip? Because the

Starting point is 00:02:47 opening clip is iconic. It's It's very funny. It's a bit of a hot mic moment for Ilya. And I think we should pull it up and play it because it has a fascinating, just insight into, it feels very like, oh, where this is the real Ilya. He's not even thinking that he's on camera,

Starting point is 00:03:07 and he gives his real feelings. So let's play this from the very start. It is crazy that all of this is real. Yeah. Don't you think so? Meaning what? Like all this AI stuff and all this base. The area, yeah, that it's happened, like, isn't it straight out of science fiction?

Starting point is 00:03:25 Yeah. Another thing that's crazy is, like, how normal the slow takeoff feels. The idea that we'd be investing 1% of GDP in AI. It hasn't even set up the cameras yet, there are a bigger deal, you know, where right now it just feels like... And we get used to things pretty fast, turns out, yeah. But also, it's kind of like it's abstract, like, what does it mean? What it means that you see it in the news. Yeah.

Starting point is 00:03:48 That such and such company announced such and such and such dollar marks. right that's that's all you see right it's not really felt in any other way so far now should we actually begin here i think this is an interesting discussion sure it's like one of the greatest podcast intro from the average point of view so good really good so good anyway we're not going to watch the whole show that's that's going to be a new meta yes yes i mean you you can't you can't fake that it's amazing also it's just funny because uh you know it's effectively getting caught on a hot mic. But I was joking. I was like, of all the things that you could say on the hot mic before you sit down, oh, okay, we're actually recording. His is just completely reaffirming everything

Starting point is 00:04:29 we know about Ilius at square. It's just completely the same. Like, okay, he is a true believer. It's not like he was sitting down and being like, like, North Cash. Like, we got to go on my private plane. I just sold so much secondary. It's crazy what's going on with this stuff. Like, if people really think this AI thing's going to pan out, I'm. making billions of dollars. I'm basing out. I don't believe any of this stuff is real. No, he wasn't caught on a hot mic like that. His hot mic moment is like, wow, it's exactly like science fiction. Everything is all real. It's all real. Yeah, which is just iconic. Well, you can go and listen to that in the Dorcasch Patel, RSS feed, and on the YouTube channel and on X. He put

Starting point is 00:05:10 the full thing up. It's 95 minutes long. Tyler, did you have any other takeaways from your speed run? You're listening to it at 5X, right? Well, on X you can only do up to 2X, so I was on that, so I still have like 10 minutes left. But yeah, a bunch of good stuff in here. Does he pop the scaling bubble? Does he give a bearish take about

Starting point is 00:05:28 NIA? Is it over? At any point? While you're thinking about that, let me tell you about Gemini 3 Pro. Google's most intelligent model yet. State of the art reasoning, next level vibe coding, and deep multimodal understanding. Continue. So I wouldn't say he's like anti-scaling, but he does kind of give this interesting take,

Starting point is 00:05:45 which he basically says, that like now AI companies are like there's too few ideas for the amount of companies and for the scale that we're at where he basically like you can think of AI progress as being in these kind of distinct ages right so he says 2012 to 2020 was like the age of research where you're trying all these like different ideas and the scale of things is very small right like to train the original AlexNet was like two GPUs to do the original transformer was like like eight, maybe 64, but like, you know, very small amount of PPUs.

Starting point is 00:06:23 And then once we kind of figured out that Transformers work, we enter this age of scaling. And that's basically from 2020 to 2025. And now we're basically at this point where, like, yes, you can keep scaling and models will get better. But even if you scale 100x, like, are we really going to get super intelligence? It'll get better on the benchmarks, and they'll become more useful. But it's not like this, he doesn't think that just raw scaling alone is basically what's going to bring us there.

Starting point is 00:06:50 I mean, this has been, I could buy a lot of people. This was, I think, Carpathie said this where we still need a couple different kind of paradigms for this to work. And even, like, this is even kind of what Schulte said yesterday, which is like, pre-training, it's not dead, but it's like, the reason that Opus 4.5 was better is not just because they scaled pre-training. scaling generally, but even then, like, the scaling has gone from pre-training, and now it's RL. Yeah. And so we basically, we need to find another paradigm. And the way you do that is just

Starting point is 00:07:26 doing, like, research. And so he talks about SSI as basically being this, like, all they do is research. Return to research. Yeah, it's small kind of training runs, even though, you know, they only raise $3 billion, which is like small compared to other research institutions. the fact that they're basically putting it all on these kind of I mean, I don't know if they're moonshots but they're these small training runs where they're doing experiments

Starting point is 00:07:51 and then they're going to scale it up eventually but they're not just basically trying to win the AI race by just scaling up and doing the same thing as everyone else. Yeah, yeah. They're trying to find a new a way to actually bend the scaling curve, find a new scaling law

Starting point is 00:08:06 or find a new technology that they can scale against. I was thinking about Ilya's talk at Nureps last year, he pulls up this chart of the relationship between the mammal's mass and the brain volume, and it's a pretty linear graph. And so, like, the elephant is a lot bigger than the mouse, and so it has a proportionally larger brain to its body volume. And it's this perfect, it's this perfect linear curve. I should just try and figure it out if I can maybe text it in

Starting point is 00:08:42 I took a picture of it because it's a very cool chart here it is. Where do I send this? The timeline? Let me see. Ship it. Share. Let me see. Timeline. Sorry. Boom.

Starting point is 00:08:59 So basically the mammals have this like very clear linear trend but then the non-human primates are a little bit higher up on the chart and they're just doing a little bit better but then hominids the actual humans have a different there's a very distinctly different curve and so there's this interesting like it was making me think like like maybe that's what we're supposed to see when we think about yeah this this it's like it's like when we say like straight lines on log graphs when we say

Starting point is 00:09:35 we are seeing scaling happen with the current architectures which line are we scaling against? Are we actually scaling on the human curve or are we waiting for divergence from that current scaling

Starting point is 00:09:51 law? Yeah, he has this good quote where scaling has taken all the air out of the room where like basically like we have more than enough compute to try these like different ideas. Yeah. But they're just all going straight into training the next big model using the next paradigm and maybe it's slightly different right you have a different way of doing RL or whatever but it's still

Starting point is 00:10:13 fundamentally the same thing right and he talks about maybe continual learning is really the better approach right we've been in this era of like having a pre-training thing for so long that we think of like AI as like you train this thing and then you release it and it's like done and RL's like a little bit different now because there's this idea post-training and you can kind of integrate different things into it. I also thought the interesting thing was with pre-training, you use the whole internet so you don't have to decide anything. You're just applying this algorithm to just all the data, all the compute, and there's no

Starting point is 00:10:44 decisions. But then with RL, you have to decide, okay, we're putting in these math equations and we're maybe not putting in something else because we're actually creating the data. And it's not just this simple thing. This is maybe why we see these kind of like models that are super well. They do super well in e-vails, but not so much. Yeah, somebody overfitting and... And the reason is because the data that we choose is not the correct data.

Starting point is 00:11:06 Yeah. Because researchers are basically being reward hacked maybe into, like, just solving for benchmarks. Yeah. It's interesting. It's interesting to hear this, like, the conclusion is we need another breakthrough and then simultaneously consensus be, like, but like we're definitely going to get that breakthrough in like the next decade. it's like it's hard I mean I feel like it's hard to predict I feel like it echoes a lot of even what Mike Newp has been saying like we need new idea

Starting point is 00:11:38 yeah totally saying this for for months but it's way it's way harder to predict the rate at which breakthroughs will arrive as opposed to like you can actually chart out okay the formation of capital the time it takes to build a data center how long it takes to you know

Starting point is 00:11:52 manufacture a bunch of GPUs rack them run the training round like that's much more predictable than like human came up with new algorithm, like that's sort of random. And he brings us up as the reason why you see companies doing this, because it's just, if you're raising money, it's so much easier to justify the raise by saying,

Starting point is 00:12:09 we're going to buy this data center and we're going to do this training run. It's going to cost exactly no as much. It's going to monetize this way. Yeah, and then the model will be this good, and then we can use it to monetize this way. Where if you're just saying, like, oh, yeah, we're just going to pay a bunch of, like, really smart researchers to do a bunch of research, and then they'll figure something out. You can't really be... Yeah, in some

Starting point is 00:12:25 ways it feels like SSI is set up for, like, somewhat of a mini AI winter or like at least riding the hype cycle down because it doesn't sound like he's sitting there being like we raise three billion and we're spending it in the next 12 months it's like 2.9 was that no not not not no no no that's the point is not it's like equity it's just sitting there it's like he can you clearly pull back I'm going to give each research researcher all these different teams like shots on goal I love it we're going to keep taking those shots until obviously he'd be able to raise like another 10 billion dollars whenever

Starting point is 00:12:59 he wants, especially if he has, like, a key breakthrough insight, and they can be first to scale that. Yeah. Well, let me tell you about cognition. They make Devin, the AI software engineer, crush your backlog with your personal AI engineering team. NVIDIA has posted... They hit the timeline. They hit the timeline. Break this down for me, George. They said, we're delighted by Google's success. They've made great advances in AI, and we continue to supply to Google. NVIDIA is a generation ahead of the... industry. It's the only platform that runs every AI model and does it everywhere. Computing is done. Invita offers greater performance, versatility and fungibility than A6, which are designed

Starting point is 00:13:43 for specific AI frameworks or functions. That is a crazy thing to post. Crazy, crazy, crazy, crazy thing to post. Sometimes we get stuff. I don't know, boys, but having the largest company in the world sending tweets to defend their main product is not very reassuring. Yeah, it's just an odd. I feel like this would be so much better delivered. I actually don't have that much of a problem with the actual text here. I just think this should be delivered by Jensen with some nuance in a conversational setting. It just hits a lot different when this is in at exactly 9 a.m., like clearly scheduled, clearly typed out in a document. You know, it's like it feels like a press release, which is just an odd. thing when it should be, you know, there should be an answer to a question which someone, Bobby Cosmic in the chat was saying like, oh, the mainstream media is just now picking up on the Gemini three story. And there's articles in the Wall Street Journal and other places saying like, oh, maybe Google's back. Like, you know, buy Google. Like, it's very exciting.

Starting point is 00:14:50 And so, Nvidia feels the need to respond to that. But it's a lot different when it's actually a response instead of just like, we're putting out a press release. Like, who knows why? Yeah. As opposed to like Jensen saying like, well, since you asked, you know, talk show host or news anchor or whoever he's top podcast host, whoever he's talking to, Dwarkash, you know, whoever he's talking to, maybe us, you'd love to have him. I can ask him that question. He can defend this here. Yeah. Well, the timing is, is, is, uh, seems important because they are coming under a huge amount of pressure right now. There's an article in Barron's this morning by, Tay Kim. The headline is not what Nvidia's comms teams would have liked it to be. The headline is

Starting point is 00:15:35 NVIDIA says it's not Enron in private memo refuting accounting questions. That's a crazy thing to say. Of course it's not Enron. Let me get it to the coverage. So Tay says a series of prominent stock sales and allegations of accounting irregularities have put Nvidia in the middle of a debate about the value of artificial intelligence and its related stocks. Now Nvidia is pushing back in a private seven-page memo sent by NVIDIA's investor relations team to Wall Street analysts over the weekend, the chipmaker directly addressed a dozen claims made by skeptical investors. NVIDIA's memo, which includes fonts in the company's trademark green color, begins by addressing

Starting point is 00:16:12 a social media post from Michael Burry last week, which criticized the company for stock-based comp, dilution in stock buybacks, Barry's bet against subprime mortgages before the 2008 financial crisis was depicted in the movie The Big Short, of course. NVIDIA repurchased $91 billion shares since 2018, not $112 billion. Mr. Burry appears to have incorrectly included RSUs, RSU taxes, employee equity grants should not be conflated with the performance of the repurchase program. NVIDIA said in the memo, employees benefiting from a rising share price does not indicate the original equity grants were excessive at the time of issuance.

Starting point is 00:16:49 That makes sense. Barron's reviewed the memo, which initially appeared in social media posts for the weekend and confirmed its authenticity. Burry told Barron's he disagrees with NVIDIA's response and stands by his analysis. He said he would discuss the topic of the company's stock-based comp and more details.

Starting point is 00:17:06 Burry is, of course, now over on Substack. He's charging $380 a year. And if you are a perma bearer, I can't. This is like Christmas coming early. NVIDIA didn't respond to Barron's for a request for comment, but they also responded to claims that the current situation is analogous to historical accounting frauds, Enron, WorldCom, and Lucent that featured vendor financing and SPVs.

Starting point is 00:17:34 NVIDIA does not resemble historical accounting frauds because NVIDIA's underlying business is economically sound. Our reporting is complete and transparent, and we care about a reputation for integrity. Unlike Enron, NVIDIA does not use special purpose entities to hide debt or inflate revenue. There's no market. it's like there's 25 examples of how this is not the same. Nvidia also addressed allegation that it's customers, large technology companies aren't properly

Starting point is 00:18:00 accounting for the economic value of Nvidia hardware. Some of the companies use, we've talked about this, use a six-year depreciation schedule for GPUs. Burry said he believes the useful lives of the chips are shorter than six years, meaning Nvidia's customers are inflating profits by spreading out deep depreciation costs

Starting point is 00:18:16 over a long period. And video's customers depreciate GPUs over 40 six years based on real-world longevity and utilization patterns. Older GPUs such as A-100s continue to run at high utilization and generate strong contribution margins, retaining meaningful economic value well beyond the two to three years claimed by some commentators. So again, under fire on the TPU front and from the Michael Burry camp. But again, I think their answers are totally valid. Matt over on X said, had a post here. He said the TPUs equal bad for NVIDIA take is up there with the dumbest, maybe worse than Deepseek, as it completely misses what

Starting point is 00:19:04 actually happened in the last six weeks. And I will remember who is who in the zoo, my view. One, demand for AI is bananas. No one can meet demand. Everyone is spending more. Google said just yesterday they have to double capacity every six months to keep up. Two, scaling laws are he's referencing Gemini 3. The flywheel is about to speed up. Somehow the mid-curve crew thinks this is zero-sum competition. None of this suggests that. If you think the race is hot now, wait until you see what comes out of large, coherent blackwell clusters. All the magic from the quote God machines is pretty much still hopper-based. Lastly, a quick GPU, less than the cost and performance specs on the box aren't what you get in real life. And Google is going to get a bat margin too. doubled up. What matters is system-level effective tokens to watt to dollars in TCO. InvidiGPUs have higher FMU because they are they're already embedded in workflows slash the ecosystem is massive. By the way, this is a good test. If you have an opinion on this topic, but you have to look up FMU, then perhaps curate better source. What?

Starting point is 00:20:11 MFU. MFU. What I said? FMU. Sorry, sorry. The above effective token watt gap also likely widens with Rubin. add in that Jensen can actually deliver volume in a tight market, plus future flexibility, multi-cloud capable, programmable for paradigm shifts,

Starting point is 00:20:28 and he'll sell every GPU he makes for years. Google will too, since everyone wants a second supplier and TPU is a fantastic chip, but this is as far from either or as it gets. The one benefit of this confusion is that it is likely to give Google a brief stint as the world heavyweight champion, the most valuable company. I would guess the midwits put the strap on them, in less than two weeks. So he...

Starting point is 00:20:51 Put the strap on them? What does that mean? Just like... Like pile in? Is he saying just like... So he... It seems like he's predicting that that people will overplay the Nvidia bear take

Starting point is 00:21:10 and overplay the Google's opportunity. And that will result in Google becoming the most valuable company in the world. and he uses the phrase put the strap on them in multiple in less than two weeks. Interesting post. In other news, David Sachs has hit the timeline. He says, according to today's Wall Street Journal, AI-related investment accounts for half of GDP growth or reversal would-risk recession.

Starting point is 00:21:38 We can't afford to go backwards. The article is how the U.S. economy became hooked on AI spending. And we will be chatting with Kratzios in about an hour on this very topic. So we can get into a little bit. Before we move on, let me tell you about Adio. AI Native CRM. Adio builds skills and grows your company to the next level. Fact sheet from the White House, President Donald J. Trump unveils the Genesis mission

Starting point is 00:22:13 to accelerate AI for scientific discovery today, and this is yesterday. Today, Trump signed an executive order launching the Genesis mission and new national effort to use artificial intelligence to transform how scientific research is conducted and accelerate the speed of scientific discovery. The Genesis mission charges the Secretary of Energy with leveraging our national laboratories to unite America's brightest minds, most powerful computers, and vast scientific data

Starting point is 00:22:39 into one cooperative system for research. The order directs the Department of Energy to create a closed-loop AI experimentation platform that integrates our nation's world-class supercomputers and unique data sets to generate scientific foundation models and power robotic laboratories. The order instructs the assistant to the President for Science and Technology to coordinate the National Initiative and integrate an integration of data and infrastructure from across the federal government. The Secretary of Energy APST and the Special Advisor for AI and Crypta will collaborate with academia and private sector

Starting point is 00:23:12 innovators to support and enhance the Genesis mission. Priority areas of focus include the greatest scientific challenges of our time that can dramatically improve our nation's national economic and health security, including biotechnology, critical minerals, nuclear fission, and fusion energy, space exploration, quantum information science, and semiconductors and microelectronics. Next, harnessing AI for our national security and economic development with the Genesis mission, the Trump administration intends to dramatically expand the productivity and impact of federal research and development within a decade. So there's one more note here on strengthening America's AI dominance. Trump continues to prioritize America's global dominance and AI to usher in a new golden age of human flourishing,

Starting point is 00:24:01 economic competitiveness and national security. And so we will get into more of this with Cratsios. Yeah, I'm very interested to hear how, like, how the public-private partnership actually works here. There was a time when every, basically, every cool technology was coming out of DARPA, coming out of the U.S. government. The U.S. government landed on the moon. And since then, you know, I think a lot of people in technology have lost faith in the U.S. government overseeing the development of technology. Even academia, I mean, people think, like, you know, AGI will emerge from a private C-Corp. Like, that's where people believe that the best work will be done with, you know, give Ilya Sutskever, give the best scientist $3 billion, let them go cook. Like, that's the thesis

Starting point is 00:24:54 currently in tech. This feels like somewhat of a rejection of that in some ways. There's obviously lots of different places where having AI resources, having science and technology resources, within the government make a ton of sense. But it'll be interesting to see, like, where are the interfacing points between the two categories, the public and private sector? Because by default, I think most people in our audience, in technology, would say, hey, let's leave the space travel

Starting point is 00:25:29 and the AI research to the private sector. And this is, you know, potentially a different direction, potentially just very synergistic. So it'd be interesting to see where it breaks. Well, should we run through the Astral Codex 10 piece on trait-based embryo selection to tee up our discussion with Kermew and Keon from nucleus and go through that? So this is from Scott Alexander in Astral Codex 10. He says, suddenly trait-based embryo selection. When a couple uses, I, so in 2021, genomic prediction announced the first polygenically selected baby. When a couple uses IVF, they may get as many as 10 embryos.

Starting point is 00:26:13 If they want one child, which one do they implant? In the early days, doctors would just eyeball them and choose whichever looked to the healthiest. Later, they started testing for some of the most severe and easiest to detect genetic orders, like disorders like Down syndrome and cystic fibrosis. The final step was polygenic selection, genotyping. embryo and implanting the one with the best genes overall. Best in what sense? Genomic prediction claimed the ability to forecast health outcomes from diabetes to

Starting point is 00:26:41 schizophrenia. For example, although the average person has a 30% chance of getting type 2 diabetes, if you genetically test five embryos and select the one with the lowest predicted risk, they'll only have a 20% chance. So you get a 10% bump there. That's nice. Since you're taking the healthiest of many embryos, you should expect a child conceived via this method to be significantly healthier

Starting point is 00:27:01 than one born naturally. Polygenic selection straddles the line between disease prevention and human enhancement. In 2023, Orchid Health, founded by Noor, who we've had on the show, enter the field. Unlike genomic prediction, which tested only the most important genetic variants, Orchid offers whole genome sequencing, which can detect the de novo mutations involved in autism, developmental disorders, and certain other genetic diseases. Critics accuse GP and Orchid of offering designer babies, but this is only true. In the weakest sense, customers couldn't design a baby for anything other than slightly lower risk of genetic disease. You're basically just selecting out of what you already got. They're not editing the genes. They're merely sequencing them

Starting point is 00:27:46 and then allowing you to select. These companies refused to offer selection on traits, the industry term for the really controversial stuff like height, IQ, or eye color. Still, these were trivial extensions of their technology, and everyone knew it was just a matter of time before someone took the plunge. Last month, a startup called Nucleus took the plunge. They had previously offered 23 and Me style genetic tests for adults. Now they announced a partnership with genomic prediction focusing on embryos, although GP would continue to only test for health outcomes. You could forward the raw data from GP to Nucleus, and Nucleus would predict extra traits, including height, BMI, eye color, hair color, ADHD, IQ, and even-handedness. And

Starting point is 00:28:28 it's worth noting that nucleus is now being sued by genomic prediction. Even though they have this partnership. I'm assuming the partnership is no longer. Yeah. Well, we can ask. Yeah. But I'm assuming it's no longer because one of GPs' co-founders left the company to join nucleus. Interesting.

Starting point is 00:28:50 And allegedly turned off all the security cameras. Is that metaphor? Or is that actually? The lawsuit alleges that he turned off all the security cameras on his left. That's not a metaphor for, like, you know, sharing a Google drive of PDFs. You literally mean... It's his last day at work. Okay.

Starting point is 00:29:10 And he was allegedly, like, rounding up. Okay, so he turns off the cameras, allegedly, and the implication is that maybe he was rummaging around, like, literally taking documents or something like that. That's at least what the timeline is accused. Okay, wow. That's wild. I did not know that that was a literal accusation. And then another part of it, apparently Nucleus,

Starting point is 00:29:37 people at Nucleus were emailing the former co-founder at his old email address, evidence of them violating the agreement that they had. So anyways, it's very, very, very messy. We can ask. There's like four or five companies involved in this. And all of them are controversial, because this is the most, I think, the most controversial, probably, like, category that you can be in. Yeah, it's certainly up there.

Starting point is 00:30:11 Health is already, like, one of the most controversial topics. Yeah, everyone has an opinion on it. Health influencer that you've gotten into. Totally. Various debacle. Yeah. And also there's just, like, the, there's just, it's so easy to throw. I mean, in the same way that people are throwing Enron.

Starting point is 00:30:27 at NVIDIA, like, it's so easy to throw Theranos at any biotech company that's not, you know, that's accused of anything. And also biotech, it's like, it's pretty hard to understand the underlying science. It's not, it's not a, it's not as popular as, okay, like does the website work? Does the business make money?

Starting point is 00:30:47 You know, what's the cash flow like? It's way more complicated. And so it does attract even more attention. So one of the other companies in the space is Heresite, and Astral Codex 10 continues here. They enter the space with the most impressive disease risk scores, yet an IQ predictor worth six to nine extra points and a series of challenges to competitors

Starting point is 00:31:10 whom they call out for insufficient scientific rigor. Their most scathing attack is on nucleus itself, accusing its predictions of being misleading and unreliable. Let's start with the science and then move on to the companies to see if we can litigate their dispute. in all theory, in theory, all of this should work. Polygenic embryos, polygenic embryo screening is a natural extension

Starting point is 00:31:31 of two well-validated technologies, genetic testing of embryos, and polygenic prediction of traits in adults. So genetic screening of embryos has been done for decades, usually to detect chromosomal abnormalities like Down syndrome or simple gene editing disorders like cystic fibrosis.

Starting point is 00:31:47 It's challenging. We've talked about this before. You need to take a very small number of cells, often only five to ten from a tiny protoplasenta that may not have many cells to spare and extract a readable amount of genetic material from this limited sample. But there are known solutions that mostly work. And so the companies that we're talking about today aren't necessarily doing like the fundamental lab equipment, development, building the machine, figuring out how to sequence data from the first. It's more about the analysis that happens on top of the results.

Starting point is 00:32:22 And the recommendations, which is probably, which I would say is the most controversial part of this. I don't know that any of them are recommending, hey, we think you should take, we think you should pick this baby. They're more just saying like, we think that according to the data, this baby might be taller than this one. But if you're giving somebody risk fact, if you're, if you're. Yeah, but that's not a recommendation. Like, if I tell you, this car is 700 horsepower and does zero to 60 in two seconds, and this one does 800 horsepower and does zero to 60 in 2.4 seconds. This one's faster in a straight line. This one's faster on the curves.

Starting point is 00:32:54 And then, like, you pick. Like, I didn't make a recommendation. I just told you the stats. Right? Yeah, but from when you look at these companies from what they're marketing to consumers of why you should care about the service. Sure. And then the way that they deliver the information, if they're advertising, we can effectively

Starting point is 00:33:15 advertising, we can help you have a smarter, healthier baby. And then they're saying, like, hey, we think this. direction is going to get you a higher IQ. I don't think it's a recommendation. It's not an explicit recommendation, but I think people are trusting the service to try to get them what was marketed to them. Yeah, people want the data, and they want the data to be accurate because they're going to make a decision based on that.

Starting point is 00:33:41 But, I mean, here, Scott Alexander actually gets into some of the complexity of the actual trade-offs because there are, so most traits are polygenic requiring information about thousands or tens of thousands of genes to predict. These are too complicated to understand fully at current levels of technology, but some studies have chipped away at the problem and gotten to a partial understanding. Often, this looks like being able to predict a few percent of the variance in the trait to determine whether someone's genetic risk is slightly higher or lower than average. And so some people might genuinely want to select on a single condition. For example, people with a strong family history of schizophrenia might want to minimize their

Starting point is 00:34:26 chance of their children getting the disease. For these people, reducing schizophrenia risk by 58% while keeping everything else constant sounds pretty good. Everyone else probably wants a genetically healthy embryo with low risk of all conditions. Exactly how this works depends on the customer's own value. Would they prefer an embryo with lower cancer risk to one that will have fewer heart attacks? Like, that's a trade-off that you have to pick. And the exact benefits will depend on how parents make that decision.

Starting point is 00:34:56 Genomic prediction and heresite, try to help by providing semi-objective measures of which embryo is overall healthiest according to different conditions, effects on longevity and patient-rated quality of life for genomic prediction. That's the embryo health score. This is, you know, that's close to a recommendation. I think you're getting close. Yeah. And Nucleus's subway campaign is have a healthier baby.

Starting point is 00:35:23 Yeah. Yeah. It's, it's, the, the marketing claims are a big, big piece of this. I think, I think the scientific claims are potentially just as important. But it's both, they're both understanding where the science actually is, both broadly and then also within the companies. And then how it's marketed, like all of that is important to get like a complete pick. sure what's going on here. So for Heresite, it's a polygenic longevity index. They don't give exact risk reduction numbers for each disease, saying that it depends too much on a couple's

Starting point is 00:35:59 specific family history, but say that most people gain one to four years of healthy life. When I test it on a set of 20 embryos, the healthiest gets an extra 1.66 years. And so how much would you pay to give your children an extra one to four years of healthy life? This is no longer a healthy a hypothetical question. Here are the costs. Genoic prediction is around $3,250. Orchid is around $12,500. Nucleus is around $9,249.249. And Herosite, $53,250. That is expensive compared to the rest, five times the price. Is it worth it? Well, if you're already doing IVF, the claimed risk reductions are accurate. You value your kids' health as much as your own. You have low, time discount rate. You're well off enough that these aren't extraordinary sums of money to

Starting point is 00:36:51 you. You're okay using expected utility calculations where a 50% chance of preventing X is half as good as fully preventing X. Then I'll go out on a limb and say, yeah, it's obviously it's worth it. Consider genomic prediction, which costs $3,500 for five embryos and claims to lower absolute risk of type 2 diabetes by 12%. That implies that not getting type 2 diabetes is worth $27,000. Ask anyone dealing with regular insulin injections, let alone limb amputations, whether it would be worth $27,000 to wave a magic wand and not have type 2 diabetes. It's not a hard question. And that's just one of a dozen conditions you can lower the risk for.

Starting point is 00:37:28 Other ones, like not getting breast cancer, might be so valuable that it's hard to even attach numbers. So what about IQ? Six extra IQ points, which is Harrisite's estimate with five embryos, is about a quarter of the gap between the average person and the average Ivy League student. The benefits of intelligence are hard to quantify, but it's been shown to have probably causal positive effects on income, mortality, and achievement. Probably the income effects alone make up for the cost of the intervention. Again, assuming total parent-child altruism and a low discount rate. So if we accept

Starting point is 00:38:04 all of these claims and assumptions, the choice seems obvious. It probably even accounts, it's probably even obvious for governments to pay for all citizens to get. these given how much they'd save on health care costs, says Scott Alexander. But in practice, it's complicated. Critics have raised both scientific and ethical objections to polygenic embryo screening. Most significantly, it's been condemned by various bodies, including the Society for Psychiatric genetics, the European Society of Human Genetics, and the Behavioral Genetic Society. Their statements are not good. They tend towards vague language about how people are more than just their genes or how no genetic tests can be perfect or how embryo screening is not exactly the same

Starting point is 00:38:45 thing as some other form of screening, which has a longer history and more proponents. Although, quote, although in general, higher scores mean you are more likely to have a condition. Many healthy people will have higher scores. Other might other people's, others might develop the condition with even with a low score, says the Society for Psychiatric Genetics, as if they have just blown the lid off of some dastardly conspiracy. Screening embryos for psychiatric conditions may increase stigma surrounding those diseases. They continue. An objection which, taken seriously, could be used to ban every form of medical treatment. Because if you take care of something, you remove them from the population that might increase the

Starting point is 00:39:21 stigma, but we should still treat these. So he says, we will mostly ignore these people and try to imagine the implications of the objections that mildly competent critics might raise, some of which will coincidentally overlap with the content of the non-hypothetical statements. So the big question he wants to answer is scientific objection, the scientific objection around efficacy, does this, are we sure where this works at all? Are we sure this works? So a typical polygenic score is created by collecting thousands or millions of adult genomes, then matching genetic information with surveys about who has the trait slash condition of interest. Reputable studies then test these scores on holdout samples, adults who are not used to make the score, to see if

Starting point is 00:40:06 they still accurately predict who has the trait slash condition. Polygenic embryo selection depends on an assumption that the scores which work in these kinds of retrospective tests will also work on prospectively on embryos. This assumption hasn't been formally proven in studies, which would require years or decades to conduct, but seems commonsensical. The strongest challenge to the application of polygenic scores for embryo selection comes from a recent body of research showing that most scores combined. causal genetic effects with population stratification and therefore can be expected to lose

Starting point is 00:40:42 much of their predictive power when comparing two members of the same family. There is an increasing agreement in the field that unless scores are validated within families, headline results like decreases risk of X by Y percent will be large over estimates. When I talked to company representatives, they all said that they took accuracy extremely seriously and had various white papers and journal articles where anyone could double click, could double check on their methodology. But I attended an industry conference a few months ago, and the gossip level was comparable to a high school cafeteria. It says minus the sex rumors. Most of the attendees were having their kids via IVF. Everyone had some story about someone being careless or fudging their numbers. Some of the

Starting point is 00:41:29 conflicts broke out into the open on Wednesday when Heresite left stealth and published a white paper and associated blog post. They criticized genomic prediction for reporting between family rather than within family results and Orchid for smuggling a term for age into their Alzheimer's predictor. Unsurprisingly, this makes it work better. We'll get to their accusations against nucleus below. Note that this was recent enough that competitors haven't had time to err their own criticisms of Harrisite if this happens. I'll try and keep you updated. And to be clear, this article is from around

Starting point is 00:42:08 five months ago. Yes. And since that time, nucleus has been accused of plagiarizing the paper from Harris. From Harris. From Harrisite. Yeah. And then also accused of stealing IP from genomic prediction. So there's, again, a bunch of different accusations. We'll let Keon. So yeah, yeah. I mean, The goal here is to just give an opportunity for, you know, Cremew and Keon to answer some questions, try and contextualize it, try and make their case to a broader audience. They're, you know, I've read through as much as I could, I can, but without actually getting in the lab and rolling at my sleeves.

Starting point is 00:42:54 I don't think I could come to a firm conclusion here, but I can certainly talk to them on this show and hopefully get some more information that the community can do with what they will. So, Scott Alexander concludes this section talking about his strongest opinion of the scientific criticism. He says, authorities on all sides have cited Alex Young as an authority on how polygenic scores can be confounding or misleading. Last week, Alex Young revealed that he had been working with Heresite while it was in stealth

Starting point is 00:43:26 mode and endorses their research. Three, L.O.L. probably that means harrisite's products are okay that serves as proof of concept that this technology can work and means other company's claims are at least plausible so lots of back um back and forth and we will be joined by with by cremew in just a few minutes i actually need to message him and make sure that he has the information is there anything else that you think would be worthwhile to discuss before we hop on let's yeah and i can i can just go through i mean the original accusations came from an account called sishuan mala sishuan mala yes uh who wrote an extremely

Starting point is 00:44:12 lengthy blog post on uh on a bunch of the issues that they felt they had found uh with with nucleus nucleus ended up firing back and saying that Sichuan Mala was, or sort of implying that Sichuan Mala was funded by a competitor or competitive service, as well as making those allegations with Kramu. Yes. They go into issues around potentially fictitious customer reviews, which will ask you on about AI generated blog posts, accusations of intellectual property theft, saying that the nucleus origin, white paper is plagiarized,

Starting point is 00:44:56 saying it that has a bunch of errors. Nucleus has responded already to a lot of this stuff. Well, our first guest to the show is here. Let me tell you about linear. Meet the system for modern software development, purpose-built tool for planning and building products. We will bring in Kermu from the Restream waiting room into the TVP and Ultradome and have him set the table for us.

Starting point is 00:45:19 Kermu, how are you doing? to the stream. How are we doing? Glad to be here, guys. Thanks so much. Good, as always. Looking good. By the way, I can go face docs if you want. Let's do it. Let's do it. We can show his actual video this time, which is great. We've had him, welcome to the show. Hey, hey. Good to see you. How you doing? What actually kicked this off for you? Do you know Sishwan Mala? Separately, independently? Did you know that this was coming? Set the table for us, like, why did Nucleus come to the top of your mind? So, can I actually go back to Hereticon?

Starting point is 00:45:58 Please. With this? Yeah. All right. So for about a year, we've told Nucleus about issues with their products. Can I just actually give a big, like, I can model log on this for a minute. I can tell you a lot of the details that you want to go into it. Go right.

Starting point is 00:46:13 Okay. So one of the early things that really peeved a lot of us who are aware of how this tech works, is that nucleus claim to provide parents with information about rare variants based on microarray files. Their website's wording is incredibly ambiguous. So the excuse when I raised this to key in in person was that they were referring to imputing a child's embryo or a child or an embryo's microarray based data with parental whole genome data. But this is not sufficient for rare variants like they claim it is. And it only works out in very well, well and narrow, well-behaved cases. It's not a clean substitute for sequencing an embryo or child.

Starting point is 00:46:51 The rare variant information they can offer is limited, and their claim is highly misleading because it sounds like they can achieve coverage of de novo rare variants and ultra-rare variants reliably for children and embryos, but they cannot do this because of crossover that happens during recombination inside the haplotype blocks you're using through the imputations and because of mutations.

Starting point is 00:47:10 They can even get high-confidence coverage of rare variants more generally, which is what their big claim is about in specific wording, with the imputation-based methods they claim to be using. So everything they say has a huge error bar on it and they shouldn't be advertising it, basically. It's something incredibly misleading.

Starting point is 00:47:27 So just to like actually zoom out, like, what do you think they are capable of doing? I think they are capable of microarray sequencing. I think they're capable of... I mean, I mean in terms that you could advertise to a parent. Like, if Keon went to a prospective parent, the parent says, I'm doing IVF. What can you do for me?

Starting point is 00:47:53 Is this category even ready for large scale, out of home advertising? Do you think, do you think? It absolutely is. The issue is that nucleus should not be doing it, because nucleus has produced scores that are invalid. They're incredibly invalid. Like, for example, they used an ADHD score that included 12 single nucleotide polymorphisms. They claimed it explained 4% of the variants in ADHD, which means it's a pretty good predictor, and you can use it to get some improvements on the margins. But 12 SNPs means that there's just no way.

Starting point is 00:48:25 They could explain less than 1% of the variance, and the best current ADHD polygenic score uses more than a million SNPs and explains about 1% of the variance. So they just basically lied. They made up these numbers, and Harrison had to go through and show that, oh, actually a bunch of their numbers are made up. Yeah, but I mean, how do you know for sure that they're, I mean, like that claim, you know, the state of the art was, was a million for one percent. Now it's 12. Nuclase is claiming 12 for 4 percent. That seems like a huge exponential, you know, growth in efficacy of that particular test. But just an exponential progress, it is not necessarily evidence of malfeasance, right? This isn't progress. They've never been able to show that they can use 12 snips, and no one can. It's impossible. They don't explain this much variance.

Starting point is 00:49:21 It's literally physically impossible. It's not possible in any way. They should never have claimed it. They should never provide score reports to people based on it, and they did. They provided customers with score reports that have to be fake. Not fake, per se, but they have to be incorrect. They have to be wrong. There's no way they could stand up to scrutiny.

Starting point is 00:49:37 What I'm saying is that after they made this 12 snips to 4. percent variance claim. People have looked and they've shown the latest ADHD polygienic score, which is more recent than what they've claimed to have on offer, uses more than a million snips and explains about 1% of the variance. It's just, that's the state of the art. They are claiming to have better than the state of the art with completely implausible parameters. It is impossible. Okay. So back to the original question. Like, what do you think they actually could offer, even if they say, hey, you know what? We're not. We're not necessarily state of the art.

Starting point is 00:50:13 We're partnering with a lot of different labs. We're standing on the shoulders of giants. We're using the tools that are available. What is a reasonable claim that if they made it, you would be like, yeah, that sounds reasonable. A reasonable claim from Nucleus is that they are pulling polygenic scores from the polygenic score catalog, a publicly available resource that lists a bunch of different polygenic scores

Starting point is 00:50:35 from different genome-wide association studies. That would be reasonable. But that is not a, why would you go? with them then. They have nothing unique to offer. What they have offered that seems to be unique are claims that don't hold up to scrutiny or which are clearly plagiarized from one of their competitors. Yeah. Well, I mean, there are plenty of, there are plenty of services that are like, you know, in, you know, effectively wrappers around Quest Labs. And, you know, I get a better UI, and it tells me that my cholesterol is in the range. And yeah, I know it's just looking up the

Starting point is 00:51:05 range from the data, but it has a nice UI and a reasonable billing system or whatever. And, people pay for that and they get, you know, they make decisions about their diet based on that. And I don't think that there's anything necessarily wrong about providing something that's a commodity or not a scientific breakthrough. As long as you're up to, as long as you're honest about what you're doing and you're not inflating the results. Yeah. The problem is that they're inflating the results. Okay. The problem is that they are effectively making up the results. They have actually, they've, in their latest report, they have claimed to basically match Harrison, not claimed it directly. They've copied their very unique citations, which no one else

Starting point is 00:51:44 has made. They've copied their method. And copying their method is incredibly weird. You mentioned there that Scott Alexander noted that Alex Young is seen as a big authority in this space. And it's true. Alex is seen as like the go-to guy. If you want to learn about family-based sampling, trio imputation, if you want to learn about within family validations or imputations or quality control, you ask Alex. That is. the thing to do. And they apparently copied Alex Young's quality control pipeline. And it's very unique. So doing this is unusual. It's unheard of. It's not likely to have actually been done. It is like saying that you woke up one day and here's my morning routine. I woke up and today I was

Starting point is 00:52:27 John Kugan. And I brushed my teeth exactly three times each time around and I went and prayed to my household God. And I did all sorts of things. It's just, it's incredibly specific in a way that is very unlikely to be real. They've very likely just kind of mimicked exactly what they said. And they said they did it on additional data, but their results came up very close. Yeah, yeah. Which is not likely.

Starting point is 00:52:48 We have strong theoretical genetic reasons to expect that if they have this additional data that Heresite did not, the results would have actually looked different. So they have all these signals that they didn't do the analyses they said they did and lots of indications that they plagiarized. And when Sichuan Mala called them out, they responded. The response they generated was amateurish. It was kind of embarrassing because they admitted simultaneously to denying they did the plagiarism. They admitted to doing the

Starting point is 00:53:19 plagiarism. They admitted they copied things directly from Heresite. They admitted they used resources that are unusual to use and they effectively got the data from them. And they did nothing that was really unique. But they did ate everything that committed. So copying might be, you know, look down upon in tech people copy stuff all the time the stories was copied into instagram from from uh snapchat different different machine learning architectures are being copied constantly some stuff can be patented some stuff can be trade secrets uh are we looking at anything that's that's that goes beyond just like yeah it's kind of bad form to copy or or is this something that's actually like a problem beyond that yeah would you would you be less

Starting point is 00:54:05 angry if they copied it perfectly instead of like sort of copied it directionally? They copied a lot of it perfectly. And this leads to weird results because they should not have done that given the cohorts they used. They used separate data for their validations. So the results should have been different. They copied details that made it apparent that they were not using the data they claimed to or they were just fudging the results. It has to be one or the other. There's no way they could have gotten the results they did with the data they did and the copying they did. The copying tells us that they lied about something somewhere along the way. There is something fishy here. And we don't know what the exact error is. We just know they have

Starting point is 00:54:44 made an error because there's no way that the additional data they had access to was identical to all the data that their competitors used. It would have delivered different results. So I wanted to ask, they released open weight models, the origin models. Part of their Well, part of their pushback against Sichuan Mala's critique is that anybody can just download the models and play around with them. Have you downloaded them? So you have to ask for access and they do not give out access. We've had multiple people go in and try and ask for access and they've not perceived it at all.

Starting point is 00:55:20 And one person who asked for more information was told, stop contacting us. So no, they are not open at all. They're open in the sense that open AI is open. They're not very open. Okay. Got it. Shifting gears to the marketing claims, what stuck out to you there is particularly in need of addressing? They very much need to address the fact that they seem to have fake reviews.

Starting point is 00:55:49 So when they started nucleus embryo, they launched it in June. They weren't offering any sort of embryo screening services beforehand. And if they were, then it would have been, how. They would have, they have to specify what lab they use for all this stuff. It's, there's a lot of details that should go into this that they can't actually specify because they didn't do that. So they claim to have had customers that have already been served by this. Well, as everybody knows, it takes about nine months to serve a customer in this at minimum. Yes.

Starting point is 00:56:15 And it is not been nine months since June. You know, it takes nine months in one month. You know, it takes nine months. So I'm, so I'd love to see these three month old babies that came out perfectly and were, uh, made their customers so happy, but I don't think they exist. I think they're not real. So why do they have these reviews? I don't know. And the reviews are also, they have a lot of fake elements. There are some that are clearly fake. So they use stock images in these reviews to show their customers. Which to be clear, you can imagine a scenario where they use stock

Starting point is 00:56:46 imagery and fake names, and they put an asterisk and say, due to HIPAA compliance reasons, we're not just publicly displaying, you know, the names of any of our clients, but... There's no issue revealing this stuff, and other companies do actually reveal the real customers. So Orchid has revealed real customers. I've been introduced. I've met Harrison customers. Oh, yeah. Didn't Jason Carman do a whole video with Norr in the first, in the first Orchid baby? And this has been, like, basically making a documentary about that person's journey. Like, there's no, like...

Starting point is 00:57:18 Yeah. And I feel like I see in drug commercials all the time. it'll be like, this is a real customer who loves this hair loss meds, and they're like, yeah, it looks great. Yeah, but there's regulations. It seems much harder to get, it's probably much harder to get somebody to opt into that than just opt in to generally providing

Starting point is 00:57:33 sure. Clearly, clearly it's possible. Is that, is that legit, you think? No, they still could not have had the babies in time. It doesn't fit with the time. The chronology doesn't work here. These customer reviews are not really physically possible, and I'd like to see an explanation from them because

Starting point is 00:57:49 it doesn't make any real sense to me. They could be, I don't know, making some sort of representative review that they've maybe hedged in some fine print on some page somewhere, but I haven't seen it and their entire site has been archive now. So if that page exists, they'll have to show it to us on the archive. One thing Jordy and I were debating was this big question of like, is nucleus making a recommendation or not? I don't know if this is relevant, but I would love your take on this, this idea of like, you know, I see a, see an ad that just says like, high. is 80% DNA-based or genetically inheritable, and then you go into the dashboard and it just says, here's the predicted height and here's the predicted IQ. And then you make the decision. And it's not necessarily they're recommending one or the other. It's more of a diagnostic and you can do

Starting point is 00:58:39 that information what you want. Does that absolve them of some sort of responsibility there? No, it doesn't. So if you still provide wildly and accurate scores, then it doesn't matter what you're recommending. You are effectively just recommending something that doesn't matter. I mean, you are giving them incorrect recommendations. You have to give some range of uncertainty within the best of your ability. You have to give them something that is to the greatest extent knowable, reliable. And they can't have done that. They've provided scores that they know, they must know are incorrect. You can't explain 4% of the variance with 12 snips. It's not really feasible. There's no possibility of it, really. There's also no possibility of getting high

Starting point is 00:59:20 confidence coverage on rare variants to make from the imputation methods they've described. So they can't make parents certain about this stuff. Like they're saying lots of things that are, that would require them to basically advance the science 20 years. They would have to be leaders in innovations here. And I just kind of doubt it. The people who are actually leading on the innovations here are Heresite. Orchid is doing so as well.

Starting point is 00:59:42 They're doing a lot of orchid's whole genome sequencing of embryos is the only one of the industry available to do this. And Heresite's innovations that they've made. this stuff low cost for what they have admitted is a reduction in quality if you get PGTA-based imputation. Now, another thing I do want to mention, though, is that nucleus might, and we're still looking, I'm still looking into this, I'm collecting some patient reports. I found one from my friend Dylan.

Starting point is 01:00:06 She got a report from Nucleus, and she got one from an invitating. It was a whole genome sequencing thing done. And Nucleus, and I did see the report, Nucleus said she had a medealian disorder, a monogenic disorder, called by one gene, and Infante said she didn't have it. Okay, this is weird. So what's the inconsistency we don't really know? Nucleus recently changed this result. Now, per the law, you have to notify your customers if you change their sequencing results.

Starting point is 01:00:35 This is a CLIA regulation, the CMS regulates this, and that is a potentially major issue that they might be out of compliance with. And I understand that startups are often out of compliance, and a lot of them fake it until they make it in terms of following the letter of the law. but I'm of the opinion that they really shouldn't, especially because this is serious tech with major implications for people, their families, and I mean, all future generations of their families.

Starting point is 01:00:58 This is, as I think Jordy called it, bloodline optimization. And we have these rules in place for a reason. You need to show you're abiding by them. Yeah, how much do the other players in the category, how much are they worried about nucleus just kind of setting back the entire category? So I have gotten, it's on video and because Kian tweeted about being in and I'm allowed to say that we were at a private conference and I did lead a panel on this topic where Kian was one of the panelists and the heads of some other embryo selection companies were also on the panel as well. And I will say that everyone was worried. Everyone was worried about one of the major missteps that actually has already been addressed by another one of the companies and the series of major missteps made by Nucleus that.

Starting point is 01:01:47 they have failed to address. I warned Kean about this stuff many months ago. I've warned him about problems for more than a year now, and he has simply not addressed them. They're still there on the website. You can go find them. I've archived these pages a few times to see, are they getting to them? Are they getting to them? The answer is no. They are still making claims that are either not possible, not possible with the current tech, might be possible in 20 years, or just don't seem consistent with the evidence that they provided their customers. Okay. Last question. And then we're going to hop on with Keon. I'm a big believer in redemption arcs. You're obviously unhappy with Nucleus's behavior in the industry. What does redemption

Starting point is 01:02:29 look like? What would Keon have to do or show you to get back in the good graces of the industry in your good graces? So the whole industry knows that Kean has been a problem for a while. And we actually, a lot of people have tried to give him a redemption arc already. They have tried to come to him. They have given him clear advice on what he needs to do. They have told him, you need to stop making X, Y, Z different claims. They have told him you need to offer scores that are vetted. You need to be open about your vetting.

Starting point is 01:03:00 You need to be open about your sources. You need to qualify everything like the other companies do, but you haven't. He has been told this for a long time. I sent you some pictures earlier today. You can see the panel we're on. where clearly like this has been a thing that's come up a lot and we wanted to give them the arc already but it came to this it came to a person going online a blogger deciding hey i'm going to look into this after reading the various blog post and saying well shoot so to make up for it

Starting point is 01:03:27 i think they would need to be incredibly open and they need to apologize and they need to admit to what they did wrong and they need to say that they were out of compliance with cleo rules and regulations and they need to say hey we might have misstepped here or there we didn't know we were doing this or whatever, like whatever it takes to be accurate. Document everything. Tell us everything you've done. Tell us all the missteps. Don't exaggerate. Do not lie. Just be upfront with everyone and submit yourself to regulation. Not in the sense that you have to go and tell the regulator you want to like implement whatever new rules and regs, but submit yourself to openness. Be really open. Stop this whole thing about not telling us your methods, which they've done. And

Starting point is 01:04:06 Citron Mala has documented that in the latest blog post. And give us your data. Give your data out. stop providing, like saying it's going to be available upon request. It doesn't matter if your competitors have it. Offer better pricing or something. Compete on some other margin because we shouldn't have to compete on trusting you. I'm saying we, I'm not talking. I don't have a company in the space, but everyone should be trusted. All the companies in the space need to shape up a little bit and they need to be a little more open.

Starting point is 01:04:32 And Keen needs to do that the most. Thank you for coming on the show and breaking it down for us. We appreciate you taking the time and walking through all of that. Have a great rest of your day. who knows, you might be on the show very soon as this debate continues. So we really appreciate you taking the time. Thank you. Talk soon. Have a good old, guys.

Starting point is 01:04:49 Bye. Before we bring you on Kian, let me tell you about fall, build and deploy AI, video and image models, trusted by millions, to power generative media at scale. And we have Kian from Nucleus in the Restream waiting room. Let's bring him in to the TBP and Ultron.

Starting point is 01:05:06 Kian, good to see you. I wish it was less dramatic circumstances, but, you know, it makes for good TV. And we're happy that you're here, and we can chat about this. And, I mean, I'd love to just give you the floor. I'm sure you saw, you know, some of the early segments. Where do you think it's important to start? Where do you think it's most important to set the record straight as the first point?

Starting point is 01:05:27 And then I'm sure we'll have a bunch of questions. Well, I didn't see what Kramu said. I was busy helping a patient. But I think the key thing to remember is that Kramu and I, we are definitely aligned on doing great science. At the end of the day that we want to do, we want to serve the patient want to do amazing science. I think what we're not aligned on is Kramu basically for several months has not disclosed that he's been affiliated with a competitor. And, you know,

Starting point is 01:05:54 that wouldn't be so much of a problem unless they're basically concerning together. And so that's on the Kramu side of things. But honestly, that's like the less important thing to me. Yes, I agree. I think that's less important. So I've seen him post positively about your competitors. I've not seen any proof that he's actually being paid or has equity in that competitor. But to me, it almost doesn't matter. It could, like, every single post from him and Sichuan Mala could literally be from Norrit, Orchid, or someone at one of your competitors, you still need to address it, right? A hundred percent.

Starting point is 01:06:28 Okay, cool. And so first and foremost, I'm going to say that our science is completely public and it's been completely public. So one thing that it's really important to say is that anyone, and by the way, we've at this point, have shared our models with over 15 different entities. which includes, by the way, people affiliate with our competitors, several of them. Just to be clear, I mean, Kermew said that a number of people that he's aware of have requested access to the models and not been given access and been told to stop reaching out.

Starting point is 01:06:59 And so I do think, I don't know, who's... Completely utterly inaccurate and false. There is not one person who has filled in the nucleus origin type form, which is a type form, to fill in a type form that has not gotten access to our model weights. Okay. And by the way, that includes people affiliated with the competitor. Okay. And so I think what's really important here is the science is public.

Starting point is 01:07:21 The message of the community is go and test it. In fact, our science is public. The competitors is not. So what I would propose is they should make their science public and let's have a third party independently evaluate the rigor, the quality of the science, and let's do it for everyone to see. Instead of he said, she said, they tit for the tat, you know, put the science out there, have a third party independently evaluate them. That is my message to our competitor. We are happy to stay behind our science and we know that it's the highest quality science that

Starting point is 01:07:54 can exist today. But by the way, John, that's not even the point either. You know what the point actually is? The point is about the patient. It's about having the empathy with the patient so they can know when they do embryonic selection, they can feel comfortable and confident in the results. And this Twitter back and forth, this tit for tat, this, oh, this person's race changed on the nucleus landing page, it's ridiculous. Yeah, yeah. It's really ridiculous. And so that's my message. Speaking of the patients on the landing pages, what about the, what about the chronology here, this idea that there's a review of a baby with three months old, takes nine months, it works, Maybe should have happened a year ago.

Starting point is 01:08:35 Was the service available a year ago? How do you square that particular allegation that the review, the timing of the review just doesn't line up with what must have happened in the real world had they used your service? So there were several claims about the reviews. Let me address each. Please. First and foremost, obviously as a HIPAA covered entity, we cannot disclose patient name, much less their picture.

Starting point is 01:08:59 Okay. If a patient chooses to, they can publicly. endorse the company and they can put their name in their picture. Otherwise, a patient can submit an anonymous, anonymous review, and then we'll put that according to the landing page. Okay. And so maybe perhaps the people on the Twitter timeline, maybe they never run a company that involves any protections to the patients.

Starting point is 01:09:19 Maybe they don't know about this. Maybe in like the broader tech community, it's like unfamiliar. If you run a software company, it wouldn't make sense necessarily, does not disclose the patient name? And I'm going to say, too, John, this is really important. Yeah, do you think you have to disclose the fact that you're using an AI generated image is that best practice or is that legally required you know what i think we should do though is now that the community gave us this feedback is we should update it make it more clear hey

Starting point is 01:09:42 this is clearly not a real picture sure and this is both not a real name that's reasonable yeah yeah okay now fraud this that guys really come on okay we'll update it we'll make sure that the picture and also the names are more clear yeah but again we're a hippocovered entity you can imagine when you launch an embryo product specifically people do not want their name to be affiliated with it. I mean, you have anon accounts that don't want their names affiliate with these things. Imagine a patient actually underwent Nucleus's services. So now regarding the timeline thing. That's the second thing you mentioned. I want to directly address that as well. Obviously, a company like Nucleus can start providing services to patients

Starting point is 01:10:18 earlier than we publicly launch a product. Moreover, you would imagine the services that you provide to patients would be the ones that actually, the beta service that you provide to patients would be the ones that you'd have reviews for. Obviously, because they do the services prior to the company actually launching the services publicly. So that's it. That's the answer. So you were using the service before, maybe a year ago or something, then you announced the service.

Starting point is 01:10:45 And that's why the... We have a question, John, to do embryo analyses probably three years ago. Yeah. I would wager that, you know, that actually was probably one of the first times before any of these companies to actually provide a sort of this sort of services. So we didn't think about this for a long, long time. Yeah, of course. Yeah, it makes sense.

Starting point is 01:11:05 It's a very logical place to go. It's also a very competitive industry. There's a bunch of reasons why you'd want to play in that space. What about the delta or the perceived gap between the marketing claims? What's on the billboard? What's in the New York subway? I'm seeing 50% IQ, 80% height. They feel like bold claims.

Starting point is 01:11:26 What's actually possible? What can customers actually... get from nucleus today? What could they get a year ago when you were beta testing the product? I want to interrogate the gap there. So to be clear, there is no gap. Right. Have a healthier baby. Have a taller baby. Have a smarter baby. IQ is 50% genetic. High is 80% genetic. These are just facts of the matter. The latter two are herdability estimates. They are what they are. The former two are basically describing what you can do with nucleus. What I think is interesting here is actually a broader commentary.

Starting point is 01:12:08 Nucleus is bringing the science mainstream. We are taking it out of the little echelons of the rationalist community, the little echelons of the scholars going back and forth out of it out of the Twitter out of the way, and we're bringing it to the actual people who will benefit and use these services. I cannot tell you when you actually talk to a patient, not somebody on tech Twitter. When you talk to a patient, they have no idea. idea of this technology exists. Yeah.

Starting point is 01:12:32 And the first time they're discovering it is when they go and they actually see the campaign in the subway. And now I think it's personally good for the entire industry, right, where you actually bring broader awareness, it lifts everybody, us, our competitors, and makes this actually more and more into a space. It's very early. My advice to our competitors is focus on serving your patients because at the end of the day, the market's huge and this market is extremely in its infancy.

Starting point is 01:12:57 And I think they would push back and say the industry can't afford to be sloppy. And I think that the, I agree. I think it's a fair allegation that some of the ways that nucleus materials have been presented have been sloppy. Do you, I guess one question I have is to think. What are you specifically talking about? What has been sloppy? specifically the reviews the reviews are sloppy i i i think specifically using using an ai image or using a stock image without making it clear that this is anonymized because of hippa if it just said

Starting point is 01:13:40 if it said at the bottom there's little asterisk that's anonymous because of hippa i think everyone would be like oh yeah that makes sense like they made a choice and i think that's like the first thing that i would count as like sloppy that's that's fair that's fair okay and we're going to update that Do you think that, you know, that's proportional to the temperature on Twitter? I think this is the most politically charged category in technology. And you can't afford to, you can't, like, basically the industry as a whole, I don't think can afford to make a lot of mistakes, right? I completely agree. This is the, this is going to involve the health of the children of all the industry's, you know, clients.

Starting point is 01:14:23 There were allegations, too. that you guys were sort of updated a test result on the fly. I have no idea if this is true, but I saw the claim going around. Somebody had gotten a certain test result, and then it had been updated two weeks later. Yeah, what's going on there? What's happening? That's super interesting. Nucleus, remember guys, unlike these other players, we've served thousands of patients. Sure. And then we can go on our website right now and use our product. They can see our services, right? I mean, I think it's really funny what's flying around when someone can just go and buy a DNA kit and see the product for themselves. Yeah. Okay.

Starting point is 01:14:55 So the idea that, oh, we updated a model, we've updated models for last several years. I mean, results will change. We make that very clear. And by the way, any embryo selection company, nucleus is full stack. We do adult DNA testing analysis. We also do the embryo and we also do a full end-to-end IVF experience. We're kind of multi-product. But these models will evolve.

Starting point is 01:15:14 One point here is important John to mention. People have a very good intuition when it comes to AI that chat TPT is going to be better next week. GROC's going to be better at two months or now. the same expectation has to be communicated to the genetic optimization industry. The main limitation for building polygenic predictors are, is data. It's a data problem. And so what's going to happen is all the different polygenic predictors are going to be approximately equivalent, okay, until we get more and more and more and more and more data

Starting point is 01:15:44 or people get more and more access to data. That is the fundamental bottleneck of the industry. In other words, similar to actually the AI situation, all the value is going to shift downstream to the application layer. The reason why, the reason why people are so upset is because Nuclase has excellent science, rigorous science, and we have excellent marketing, excellent product, and we've served more patients than all the companies put together. And so I think the position is something. But if somebody gets a result and they make a decision based on that, and then a few weeks later, you come in and you say, actually, it was the opposite.

Starting point is 01:16:21 It was different than what we said, but they've ultimately made a decision that they cannot correct. Is there a regulatory framework? Yeah, that's a tough, that's a very tough situation. Is there a regulatory framework that requires you to communicate to patients that... 100%. So, do you guys do that? Of course, we're a Cleo-certified cap-of-credit laboratory. Any single time an update is made, it would be reflected in the physician report of the respective customer.

Starting point is 01:16:49 And also importantly, remember, there's a lot of decisions. disclosures, a lot of different consents people have to sign when they sign up for a genetic testing services. One of them is, of course, that the results can and will be updated. And that applies to everybody in the industry. And that, by the way, applies to any lab developed test. The tests are getting better as more data comes in. Software improves. That is the nature of how these things go. That's something that Nucl has been very upfront with and we're very upfront if results do update or change. Yeah. Yeah, I'm super. Yeah. I mean, we were talking about like there, there is a world. I think this is why the marketing is really like so important for what you do.

Starting point is 01:17:27 There are so many other industries where if, you know, if I'm watching and if I see a billboard for Instagram and it says like the most entertaining images on the internet and I downloaded and like, I'm like, this isn't as entertaining as YouTube. Like I feel like, yeah, whatever, you know, it's not that big of a deal. But the worst thing that could happen is, is I, you know, someone does nucleus with their, their embryo. It's accurate, but there's a minus sign in front of everything. So instead of like the tall and smart one, they get the shorter, dumber one, and then they can never go back from that. And this is this weird scenario. And so I do think that there's a, I think that you're,

Starting point is 01:18:05 you're potentially correct that the, the temperature on Twitter is, is high, but it is a very high stakes industry. I think this is, I think it is appropriate that you will just kind of need to deal with this for a while. I would love some background on this idea of the implications of scores changing. I feel like that happened with 23 and me. I feel like that was the promise of 23 and me was that you do it and then as more data comes in, you can opt into extra studies. How did they solve that? Have you studied that business or any other previous DNA testing businesses to understand how to actually wrestle with that issue of the changing results over time as more data is created? So there's a couple things to unpack there.

Starting point is 01:18:56 The first is that you don't want to be a widget. Fundamentally, 23 is a widget. They thought that with their data, they could be able to actually eventually mine drugs, which because of the limited nature of their data is actually. not best for drug discovery. But on the whole genome front, maybe you could argue, okay, that that won't be a problem if ever a company wanted to do that. But actually, there's a meta point here about the genetics industry, which is, I think,

Starting point is 01:19:23 and this is, you know, kind of for me, which is I think that too often people focus on the genetic test. It's not actually about the genetic test, right? Usually people have a very specific acute problem they want to solve. In the IVF context, it's about, for example, having their best baby. So you'll see more and more patients are signing up for new. nucleus to do IVF. And it's IVF that's full stack powered by our genetic stack because what's really unique about nucleus as a business is because we've done the adult DNA testing

Starting point is 01:19:50 for many years now. And then now we also do the embryo analysis. So we can actually build a new kind of IVF experience called IVF Plus that is centered around giving parents as much information as possible into their embryos. And notice that shift there, John. You go from being a widget to actually trying to serve a process, a workflow that exists, and that if anyone who wants to have a baby has to undergo. One thing that Kremu was kind of hammering that I hope I don't botch was something about 12 snips versus a million snips. Previous data required a lot more genetic information.

Starting point is 01:20:30 You nucleus was making the claim that you could get just as much signal from much less data. Is that right? What is your reaction to his take that it would be impossible for nucleus to derive such a strong signal about the impact of genetic information from such a small data set? My response is the models are public, go check them. I mean, the models are public. To be clear, he also said that he tried to get access to the model and that he couldn't, but you're saying that he's lying about that. Okay.

Starting point is 01:21:05 It's a complete, a complete lie. Okay. It is false. So, so, so it is incumbent upon him that he proved that he, that he applied or something like that. I don't know. I doubt he took screenshots of his. I don't know.

Starting point is 01:21:16 I mean, he wants, yeah. I mean, also, by the way, we'll give the, we'll give the weights. Okay. Just go, literally mynuclac.com slash labs. Yeah. Go on there. It says origin, open weight. Like open weight.

Starting point is 01:21:27 You know, instead of all this debacle, all this Twitter nonsense, the message to the scientific community, the message to Twitter is go check the weights. And not just that. Our models are public. Okay. Yours are not. Make your models public and we can actually have a conversation then. What about the allegations of, of plagiarism,

Starting point is 01:21:46 copying from a different paper. When I looked at it, it didn't look like, you know, I'm not equipped to evaluate if it's a direct copy or not. Is it scientific breast practice to stand it on the shoulders of giants and pull from, you know, what the best research you're doing even if they're at your competitors or with something else more sinister going on or maybe a mistake? I think the thing to remember is like, you know, a lot of research, like if you think about

Starting point is 01:22:16 like the in AI context, right, it's like attention is all you need. Yeah, they will use that paper. Yes, yes, yes. The transfer paper has gone everywhere. It's the same thing. What are you going to say? A. Google stole, like open AI stole it from Google.

Starting point is 01:22:27 Yes. Come on. You're not going to say that because it's public research. And so it actually comes back to the fact that nucleus is bringing this technology mainstream. Our campaign is succeeding in every metric. Sales are up, signups are up. We have multiple articles covering the campaign. People are starting a national conversation as we saw on Twitter. And I think people want to try to tear nucleus down because we have excellent science. And science, excellent science, is the first step to building a very successful

Starting point is 01:22:57 biotech business. And this is what I think a lot of the armchair philosophers on Twitter don't fully grasp, which is in order to actually, in order to actually, actually build a successful business. You need to have each and every single one of these components, John. It's very easy for a scientist that's never actually built a business, that's never actually tried to sell to a normal person, right, to say, oh, the marketing is extreme. Oh, the marketing's over the top. Oh, the marketing's this. The marketing is helping a patient identify a genetic risk that they would unknowingly pass down to their child. There was a patient recently who they were doing IVF with nucleus and they identified that

Starting point is 01:23:29 they actually had a gastric cancer marker, a coerctor cancer marker actually, excuse me. And that marker doesn't just impact their health. It also impacts their future child's health. So what do they do? They chose an embryo without that specific genetic marker. And they took themselves precautions to make sure that they don't get cancer. And so that's what nucleus is about. And we're going to keep serving patients.

Starting point is 01:23:53 Yeah. Makes a lot of sense. Sorry to everyone. We're having a little bit of trouble with the stream. This has been a fascinating conversation. I like the gauntlet being thrown down on the next step here, actually being let's get the models in everyone's hands because that seems to be a fundamental disagreement here.

Starting point is 01:24:16 You're saying that you'll give them the model, give them the data, they're saying that they can't get it. Well, the next step out of this debate, I think that could silence a lot of the armchair philosophers. John, do you care about rebuilding trust with the broader community on X because... Yes, absolutely. And that's why you came here.

Starting point is 01:24:37 Yeah, yeah. And then I guess like what are kind of the... What are the handful of things that you're committed to doing to make that happen? A lot of things. One is the AI thing. I think, you know, we should fix that, as mentioned, right? Make it more clear that it's AI images that people want to be under prex under HIPAA. That's one thing I think we should do.

Starting point is 01:24:59 Another thing is that, and this is originally why we made the models actually public, is because we need to be more clear that people can go and actually test and check this stuff, right? In some sense, like, a company can do something, but if people aren't aware of it, that's also my responsibility as CEO, right? And so part of the reason is why I'm saying, hey, the models are public, come, submit to get to them, is clearly there's an information gap there that for whatever reason wasn't filled in my nucleus, I'm going to have to do better. And the last thing I think, which is really important is sometimes, on Twitter, this one personally hurt me.

Starting point is 01:25:35 I, the amount of time we've spent making the product in a way that people can try to understand where it shows the error bars, for example, where you can see that there's signal, but there's not a lot of signal. Or in some cases, there might be a lot of signal with something like height, right? The amount of time and effort and product design

Starting point is 01:25:57 and science and and and sort of like like science scientific communication specifically it's been years nucleus been around for six years i've done this for years trying to get that right when you see the embryo on your smartphone it's easy to look at that and just write it off right i encourage everyone to go on pick your baby dot com go to them the flow click around look at our disclosures look at the way we present information right it really is um it's it takes a lot of time It takes a lot of effort, right? And I think that is something that I'm thinking about how do we, how can we better put the product out there? Because I think when people can't see something, when they can't see the product or they can't see the science, then they start ruminating.

Starting point is 01:26:38 So that's what I really, really want to do and how do we better kind of connect. And also, honestly, I think I need to be also more mindful. I mean, clearly, I inadvertently, I've pissed some people off. And that's understandable. That feedback's taken in. Okay. I'll be more mindful because I really, really, really, nothing is more important to me that a patient coming to nucleus getting the highest quality care in every sense. Yeah. Well, thank you for taking the time to talk to us. Thank you for taking the time to break all of this down. We are fighting through some technical issues, completely unrelated to what's going on. Somebody didn't want this interview to happen. Somebody didn't want this interview to happen. I don't know. But we really appreciate you taking the time to hop on and set the record straight.

Starting point is 01:27:24 give your side of the story and explain what you think folks are getting wrong on the various issues here. It's an incredibly detailed and incredibly important discussion, and we really thank you for taking the time to talk to us. Thank you, John. Thank you, Jordan. Have a good rest of your day. We'll talk to you soon. Let's go back to the timeline. We do have more guests if you are tuning in because of that debate. The chat is active, but we don't. We do. We do have more guests. We do. don't know the status of the stream, the, the chat is saying that the stock is crashing because, of course, we're having stream issues. I don't, I don't necessarily think we got to put this on us. You know, there's all sorts of, you know, independent state actors that

Starting point is 01:28:13 have attacked us in the past. We've had major outages from the hypers, you know, NVIDIA put out a statement saying that they're not Enron, but that doesn't mean that they're not somehow responsible for our stream lagging in quality. Who knows? You have no idea what's going on why our stream is having issues. All I know is that it's not our fault. That's for sure. There's definitely nothing that we did. We are. Rameu says Kean is now dodging the question about the Mendelian disease. He flagged for a patient and then revised on this is illegal without notifying the patient. He's saying it's just a model update. This is not something you update on. Medellians are

Starting point is 01:28:53 definite. You have it or you do not. Yes. Mendelians. Interesting. Well, we have Joe Wisenthal from Bloomberg joining. I'm very excited to get an update from him. Oh, he's here. Welcome to the stream. Thanks for having me. It's been too long. It has been too long. We're sort of fighting for our life on the stream, but the conversation between us should be fine. We had a very dramatic debate in this in the bio context i don't know if any of this uh bubbled up to you but

Starting point is 01:29:24 this uh anonymous poster cremew is alleging that this company nucleus you might have you might have seen some subway ads have a healthier baby have a taller i've seen these i don't know if i've seen the ads i've seen the photos of the ads and i know that they're the subway stops that i sometimes go to yeah but i actually don't know if i've seen the ads i'm going to look for one tomorrow morning i'm going to go to the uh the stop that i've seen those photos I'll take my own. I'll verify my own photo. We were joking about this that there's something about New York subway advertising that's

Starting point is 01:29:56 like uniquely viral. And every company, no matter where they're based, just wants to, they see New York only as a place to run out of home and then go viral. Because this has happened a number of times, friend.com. I'm sure you, did you actually see the friend.com campaign in person? I definitely saw. And without feeling, every single one was vandalized, period. There was not a single exception.

Starting point is 01:30:15 Really? That's true? Every single one? 100%. 100% of the ones that I saw are Vandal. That's hilarious. Zero acceptance. They were all over L.A.

Starting point is 01:30:24 There was one that it's right on the drive from the gym to our office, to our studio. But it's hilarious because it's a billboard that's directly up against a wall. So you can't see the full ad. You just see end.com, and it just is very ominous, like, this is the end. And it's clear because the buy was so broad that they didn't think that, oh, yeah, that one's actually not valuable at all. I love that. I love that they literally, they just spammed real life. That's what happened, right? They spammed the physical world. And it was getting worse and worse. They were spamming Los Angeles. And then I live in a

Starting point is 01:30:58 suburb of Los Angeles, Pasadena. And then one day, I'm driving around my hometown, which is very quaint and sort of out of the loop. It's not, it's not the San Francisco hubbub. It's not teapot. It's, it's Pasadena. It's a very chill suburb. And I see a friend.com at billboard, and I'm like, I can't believe you followed me here. It's following me everywhere. It follows me on the internet follows me to L.A., follows me to Pasadena. I can't get away from it. It's like clicking on an ad, you know, it's like, or something, and then you just get, you know, served that same ad over and over again, except this one is real life.

Starting point is 01:31:29 I do wonder if the, if the rage bait, if the, if these, like, because there's these small campaigns from these, like, tech startups that are, in one way, you know, I, I criticize them, we criticize them here on the show, but, you know, I think that in some ways these are just, you know, pranksters on the internet. These are young entrepreneurs figuring out. how to get attention. I'm somewhat sympathetic. But at the same time, I do see that there are people that make political decisions based on this stuff. Have you been following the tech lash and all of this, like the data center news? I've heard, yeah, absolutely. I've heard of it. I've heard

Starting point is 01:32:03 that there's a bit of a backlash brewing, yes. How real do you think it is? And do you think it's going to be more power-driven or water-driven? I think it's going to be very real. I actually think it's going to be very, yeah, it could be all I think it's very real. I mean, I really wonder who is even in the 20, you know, 2028 or even 2026, who is going to run on anything that actually is like, oh, no, actually data centers are good. AI is an important industry of the future. You know, actually the power stuff has been overstated.

Starting point is 01:32:38 The water stuff has been dramatically overstated. I have a hard time seeing any politician actually running on that case. It seems like anyone, Democrat or Republican, feels like they almost have to be automatically against data centers for some reason. And I think you're right. Like, I think that there's, people have some intuition. They don't like AI companies. They don't like big tech, whatever. And then they fill in whatever seems satisfying. So, I don't know, I think the water stuff seems like the most out there and disconnected from reality. And so, therefore, the people who are most disconnected from reality will probably latch on to the water stuff.

Starting point is 01:33:13 the people who are a little more sort of, I don't know, Normie in their views about how the economy will maybe talk about the electricity. And then people who just dislike the sort of broader effective tech and all these different things, they're probably going to talk about the slop and how it's ruining society, which maybe it is, I'm not really sure. But it does feel like there's something, to your point, there's something for literally everyone to latch onto in the anti-AI fight.

Starting point is 01:33:39 Anyone can have their thing which resonates with them, in which case it's like hard to see where the constituency is to build these things. So I think what you end up is probably a lot of entities, you know, looking for places in the middle of nowhere in Texas where they can set up their own natural gas plant and their own gas turbines or something and, you know, stay off the grid as long as they can. We got to get a politician who's just like, you know what, like this stuff's delightful. Have you seen a studio Ghibli? I like the studio Ghibli.

Starting point is 01:34:08 I like a Suno song. yeah someone will like who is it like just even when you say that though like it's hard to imagine who that politician is or what the case that they make you know so few people right today and i mean i use some ai tool basically every day you look up something you know who speaks for us who speaks who speaks for the constituents that do one deep research report every couple days and they and they use the the thinking models to answer some questions every once in a while, and they generate some funny images every once in a while. And then they kind of move on, and then they come back to it in a little bit.

Starting point is 01:34:46 And then it's like, oh, they have a new, they have a new thing out. Let me, I'll run my battery of tests that I do with every new model, the questions that I always ask. Yeah, yeah, yeah, it's pretty good. Yeah, yeah, pretty good. No, I think it's going to be, it's a real, it's, it's going to be really tricky politically. I think, and I really think it'll be huge, because just to, you know, the labor market is

Starting point is 01:35:05 soft. Yeah. So here is this, and electricity prices are high. So, and granted, we don't know what the conditions are going to be in 2026 or 2028. Yep. But right now you have people who are, you know, in the AI industry, obviously many of them, in fact, talking about the potential for AI to displace significant amounts of labor. Who's going to vote for that? Like, how is that?

Starting point is 01:35:25 Again, I could envision some world in which human life is a lot better when we've been freed from many laborious tasks. Totally. But in today or tomorrow or 2026, it's hard to see, like, what is the first? thing that gets people excited. Yeah, I mean, pre-internet, pre-internet, it would have been a lot easier for Silicon Valley to navigate, navigate this tech trend, this like technological cycle because you could have been going to the capital markets and saying, well, you actually want to give me $100 billion because I'm going to deplace, you know, millions of jobs and all these categories.

Starting point is 01:36:04 And we're going to capture some percentage of that. And then you could go to the everyday people. You could go to the media and say, you know, we're going to create super abundance for everyone. And there won't be any jobs. And you're going to have an army of robots that, you know, run your whole life. And it's just going to be this beautiful utopia. And now people go on one podcast and they want to talk about job displacement. And they go on another podcast and they talk about utopia.

Starting point is 01:36:31 And the same people end up seeing both of them. And it just doesn't, it doesn't work. I couldn't agree with you anymore on this specific point. No, for real. I think this is like one of the defining phenomenons of our time, which is that there is no such thing is segmenting audience. And this is a phenomenon that occurs across all sorts of different realms, whether we're talking about politics or Wall Street, et cetera. Historically, people talk differently to different groups, and that's just very normal, et cetera. And you tell one story to someone.

Starting point is 01:37:00 If you can't do that anymore at all, everything in any context is implicitly understood. including internal company communications, right? Where we expect everything to leak and you expect someone to any memo that you share. So internal communications can't really have any candor anymore because that all reads as PR because that's expected to leak in some part. And every single person running a lab

Starting point is 01:37:23 has had something where six years ago they were on a mic saying confidently, it's very likely that AI will kill all of us on a long enough time horizon. And then that resurfaces today. Or even worse, they'll say, like one of my worst nightmares is that this happens and that's why I'm working to stop that scenario and then people just remember the bad yeah yeah but this is a weird thing about

Starting point is 01:37:46 AI specifically too which is that we're sort of at this point where several technologies that we maybe we were really excited about at some point in the past years down the line like oh we didn't really think this turned out to be not so great or we don't really love the effect that this is having out society and so forth AI seems very strange and distinct from anything else I can think of in memory, where from day one, even before it really existed, the people invested in it and sort of working on it have talked about its downsides in frequently dramatic terms, such as you're describing. So it's very different than any other technology where usually the downsides only become apparent years and years after they've sort of suffused and soaked into

Starting point is 01:38:25 society. Here they're talked about from day one. Okay. Last week, there were two headlines that I was trying to turn into some sort of pithy phrase or headline. I couldn't really land it, but I'd love your feedback. So, NVIDIA beat earnings, and we got a jobs beat.

Starting point is 01:38:44 And so I was riffing, I was like, demand for robots and humans through the roof? Yeah, that's right. Is that what's going on? Or am I misunderstanding the jobs numbers?

Starting point is 01:38:52 Was that less impressive than people thought? The October jobs report is canceled. Yeah. And the GDP. So, yeah, I'll say a few things. I mean, that's September, that jobs report is from September.

Starting point is 01:39:03 It feels like a lifetime ago. You know, the other thing, too, is that although it did the pace of job creation for September did come in higher than expected, once again, job creation was actually negative if you include health care and social care work, which is a thing that I think I've talked about a couple of times when I've caught up with you guys. But those are the jobs that we would really like to see AI liberate us from, right? It would be really nice if you could actually have robots change the bedpans of seniors or other these things. that are sort of low productivity jobs, jobs that for many people are low paid, kind of miserable in many cases, et cetera. So I think what's going on, unfortunately, is that we're in this environment in which job creation overall for most sectors is pretty mediocre, including finance, including tech, et cetera. The one thing that keeps powering us forward are these sort of menial, low-paid

Starting point is 01:39:57 jobs in health care, service sector, taking care of seniors, etc. And I don't know. feels like we're very long, very long until anything AI robot related is actually getting into that sector. But that would be the dream, right? I mean, that's what, that would be the dream to make progress in that front. I mean, elder care was the original pitch of Honda Asimow back in like 2000 or 1999. It was like, yeah, this robot's going to bring the meal to the, to the old person in the sickbed. Wouldn't that be amazing? Makes a ton of sense. And there are, do, do be fair, there are some robotic, like, you know, with wheels on it, rolls around, does some stuff. But certainly not broad deployments by any means. Not, yeah, not broad deployment,

Starting point is 01:40:42 not at the scale. And so what happens is we have this economy where, you know, the unemployment rate is still pretty low by historical standards, although it's gone up a fair amount a little bit in the last year. But there's just so much pressure being put on the existing labor force to care for the aging population. It creates some pretty serious strains. Yeah. How have you been tracking the just the AI spending bubble, this idea of, you know, massive growth debt coming into tech for the first time in a long time. I feel like a lot of tech people are so used to the venture capital model. Okay, yeah, it might be $100 million at risk, but this is, you know, 1% of overall allocation. There's a bunch of LPs from foundations. It's very diversified.

Starting point is 01:41:25 And if that company goes bust, that's fine because another company is going to do great. Now we're looking at serious numbers. We're getting into the trillions. There's debt involved. People don't know how to manage with debt and deal with that. How are you processing it? It was sort of a real wake-up call to me when I started seeing people online post screenshots of credit default swaps on Oracle debt.

Starting point is 01:41:47 Like that was the moment where I was like, oh, this really has transformed from the sort of equity funded, free cash flow funded environment, which is characterized tech for, you know, over two decades, really going back to until the telecom bubble. So as soon as you, and I know these companies have debt, I mean, you see them, Apple has debt. They've had it for a long time. A lot of these debt issuance, though, in recent years have been more or less like exercises in cash flow management or optimizing taxes and stuff like that. Now it feels like, okay, like this debt is at a very real scale where I'm looking at core weave credit default swabs, Oracle credit to fault swabs, et cetera. That is, to my

Starting point is 01:42:27 my mind a signal of, of course, something's changing. Blue Owl, they're publicly traded. That's the private credit company that's gone in with a meta to do some data center financing. That stock, I think, has become a bit of a proxy for how people perceive some of the financing risk in that space. The deal that, we have an episode coming out, I think sometime next week, actually, where we talk a little bit about how the credit construction of these events. the founder of this company, Noetica, which uses AI to examine credit agreements, actually, walked us through it. It's some fascinating stuff, but people are sort of looking at, okay, who is the real bagholder here? What is the risk of, you know, people are talking about, obviously, what is the risk if there's not as much value in these GPUs five years from now as we may have thought?

Starting point is 01:43:21 And to your point, that's just totally novel for tech in recent years. It doesn't feel like we've, it's been so long since the idea of debt or leverage in that way has been part of the tech story. Yeah. What did you, what did you make of Nvidia sending out that report for the sell side over the weekend? And then, and then again, they had another post. They had another post today. I'll just read it in case anyone's just tuning in this morning. Let's see, we covered it at the beginning of this show.

Starting point is 01:43:53 We're delighted by Google. success. They've made some great advances in AI and we continue to supply to Google. NVIDIA is a generation ahead of the industry. It's the only platform that runs every AI model and does it everywhere. Computing is done. NVIDIA offers greater performance,

Starting point is 01:44:09 versatility, fungibility than A6, which are designed for specific AI frameworks or functions. So again, not like these are meant to be confidence inspiring but they come off the, you know, they have the opposite effect. It's really weird. The tweet,

Starting point is 01:44:24 Like, I guess I sort of understood, okay, so they sent out that note, you know, look, as I mentioned, over the last several weeks, for some reason, Wall Street has got really taken in with this whole conversation about GPU depreciation schedules. Maybe I think Michael Burry's been talking about it or something. Okay, fine, like this has taken hold. I am sure this is the type of thing where for every 100 people who are talking about this, maybe one knows what they're actually talking about. Like, I have no doubt that there's all kinds of noise out there. And so maybe it makes sense for Invidia. They're like, oh, let's put out some information that we have. Like, I guess it's a little weird, but okay, like I sort of get it, given the degree to which this has just become a meme in the last few weeks. The tweet about Google is a little strange. Like, I typically, here's the biggest, most powerful company in the entire world, and they're doing this sort of weird. We're excited to see, we're happy for your success and it reminded me it reminded me of jensen's comment on the on the open a i amd deal where it was like minus snarky he was saying something like i'm kind of surprised they would they're so

Starting point is 01:45:36 excited about their new chip i'm surprised they would give away 10% of their company before they've even developed it so so again to me i feel like when you're operating from a position of strength and confidence you never you never talk about competitors from official channels. In fact, I had, I've given it, I've given this advice very recently where like there's a, uh, a company that every time their competitor does something, they post about it. Guess that, get, guess, guess, guess, like a quote tweet, too. It's so much easier to just be like, like, our esteemed, our, the other members of our industry or like the other big tech companies do it this way or like other handset makers or other smartphone makers. And everyone knows that

Starting point is 01:46:22 you're talking about Google or iPhone. or whatever you're comparing to. But in this case, the company that they always talk about has a hundred times their revenue. And so it screams like, it screams like we're obsessed with our competitor. But in this case, it's like the TPU is just like the first sign of an external. Right. And you immediately like pounce on them.

Starting point is 01:46:46 And you're like, you know, I've been thinking, I mean, one of the funny things about Jensen and video overall is you think about these other mega-cap tech companies and these other ultra-rich CEOs, you know, they've been dominant for, in most of these cases, for like well over a decade, right? They've been some of the most powerful companies in the entire world for over a decade, obviously. Invidia is sort of this weird case because, you know, four years ago, people were talking about, oh, this is the company that makes chips for Ethereum mining, or this is, like, that was how a lot of people talked about in video.

Starting point is 01:47:21 I think even as, yeah, as recently as 2021, people were talking about, oh, NVIDIA's, so it might be a little... Well, people don't remember before there was the Mag 7, there was Fang, and there was an end in Fang, and what was the... It was Netflix. It was Netflix. It was Facebook, Amazon, Apple, Netflix, Google. Yeah. And then crazy to not include Microsoft, right? Right.

Starting point is 01:47:45 And so they added Microsoft and they took out Netflix and then they added Nvidia and Tesla, of course. But yeah, I mean, what a wild drive for Jensen. Like, what a run. I mean, obviously, like, a video's been an incredibly successful company, but it went from, like, you know, a pretty successful chip company to the biggest company in the world in a matter of few years. Yeah. Maybe there's something there.

Starting point is 01:48:07 Maybe there's still a chip on the shoulder. Maybe there's still some culture of feeling like it's the underdog and something where it's like, yeah, it's like act like you've been here. Yeah, it's one thing. But maybe this is a start on a new to them. I mean, the 10-year thing is real. So here's the thing. If you're a startup and Google enters your market, I've seen a lot of founder or like a big company, if you get Sherlocked by Apple, you're going to hit the timeline and be like, I'm delighted by Apple's new mobile app.

Starting point is 01:48:36 We're delighted to continue serving. They validated the market. Thank you for, thank you, Apple, for validating our industry. Yeah, yeah, yeah. But I don't think, I don't think, Nvidia doesn't exactly need validation. at this point from a competitor. And the TV is a decade old, by the way. Right, it's been around for a while.

Starting point is 01:48:55 Right, which is another sort of interesting dimension of all of this, which is like suddenly people wake up and, oh, it's not like, to your point. I mean, people have been talking about the existence of Google's TPUs for a long time. Obviously, now there's been some question about the degree, what's the strategy here? Would they ever sell them? Will they rent them out there? There have been, obviously, questions, et cetera. But it is funny how, you know, you get these.

Starting point is 01:49:18 moments where suddenly, you know, you get this 180 over the last several months, weeks, on Google, the Gemini three launch, of course, bolstering this idea. Oh, actually, they're in a good space. And suddenly, oh, and they also have these TPUs, which they've had and been working on for a long time, but nothing really changed. They've had them for a while. There's not some new breakthrough announcement here or something. And so suddenly, in video feeling like it has to respond a little strange.

Starting point is 01:49:44 Yeah, yeah. I mean, it, like, I do think that, like, there's some. so much demand that it's a little bit ridiculous to just be like, oh, Nvidia is completely over. But at the same time, like, it's never great when you go from a monopoly to a duopoly. Like, it's just that, like, it's always nice to be in a less competitive market. Yeah, but at the same time, it's like if AI is literally half as transformative as the AI bowl's thing, it doesn't match, like you can have two players and the demand is so overwhelming that. And like,

Starting point is 01:50:18 Nvidia's own sales are capped, right, by capacity and by production capacity. There's only so much, which is another sort of like interesting dimension that I'm like trying to get a better understanding of here, which is that, you know, again, it's not like search or

Starting point is 01:50:34 AI or chatbots, et cetera. There is the, there are these constraining effects on the markets from how much fab capacity is at TSM or how many, some of this underlying equipment. So the ability of anyone to scale up dramatically and expand the overall size of the market is sort of tough at this point. I don't know exactly what that means, but my guess is my assumption

Starting point is 01:50:58 would be that everyone remains pretty capacity constrained for some time. I don't think there's like tons of chips laying out there for the taking right now. Yeah. Give us an update on odd lots is a people have been congratulating you on 10 years yeah did the 10 year date happened recently the 10 year date did happen we didn't even you know it's funny I think it was um I think it was November 6th I got an email because we're so we have a party which you both are invited to we have a party in New York City in December to celebrate our 10 years so we've been sort of um that's right you play the night vision sound that's right fire me up um but so we were like Getting ready for that, we've been doing some sort of 10-year episodes, we're talking to some big names, and, you know, big, big picture thinkers and stuff like that.

Starting point is 01:51:50 But I didn't realize the actual date until the morning of, and I started getting random emails from people inside the company congratulating me, congratulating us on 10 years. But it's pretty crazy. Let's get it gone 10 times. Yeah. Ten times. Wow, 10 times gone. There we go. This has been the highlight, the 10 gong smashes.

Starting point is 01:52:14 This has been the highlight so far of our 10-year anniversary. Who's the most recent 10-year guest that exemplified that bigger picture thinker? Yeah. Well, we had Ray Dalio on the podcast on Monday, which was a lot of fun. He told us about meditation. He talked to us about the importance of transparent culture, about the importance of letting junior staffers or junior employees at Bridgewater. attack the senior employees.

Starting point is 01:52:42 I loved his story about the first stock he ever bought. I know, I know. Hilarious. I don't know if he tells that constantly. I haven't listened to it every possible. He probably does. But it's such a good story. He's just like, he's just like, I saw it.

Starting point is 01:52:56 And it was the cheapest stock he could possibly buy. He was the cheapest nominal stock. It was like $5 a share. He's retail. It's a retail army. I mean, it's like in crypto, when they're like, it's zero, zero, zero, zero, zero one cent a token. And if it's a dollar, and you'll be like, I have so much money.

Starting point is 01:53:12 Can we talk about that for a second? It's really annoying with some of these crypto prices to have to, like, squint and see how many zeros there are. They need some reverse splits, I really think. Like, we need to like $1, $2. But no, to your point, right? It's basically, if we're, I don't want to be insulting, but it's basically like the dumbest. It's trying to attract the dumbest kind of traitor. There is, when you're like, oh, this is cheap because it's one, one millionth of a cent or whatever.

Starting point is 01:53:40 Still early. Yeah, I mean, I imagine if this were a dollar. I remember the doge to a dollar campaign. Everyone was a doge to a dollar. That was a very powerful meme. It was very powerful. These things are real. I thought it was going to get there because I think it got to like 30 something

Starting point is 01:53:53 cents or so I was like, oh, it's definitely going to satisfy the meme. It never made it. You're like, I thought it was going to get there because I put my whole page on it. I really thought it was, I was like, oh, it's obviously going to go to a dollar. It did seem like the meme was going to work. Yeah. It got close. It got close.

Starting point is 01:54:08 Yeah. close. It was in the run. Jordie? Are you aware of a single person that has sold their primary residence in New York and is going to, going elsewhere since the election? What's been the sentiment? I know a bunch of former odd lots guests are on the transition team, which is cool because you guys have some insight into like how these people think and the public does. Yeah. I have not heard of anyone who is leaving New York City. It's interesting. There's a, I don't know the name of the brokerage, but there is one of the brokers in New York runs these ads,

Starting point is 01:54:45 speaking of outdoor ads, in the taxis. And prior to the election, they're very scaremongery, and they were like talking about deals in Florida and all this like real estate that you could buy in Florida, et cetera. Anyway, after the election, I saw one of the ads this week, and it was talking about how JP Morgan just opened up its new headquarters in New York City and how all of the, how many employees they're going to have there. So I guess now that, uh, now that, now that, that the election is over, they can go back to selling New York City real estate because, you know, they didn't get the, I don't think they got the, those brokers did not get the outcome that they were hoping for. But yeah, we have some, we have some past guests in the transition team. Paul Williams, shout out to him, who I also play music with in my country music band, Light Sweet Crude. He's on the housing transition. Kathy Wilde, who is the head of the partnership for New York City, which is the big organization of a bunch of CEO. etc. That's like the most like sort of like died on the wool like sort of you know they're not a lobbyist group but they represent the business community even she is in the transition

Starting point is 01:55:48 team. She's also been on the podcast. So it seems like you know I've seen a bunch of people angry at some of the picks happy about some of the other picks. It seems like he has built a he's at least with these selections a little something for everyone, some moderate, some normies, some more radicals, et cetera, enough for like everyone to be a little happy and a little a little concerned maybe. Yeah, yeah, I mean, it certainly seemed like that there was that violent clip of him with Trump that was very funny where they're kind of going back.

Starting point is 01:56:16 Oh, that was crazy. It felt like they was just like, okay, there's a lot of theater on both sides here. They're both, they're all having fun. You know, they're not like, I can't even be in the same room as that person. No. So, you know.

Starting point is 01:56:26 I think that, like, for those of us not in politics, it's, I don't know. It's, like, hard for us to imagine how are these people who are, like, the rhetoric, and then they, like, get together and they smile in front of the camera and, like, feels very weird. but, you know, they're just like, they're professionals.

Starting point is 01:56:41 I mean, NFL teams have rivalries. They shake hands at the end of the game. I think that's kind of the nature of these things, which is fun. Yeah, WWE. Yeah, WWE. I've always liked the wrestling metaphor. The thing is, too, is that, like, whether we're talking about the selections of people for the transition or the, I think, very wise choice by Mamdani to, like, try to be friendly with the White House

Starting point is 01:57:05 or at least find some common ground is, look, there's some, like, pretty, hard budget constraints, et cetera. The mayor, the job of mayor is a management job. You know, the trains aren't going to run themselves. The buses aren't going to run themselves. The apartments aren't going to get billed. So there's a certain amount of like required between the budget constraints and just everything else. I think there is a certain amount of sort of forced pragmatism that's probably at play here. Yeah, that makes sense. We'll let you go in just a minute. What's next on the economic calendar that you're tracking? What should we be looking forward to through the end of the year? Obviously in our world, we're very excited about Black Friday.

Starting point is 01:57:45 You're going to be tracking those numbers. Also an interesting economic indicator. But what's on the top of your mind? You know, I think actually like basically all data through the end of the year is going to kind of be garbage because obviously we're going to get the restart of the data and we'll get more timely jobs data and so forth. But all of that is going to be. affected by the shutdown. October, we're not getting anything from October. Yeah, we're not getting anything from October, and then November is going to be affected in large part from the shutdown. So those numbers aren't going to mean anything anyway.

Starting point is 01:58:16 And so probably the next clean report where we could say, okay, this is actually well-collected data to the extent any data is there's all kinds of issues with collection and is not affected by the shutdown. So that's going to come out in January about December, which is going to be very strange. You know, I think the big development is, so I think it's December 10th is the next Fed meeting. As recently as a week ago, it was looking like they were not going to cut. The odds were below 50%. But a number of the FMC members in the last several days have come out and said they're cool to cut. So it looks like as of right now we're going to get a cut. But that is, I think that meeting is the day before we get the November jobs data.

Starting point is 01:58:55 So it's all going to be sort of a mess, I would say, until early next year. Last question. And how should people read into not getting data from the month of October? Should we say that like the simplest explanation is employees were furloughed? We don't have we don't have time to go back and collect all the information. I think in this, I think my impulse is that the simple explanation is the correct one. Due to the timing of the shutdown, the duration of the shutdown, it's like unrealistic or something like that, as opposed to something nefarious.

Starting point is 01:59:32 If there continue to be more things, then, okay, then we see what's up. But I would say this is probably pretty straightforwardly linked to the fact that, you know, the government shut down. Yeah. I'm also going to tell you about fin.a.ai that handles your customer support, the number one AI agent for customer service. Our next guest is director Michael Cratios. He is the 13th director of the White House Office of Science and Technology. policy and it is great to have him here with us on the show today. One of the best call-in setups we've seen. You look fantastic. Thank you so much for taking the time to talk to us. How are you

Starting point is 02:00:08 doing? I am great. Thank you guys so much for having me. I have followed your meteoric rise over the last year. I feel like I should have been on this show much earlier. Yeah, we would have love to have you, but we're very happy to have you today because there's massive news, but I'd love for you to introduce it and actually set up the conversation. So please take us through the announcement and then I'm sure we'll have a ton of conversation and questions to go through. Yeah, so I think maybe we can start

Starting point is 02:00:38 with the AI action plan that the president signed out in July of last year. And one of the main themes of the AI action plan, essentially to win the AI race, is all about how we can win in scientific discovery. And the question was like, how do we do that? What's like the next chapter of using AI to drive scientific innovation

Starting point is 02:00:56 in our country? And yesterday the president signed an executive order launching what is known as the Genesis mission. And I think for a lot of folks that, you know, you guys talk to every day, you know this. You know, AI has, you know, had this incredible rise over the last couple of years. It ultimately started first with large language models themselves. And those were where we scrape the totality of knowledge on the internet and we're able to then create these models that can predict all sorts of things, the next word, if you will. The next phase of it was all around coding. And you've seen these great startups that

Starting point is 02:01:26 are incredible of coding. Some of our huge model builders are awesome at coding too. But the big question that still remains is how do you apply this large language technology to scientific endeavors? How do you use it to be able to create new materials? How do you use it to create new microprocessors? How do you use it and tap it into all of this exquisite scientific technology and hardware that exists all around the world? And what the Genesis mission tries to do is to bring all of the super valuable, important data that the national labs have collected. over the last 70 years and put it to use to train large language models to be able to dramatically, and I mean dramatically accelerate the pace of scientific discovery.

Starting point is 02:02:08 So this is, I would argue, and I've said it before, and I'll keep saying it, this is going to be the single largest marshalling of the federal resources of scientific discovery since the Apollo program. So we're thrilled to kick it off, and DOE is going to be the home for it. Talk to me about the relationship between the public sector and the private sector. sector. I feel like a lot of folks in our audience have said, thank you for your service. We loved DARPA. We loved when DARPA created the internet. We love GPS. We loved the moon mission. But we got it from here. We invented the transformer at Google. We invented SpaceX and the rockets that go up and land. And we think private industry has it handled from here on the

Starting point is 02:02:48 research side. We don't need anything from the government. How do you think about, is that just, Is this just a misconception and we actually need more original new ideas, research in a government setting? Or is there more of a public-private partnership that you think will play out here? I think it's definitely more of the latter. I mean, the reality is that when some of the initial projects were launched in the 1950s, you think of a Manhattan project, that was an era where the vast majority of research and development in the United States was paid for and funded by the federal government, to the tune of almost 70 or 80 percent of it was funded by the federal government. If you zoom ahead sort of the 70 years that we are where we are today, the vast

Starting point is 02:03:27 majority has done private-private sector. You know, generally, the United States spends about a trillion dollars a year on R&D, and about 80% of it today is done by the private sector. Now, that being said, there is a critically important role for the government to play in that larger ecosystem. And it truly is an ecosystem. You have to have private sector, academia, and the federal government all working hand in hand to sort of make the very important basic research breakthroughs and ultimately commercialize those. And the secret and the sort of unique special thing about the Genesis mission is the scientific data sets that exist at the national labs. This is a unique asset that is so valuable to the American people and to all of a scientific enterprise in

Starting point is 02:04:10 United States. And the bottleneck, because you guys know so well in so many of these AI endeavors, is how do you get the right data to train these models? And for the early large language models, data was just out on the internet. You could use common crawl and scrape it. For the world of coding, you had lots and lots of coders and lots and lots of material that you could use to train those coding models. But for science, it's not that easy. You have silo data. You have pharmaceutical companies that have pharmaceutical-related data. You may have chemical companies that have chemical-related data. But what's so special about our national lab ecosystem is that it covers such a diverse and broad range of scientific endeavors. You have

Starting point is 02:04:45 biologists, you have material scientists, you have chemists, you have people who are working on space all in the same national lab infrastructure. And that depth of data is so incredibly powerful in order to accelerate scientific opportunity and endeavor. And kind of that's what the Genesis mission is all about. And to your point about, private sector involvement, the thing that we have been so excited about is the private sector is clamoring to be part of Genesis mission. We all are in this together. There's a desire to pair the incredible supercomputing infrastructure and the GPU capacity of our greatest chip companies with the great data sets we have at DOE and everything in between.

Starting point is 02:05:20 Yeah. The, can you help me understand a little bit more how a lab might actually plug into the Genesis mission because we've seen a lot of the labs say, oh, we want to start working on science. Some of the labs have already. I mean, Google has a Nobel Prize, right, for AlphaFold. Does that look like them interfacing directly with the DOE

Starting point is 02:05:42 or going through a lab? Or how does that take shape, do you think? So what DOEs ultimately, the Department of Energy is going to be creating, is a platform for scientific discovery. And there we're going to be making available the very valuable scientific data sets that can be used to run these large language models for science. And as we've seen, OpenAI had a nice tweet thread about sort of this project and how excited they are about it. We've seen lots of other companies already pursue this. Google, of course, with Alpha Fold and the Nobel Prize that they've won there. So what we're seeing is all of these large language model builders or labs themselves want to partner with the Department of Energy so we can work hand in glove to accelerate the scientific discovery.

Starting point is 02:06:26 The key is how do we, you just zoom out a second and think about it, how do we 2X the ability of us to very quickly iterate on scientific experimentation? So right now, if you have a hypothesis and you want to test it, how do we make that 2 or even 10x faster? How can we make it so that if we have an idea, we can send that idea to an AI-powered cloud lab? And the test can be run behind the scenes as we're working on a second project. So that's this sort of like big level thinking that we want to do to dramatically, dramatically accelerate the velocity

Starting point is 02:06:53 and pace of scientific discovery. Can you help me understand how you think the United States is positioned geopolitically against our near-peer rivals on the issue of AI in science? I'm pretty, I feel like I'm relatively up to speed on just the general capabilities

Starting point is 02:07:12 of American image models versus maybe Chinese image models or deep seek versus GPTOSS and the Lama models. I kind of know where we're standing in just the general industry, the industrial uses or the general uses of these LLMs and the AI models that are coming out of America and China. How are you thinking about America's competitiveness on the science and research side? So we're in a lot earlier innings of that. That's a really good question. And I think what's unique about sort of AI for science is it has to pair this exquisite scientific physical infrastructure with the models themselves.

Starting point is 02:07:52 So if you want to drive scientific discovery, you have to be able to pair what's coming off of telescopes, what's coming off of lasers, and all this stuff, to be able to match them with large language models, to be able to accelerate that loop. And we're still in the very early innings of it. And in order for us to sort of outpace and continue to keep our lead like we do have in some of these other places is we have to do something like the Genesis mission. We have to wake up a country and say, like, look, where do we have the most valuable scientific data sets and how do we make those data sets available to our model builders to be able to create the necessary tools to pair the data coming off these scientific instruments back into these AI models? And like, think about it. For us, we want to win on Fusion, for example. So for Fusion, Google's already doing this, but there's a ton of companies around the United States that are very heavily funded, that all have lots and lots of experimentation they want to do with these fusion reactors. The ability for us to be able to accelerate the modeling of that through the Genesis mission rather than each individual company having to do on its own can be really, really dramatic. So back from a pacing standpoint, I think the U.S. has all of these amazing instruments. It has the labs. It has the great private sector together. And what the Genesis mission tries to do is bring that together. in a way that's like truly American, where each of us as part of the ecosystem can play the

Starting point is 02:09:09 important role that we're best at with obvious sort of commercial goals in mind down the road. And that's what's sort of driven this great discovery we've had for years. And I think we're going to continue to see that in the years ahead. What role does academia play as part of this project in your view? So I think for us, academia has always played a very important role in pursuing early stage, basic, pre-competitive R&B. As we think of all of the most critical areas of scientific endeavor, whether it be in materials or in chemistry or in mathematics or in physics or science, there is still a critically important role in that early stage discovery science. And academia

Starting point is 02:09:51 can play a very important role there. They're the ones that have these these theses and these ideas, these hypotheses to solve some of these very early fundamental problems that can ultimately unlock great commercialized solutions in the years ahead. So academia and universities are are and do want to be part of the Genesis mission. There's an important role for a lot of these departments, professors, and thought leaders to be part of it and to introduce their ideas and concepts of things that they want to test. What ultimately Genesis is going to do, it's going to create up to 20 grand challenges on some of the biggest scientific problems we face today, and we're looking for everyone. You can be in academia, you can be in the private sector, you can be in your garage,

Starting point is 02:10:28 wherever you are. We want you to come with the best idea on how to solve those and be able to leverage a great infrastructure that the federal government has at its labs to solve those. How are you thinking about AI risk these days? I feel like we've been on a total roller coaster from sci-fi paper-clipping scenarios to some very real geopolitical competition and issues with people maybe using these models too much going somewhat crazy. There's been a wild ride that I think everyone's been on. And I'm interested to hear how you think, think about AI safety these days, is the nuclear analogy the most important? Is this a Manhattan project? How do you think about the role that the U.S. government should play in the AI safety

Starting point is 02:11:15 discussion? Yeah, you know, I think as I kind of look back and think about how the early conversations were when some of the big model lab builders came to, you know, came to Capitol Hill and, you know, talked about wanting to create kind of an IAEA for AI. I think it, I think honestly it set the wrong tone, and I think it's set sort of the industry back for a while, and I don't know how much you guys track sort of like global politics of AI policy, but for a good two years there, I think that sort of discussion on the global stage was, you know, what are the worst possible things that AI could possibly do for the world, and let's try to like figure out if we as a collective, you know, world can find some sort of like global solution to solving it.

Starting point is 02:11:59 And obviously, that's kind of the wrong approach. backed away from it. You know, the UK themselves that were kind of touting their big, you know, UK Safety Institute have renamed it, the Security Institute. We as the, as Trump administration, you know, renamed the AI Safety Institute or Department of Commerce as well and kind of moved in a direction of innovation and adoption versus sort of this nebulous catch-all safety term. But I think to me, one thing I always think about, and we always have to balance is, and maybe it's a little personal, but to me I recently became a father in July. And And I think a lot about the way that my child is going to grow up, the way that he is going to interact with this particular technology, the way it's going to impact, not only the way that he grows and learns in school, but also the types of job that he's going to have and how he's going to enter the workforce.

Starting point is 02:12:45 And I think there are very credible and real things that the American people are thinking about, about how this technology can be best used to improve their way of life and ultimately help them live a better and more fulfilled and more rich life. And we at the White House are all about finding ways to encourage adoption of this technology. And in order for that to happen, we have to have the trust of the American people. And trust is part and parcel of everything we're trying to do, whether we're building out, you know, AI for educational reasons, or to improve the amount of energy that we have in the United States or solve the biggest health crisis that we may encounter in the future, we want to build that trust. So when these solutions are coming about because of AI, it's embrace. And it's something we worry about and think about a lot. And it's something that a lot of our standards agencies are thinking about, how do you promulgate the right standards so that when these technologies are put into all these different industries, people trust them. How are you thinking about job displacement at this point?

Starting point is 02:13:43 The models are so incredible. They're smarter than me at everything. And yet I haven't actually been able to drop them in as a coworker right next to me. Every day we try to replace Tyler. We say, Tyler, replace yourself. And he's just too good. But there's still like an ambient level of anxiety that something's coming. And so I feel like you do need to be prepared.

Starting point is 02:14:04 You can't write it off entirely. But I'd be interested to hear about how you're thinking about the goals that you could even set around transitioning people through jobs. How you think about the AI job relationship? There's a couple of things. The first thing is about AI education. We have to prepare the future American work. to be able to fully leverage this technology when they enter the workforce.

Starting point is 02:14:30 The president signed an executive order in April of this year, where he prioritized K-12 AI education. The First Lady has gotten very excited about this endeavor. We're running something called the AI Presidential Challenge, which is a challenge that we're rolling out across the entire country, where we have students from all 50 states participating in it, and ultimately is going to culminate in a sort of competition

Starting point is 02:14:53 here at the White House and at the end of the school year, where we can show kind of how students are using AI to solve some of the biggest challenges they face in their local communities. It's like the presidential fitness. I used to have to run a mile and touch my toes. And now you have to use AI effectively. Now you've got a program that's an AI.

Starting point is 02:15:07 Exactly, exactly. That's actually great. The presidential vibe coding assessment. But I do think it's important because at the end of the, and I think the velocity of change that you see in how education is thinking about AI is so interesting to me. It's kind of like, and if you guys even remember

Starting point is 02:15:24 like two years ago, when ChatGPT first came out, every university out there was essentially banning it. You weren't allowed to use it. You were like, you know, violating the honor code if you ever, like, turned on ChatGPT. And now there's no college student in the country that doesn't use it. So I think that the pace of change in how these models are being used in education is just so, so dramatic. And we have to get in front of it because, you know, what we think, you know, an eighth grader is going to be doing with AI when they enter the workforce, you know, it'd be very, very hard to predict. So what we try to do mostly in the K-12 space is not necessarily teach kids how to leverage

Starting point is 02:16:00 this technology, but teach them about how the technology works. The term that was used in the executive order was demystified. This is really important. Teaching both children and teachers, like, where does it work well? Where does it not work well? Why is it answering questions like this? Why does it hallucinate? Why is it not hallucinate in these cases?

Starting point is 02:16:16 And I think the more people understand the technology they're dealing with, they'll be able to leverage it much, much better no matter where they end up in the world. Yeah. I absolutely love that because I feel like the, I mean, we talked about this, the GPT psychosis thing. There's a very big difference between understanding that what you are chatting with is a robot, is a bunch of math, and just being mystified by it. And the same thing with, you know, at some point, every parent needs to teach their kids that, hey, the explosion in that movie, they didn't actually blow up the house. You know, that's a cartoon. And now you also have to do that with AI generated images. Something I noticed, this was a major AI-related announcement, did not have a, I searched Command F for a dollar sign, didn't find it. Is that intentional? Is that like part of your focus of, hey, the government can catalyze the right sort of like progress and activity without just being a capital provider? We've seen so many, you know, announcements from this year that are really just fixated on the biggest, you know, the biggest number. And this. feels like unlocking the sort of potential within the existing ecosystem that's that's not

Starting point is 02:17:30 being properly utilized? Yeah, for the Genesis mission, I think Congress actually appropriated some money to the Department of Energy this summer and some of the legislation that passed to drive some of these AI-related efforts. And we've used that as kind of the down payment on kicking off this program. I think what you're going to be seeing in the next couple of weeks, which we're really excited about, is commitments from a number of large private sector players on what they are contributing and donating to the Genesis mission. And I think what's key here is that what makes the American innovation ecosystem so unique is the ability for us to bring all pieces of the ecosystem together to drive innovation.

Starting point is 02:18:10 You have one part academia, one part private sector, and one part federal government. And the reason why and the feature about the U.S. innovation system that allows us to be so successful and have been the home for the greatest scientific and technological breakthroughs for the last 200 years is that free market approach to innovation. And I think that's what the Genesis mission brings together and sort of highlights in the best way possible. But we are in this together with all pieces of the ecosystem working together. There's been a lot of worry about going too far, doing too much AI, too aggressive about the buildout, too much debt. There was this back and forth about the backstop.

Starting point is 02:18:47 David Sacks, of course, said we're not considering that, but have you thought any more about what the role of the government is in moderating the amount of sort of private sector AI build out and how that should even interface with the government at all? What is the framework that we should be using? I think we are very focused on removing federal barriers to folks that want to participate in the AI revolution. And for us, what we see is that if private markets and if capital markets are allocating dollars in a direction that say, like, look, we believe that we will be needing this level of compute to drive our future AI demands, we want to make sure that the bottleneck to

Starting point is 02:19:34 that deployment isn't some federal rule around permitting and allowing these things to be built out. I have faith in the way that the capital markets are allocating dollars right now, and I leave it to those capital allocators to make the best decisions possible for that capital. What we want to make sure is that we are not an inhibitor to innovation. That some arcane rules about how and when you can build some certain facility in a particular place are not the reason why the data center we really need

Starting point is 02:20:00 isn't built out. And that's the role that I think the federal government plays. And one of the three executive orders, the president signed the day that the action plan was released, was all about this. How do we get the federal government out of the way to allow the build out of these as quickly as possible. Yeah.

Starting point is 02:20:18 What about going deeper in the stack? Obviously, there's been news around Intel. There's a very rational reason to understand, prioritize, you know, having American control of the entire AI stack. How is your thinking evolved on just this idea of having, full control over the supply chain in America? Is that something you're spending time on or you're thinking about

Starting point is 02:20:48 or your thoughts have evolved on? We are. I mean, it's so critical. Back to the great day in July when we signed all of our executive orders, the third executive order was all about exporting the American AI stack. And I think, you know, we rewind, you know,

Starting point is 02:21:05 into Trump One when I was the CTO of the United States at that time. One of the challenges that I faced was traveling all around the world and trying to convince governments that they need to rip and replace their Huawei infrastructure. And at that point, you know, I think there was a big lesson learned by me and by a lot of people who served in 45 about, you know, what are lessons learned from that experience? You know, in that moment, the PRC had a technology that was good enough and was priced very

Starting point is 02:21:37 reasonably, whether it was for a subsidy or who knows what, but it was priced reasonably enough where everyone wanted it, and an ultimate was deployed for a big technological shift with 5G. Now, if you fast forward to where we are today, this next phase of AI is orders of magnitude more important than the 5G rollout. Right now, governments all around the world in very few short years will all be running some sort of AI stack. They will be having some sort of chips that will be running some sort of large language model, and on top of it, there'll be important, critical national applications, everything from the way that hospitals are running to the way the tax is collected. And we want to make sure that the U.S. is the partner of choice for that

Starting point is 02:22:25 AI stack. And right now, we're at a great spot. We have the very best chips in the world, and we have lots of them. We have the very best models in the world, and we have a competitive ecosystem that it seems like every other week they're leapfrogging each other. And then on top of that, we have the best applications in the world. So we are so committed to exporting that stack. We want to create something that is so compelling and so exciting for governments and people all around the world that they will want to use our technology. And that's what we are building here at the White House, our partners at the Commerce Department and at the State Department to get that exported all over the world, from South America to Africa to Asia, to anyone who wants it, because we believe

Starting point is 02:23:03 that the same benefits that all Americans are going to realize because of our great technology, we want to share that with the world. Yeah. Final question, a little bit of a wild card. I'm curious, are you thinking about humanoids as a category? I've been really concerned that we're going to repeat what happened with DJI, where we allowed a company to basically destroy our domestic industry and flood the country with cheap, very good, very good drones, but cheap.

Starting point is 02:23:33 drones and I worry that you know I think you can buy a unitary robot on walmart.com right now it's not it's not a concern today but if we allow the country to be flooded I would be very concerned about it in the future and I'm curious if it's on your mind at all it is actually it's funny you say that we we we were just actually speaking about this week with a number of my of my colleagues I think the it is challenging I mean there was a a large well-known top sort of shelf research university that I saw one of their, one of their, like, magazines that they published. And there was, and they were like advertising, kind of how great

Starting point is 02:24:14 they were. And on the front of it was like, you know, two unitary robots. And I was like, come on, guys. Like, honestly, like, we can do better as a country. You know, we have been the home for boss dynamics for, you know, over two decades. And I think to me, this is an opportunity where we should think about it in terms of the larger AI stack. And if you start building from chips all the way up, you know, the ultimate manifestation of all this is where a lot of these ultimately robots play out. It's this marrying of the physical infrastructure with the digital itself. And I think we need to do a lot more as a country to be able to propel that industry forward to make it economical for the adoption to happen. And it goes back to something I talked about earlier.

Starting point is 02:24:55 It's a lot about trust. You know, and I think there's, you know, if you think about kind of the dynamics oftentimes in the global marketplace, funny enough, the Europeans tend to be a lot more. more a lot more quickly, quick adopters, early adopters, a lot of this technology, just because of the issues that they have with some of their workforce. So I think for us, we want to create an environment where we can build these robots into our economic growth plans here in the U.S. And think very critically about the way that, you know, we want safe, secure, and trusted technology used by Americans in America. And if we see, you know, foreign companies that are not, you know, safe, trusted, and secure in our conference.

Starting point is 02:25:33 There's lots of tools that we have in our arsenal to protect our ecosystem from it. And the U.S. government, for example, banned D.J.I for use by the federal government. Then there's lots of other examples like that. Yep. Okay. Actually, last question. Do you fish? What's the biggest fish you've ever caught? We'd like to ask that to everyone that joins the show. Oh, man. I thought that was coming. I actually do not fish.

Starting point is 02:25:54 So if you guys ever want to take me fishing, I am totally. I don't. I haven't either. So we can go together and figure it out on the fly. Yeah, well, thank you so much for taking the time to come talk to us in your busy day. Always welcome. Congratulations on the Genesis mission. We're very excited for this. And have a great Thanksgiving. Happy Thanksgiving.

Starting point is 02:26:13 You guys, too. Let's do this again soon. See you. Goodbye. Cheers. Let me tell you about profound. Get your brand mentioned in chat. You bett reach millions of consumers who use AI to discover new products and brands.

Starting point is 02:26:24 Let me also tell you about TurboPuffer, service vector, and full-text search, build from first principles and object storage, fast, 10x cheaper, and extremely scalable. Founder Mode is right. We have Sebastian from Klarna in the Restream waiting room. We're going to bring him into the TVP and Ultradam. How are you doing? Great to see you again. It's been too long. The last time we hung out, it was just another, it was just another day at work for you. You just stopped by. NIC. You did your IPO and you jetted back home. Yes. But it's great to, it's great to be back. Yeah. Kick us off with a review. What's it been like being a public company CEO for a couple months now? Well, we said to ourselves on the management team, like, we're never. going to watch the stock price. Never, ever. And then every day it was like, what? Why didn't

Starting point is 02:27:08 go up? And then somebody like came to those guys just like put yourself next to all the other stock and like, oh yeah, it's moving the same way. Got it. Okay, got it, got it, got it. It wasn't just us. Yeah. Makes sense. Well, big news on Tuesday today. Please break it down for us. What's the news? No, I think it's, you know, I think I was, it's kind of funny. It's kind of Funny, because I, like, about a year ago, I was, like, giving crypto a second chance. I was always like, you know, I was a little bit of the, on the skeptic side over the years. And partially because, like, my lackmus test is, like, how is this going to help my mom, right? And as much as, like, I'm getting excited about, like, financial revolution, you know, kill the central banks or whatever.

Starting point is 02:27:53 Sorry, I shouldn't have said that. I'm going to re-quote. You know, bring financial liberty to the world, that's the way. And those things. The point is that, like, that doesn't get my mom much excited. She's like, what? Can I just pay cheaper or can this be faster? Like, she doesn't really care that much about.

Starting point is 02:28:09 The noise in the industry was insane. I mean, it was so many just regulatory arbitrage moments, so much just, like, fraud and grift and also just, like, pumps and dumps and all sorts of just, like, memes and things that weren't outright frauds, but had just, like, total. But the potential was always there. There was always a little bit. And the technology was always there, and it was a little bit. And I'm sure, I imagine, like, I'd love to understand how you, how you're thinking evolved on stable coins over time.

Starting point is 02:28:39 Yeah. Because back in 2021, what happened was some close friends of mine called Sequoia said, like, Sebastian, you really have to take a second look at this. And I was like, okay, fine, I'll do it. And they introduced me to some amazing founders among them, Bridge. that we now partner with. And so I gave it a second chance. And I listened to the stories and I realized, yeah, okay, so I've been wrong. Technology has now moving ahead. It's now fast.

Starting point is 02:29:09 It's now efficient. It actually solves real problems. And so we, and then I wrote that on X and I was like, nobody's going to care because we're like, we're the last fintech in the world. It's not going to make any news. But it was surprising, like over a million people views on that post. And since then, we've been just working hard. How do we integrate this into our stack and what can we do?

Starting point is 02:29:33 And today we announced that we're doing the Clona stable coin, which is one of many things which we're doing with Bridge and Tempo, which is, you know, together with Stripe and the team over there. So that's really exciting. But obviously, you know, there's more to come. It's just that like we're testing the orders and trying out what will work for us. So, yeah, walk me through what the first few implementation points look like. if I'm going and I'm paying and I'm on a checkout page

Starting point is 02:30:02 right now I can check out with Klarna at what point do I actually interact with Klarna coin the stable coin when does it actually come into the workflow so so far the start of this actually is just that like we obviously want to bring these services to the benefit of our consumers and in their day to day using a Klanah we have a big neobank ambition, right? And we have over 100 million users worldwide, but most of them will only use us for a single transaction. Thank you. Sometimes it's like, you know, a lot of those by now pay later,

Starting point is 02:30:40 but actually over 20% of them are debit, where people pay the full amount as well. So it's a mix. And now we're trying, you know, we're trying to offer ourselves more advanced services, more neo-bank services, right? So we have our card. It's pretty cool. Like, you know, we are now three million active card holders a quarter ago we had zero in the US so exactly I'm going to drop a lot of numbers and get a lot of beeps so that was like yeah so that's been great but obviously we want to add peer to peer we want to add the ability to send money and so forth and then we'll be looking at different solutions but but what's interesting I think here is that like even though that's the natural next step of you know offering uh using stable coin for that because it's a very

Starting point is 02:31:23 efficient way to send money fast and at low cost. We also actually realized that, look, I mean, we're processing over $100 billion worth of volume every year, and we move ourselves quite large money between the U.S. and Europe and so forth, and this may actually be very efficient even for our Treasury Department to use as a way to move money, right? So we've seen that there's a tremendous number of use cases, maybe more than we realized when we started looking into it. That's exciting, and that can drive.

Starting point is 02:31:53 You know, anything that obviously drives our efficiency will also allow us to offer our services at lower cost and more value to our customers. Can you break down specifically what's happening at each kind of level of the stack? So you have bridge, which is from my understanding, the infrastructure that you're using to actually issue the token. Then you have tempo, which is the blockchain in which you'll be moving it around on. And then you have obviously the product layer, which is the interface. that consumers will use to move around. Is it USDK? What is the ticker?

Starting point is 02:32:31 I think it's used decay. I think you've described it pretty well. I was like, thank you. That's a good summary. I'm not trying to understand. And then you also have, there's a corporate treasury use case where you're saying you could use the same token internally as well. Yeah. I mean, yes, exactly.

Starting point is 02:32:48 I think also then people discuss a bit on the merchant side, but the truth is that we mostly today distributed through Stripe and Adyen and the big PSP. So most of the kind of merchant relationship happens directly with them. So to us, it's mostly what we can do on the consumer side. But I think there's more opportunities. I mean, this is just like,

Starting point is 02:33:06 again, people are like, oh, why did you do it this way and not that way? Why didn't you use Solano Ethereum? And why didn't you, you know, whatever. And what about Bitcoin and so forth? And to me, it's like, yeah, but come on, this was just like one first thing that we now announced. And obviously we are working on more things and we want to utilize these technologies

Starting point is 02:33:21 to more use cases. And we have more things, but we are not yet yet to announce them and talk about them. So more to come. Fantastic. Last question from my side. Who do you think is, or what, you don't need to say a name of a company, but like, what type of, what type of player in the financial industry loses out most if stable coins really take off? It feels. like at a certain point, we're just shifting take rates around. There's a lot of different transfer services that could benefit a lot from higher speeds and lower fees.

Starting point is 02:34:03 It seems like consumers could benefit a lot. But do traditional payment rails suffer? Do government suffer? Like, what category really needs to be on their back foot right now? To me, it's really, and I actually talked about this a little bit on our earnings goals as well, is that I actually think, like, to me, I put the whole crypto thing in this wider AI. change that's about to happen, right? And if you look about it, both Finn and tech have been extremely inefficient market, which is why you've seen the excesses in profits and excesses

Starting point is 02:34:35 in, you know, as I quoted there, like, you know, the gourmet cafeterias used to be called culture and now they're going to be called overhead. The point is that like there's been this huge success because it hasn't really been that strong competition. And why has there been lack of competition goes back to kind of classical microeconomic theory, which, states that if it's hard to compare to products, if it's, if, you know, too many legal terms that are hard to understand and so forth, then people will be fooled and will take products that are less good for them, right? And AI and new technologies brings a fantastic amount of transparency. You can just ask, compare these insurances, compares these banking products,

Starting point is 02:35:12 what is the best one? Compare these credit products, which one is actually the best one for me, right? So that's going to change. The second thing is the major one, which is very connected to crypto is switching costs. So the big reason banking isn't more competitive is simply because it's such a hassle to move. You know, like I want to bring all my stuff from this bank to that bank. And people just don't have that energy, right? People don't have that energy and all the accounts and the salary coming in and all that stuff, right? So what happening is with technologies like crypto, with AI and the combination of that, what you're going to see is going to see switching costs coming down to almost zero, right? It's just going to be much

Starting point is 02:35:50 easier to move between different services. And it's just going to put this massive competitive pressure on this industry. I mean, financial services make over $500 billion in profit. That's the profit pool that we're after. And then tech in advertising and tech is another $500 billion. That's a trillion dollars in profit pool. And so what I think is going to happen is you just like, if you are willing to be customer-obsessed, efficient, you know, and with efficient, efficiency, operational efficiency, crypto plays a role because it can help you be much more efficient. You can move faster at lower cost and so forth. Then the players who take the advantage of these technologies are going to make a massive dent. And the benefit, I believe,

Starting point is 02:36:36 and maybe you can call me naive or you can call me, you know, optimist. But like, I believe that the value of this is all these excess profits that we see in banking, they're going to come back to consumers. They're just going to make consumers are going to pay less for higher quality services. And some financial institutions will lean in and they will transform themselves and they will be more customer obsessed. There will be less marble offices, less beautiful, you know, buildings and all that stuff. And there will be more of just like waking up every morning, just like a restaurant or any other business for the sake, where you actually wake up every morning, you're like, how do I bring my customers in? How do I make them happy? And how do I

Starting point is 02:37:11 run my business so I'm operationally efficient, right? And that's a huge difference. And I think crypto plays a role in that. So I think that's what's coming for Finn and tech. And if you're willing to be that customer obsessed and run those operational efficiencies and be smart and use these technologies, there's obviously tons of opportunity of growth and you can build trust with your customer base. But if you're going to sit there and still continue making the kind of profits you are today and think that nothing's going to change, well, eventually you'll wake up, right? Just like, you know, whatever, airlines did when low-cost airlines came in. We've seen this over and over again. They're like carriers when

Starting point is 02:37:44 the low-cost carrier things came in. Like, it's happened before, right? yep well thank you so much for taking the time to come chat with us we will talk to you soon have a great rest of your day thank you good to see you talk soon cheers uh before we bring in our next guest let me tell you about public dot com investing for those who take it seriously they got multi-asset investing and they're trusted by millions i'm also going to tell you about numeral dot com let numeral worry about sales tax and vat compliance compliance handled so you can focus on growth we have david from one password in the restream waiting room. Now, he's in the TBPN.

Starting point is 02:38:21 He's back. Welcome to the stream. Welcome back. We have a big update for you. Since you came on the show last time, we onboarded the whole team onto one password. John, we were using it individually. There was some passwords being stored in the notes app. We corrected that.

Starting point is 02:38:36 Yes, yes, yes. I'm glad I was able to shame you into action there. Yeah, yeah. This is real like customer development, getting your hands dirty, rolling up your sleeves, you go on the podcast with 10 employees and you physically tell them, download my app. And it worked. It worked. You can't. But anyway. One customer at a time. Thank you again for your trust, both with your business and your personal life. We're, yeah, we're very happy. It's

Starting point is 02:39:02 been a huge upgrade. And I think it's going to sail to being a, you know, valued piece of infrastructure in this organization for many years. But we're also very excited that you've been on a tear. Your business is doing better than ever. Give us a little update. What's new in your world? Yeah, so, you know, we had a little announcement just to refresh some of the statistics of the business. So a couple weeks ago, so we are over $400 million of ARR. Whoa.

Starting point is 02:39:30 There you go. Love it. The spinning hit. Spinning head for 400. We've done that. We've grown to that level profitably all the way from inception, which is really cool, you know, durable growth profile. We continue to be very profitable. We're known as a consumer app for a long time, but now nearly.

Starting point is 02:39:46 80% of our business is selling to business customers. 180,000 business customers use the product. 180,000 and 1% because of you guys, so thank you for that. And, you know, we support over 1.3 billion credentials that are managed in our system. And we have, you know, over a million developers using the product. So, you know, we feel really good about the trajectory that we've been on. We've also, in that announcement, we've also announced that we've added to the team. So we've got new leaders driving the ecosystem and the revenue.

Starting point is 02:40:16 side of the business, enhance the tech side as well. We're really just setting ourselves up for the next chapter here. And, you know, as we chatted last time when we were together, we just see the opportunity around AI is really something significant for us. And we're going all in. Yeah, I want to talk about that because it feels like there's, is there a fight coming? Because credential management on the developer tool chain side, the API keys, that's traditionally been a completely separate industry, a completely separate, you know, I don't even know if you guys run into each other at industry conferences. It feels like a very different world. One's a developer tool. One's a consumer tool that you use with your family. Then maybe you wind up using it

Starting point is 02:40:57 at work. There's single sign-on. There's all sorts of different ways to manage credentials. How much are they blurring? How quickly is this happening? How are you thinking about setting yourself up for the world where? I mean, Shulto from Anthropic was on the show yesterday saying, we're creating a co-worker. Digital co-worker. Sounds like a digital co-worker who needs some passwords every once in a while.

Starting point is 02:41:19 How are they going to get it? Absolutely. I think the traditional identity systems that have been siloed and sort of different parts of the puzzle and really hardwired into big enterprises over the years. They don't really know.

Starting point is 02:41:30 They're not set up to support this federated dynamic environment where applications spring up and agents need access that are very, very specifically scoped to what they're trying to accomplish and then they need to be sort of tracked and managed and many times revoked.

Starting point is 02:41:43 And so that system of identity and access management really doesn't exist, and we're in really good position to support that. And we're doing a bunch of things already to help people use AI tools securely. I'll give you a couple examples. So we were a launch partner with perplexities, Comet browser. I'm not sure if you've seen that. But then shortly thereafter, Open AI and ChachyPT Atlas came out, and immediately both they and we got besieged by a customer saying, I need to be able to use one password with Atlas. Of course, of course.

Starting point is 02:42:12 I use this thing and listen it. So we got on the phone with them and we started working together with them. And quickly we made one password available as a browser extension on Atlas with a little bit of a workaround. And then just today, we announced the release of a fully optimized, you know, onboarding experience, optimized for the workflow experience and the Atlas browser as well. And so what we learned is like, look, customers, they want to use these tools, but they want to be able to use them safely and securely. And they want to be able to have trust. So that's what we've always been about is bringing that. trust to the table. And so we'll be wherever our customers need to be. Last time we were talking

Starting point is 02:42:45 about the headless browsers around a browser base. But like with perplexity and Atlas, you know, our customers want to be able to use these tools without having to give their credentials over in raw form. So that's one aspect of it. The other is what's really interesting is, you know, we're moving from applications being developed by software engineers to being developed by everybody. And it's amazing how much of the agentic applications that are being created are not necessarily by software developers. And so it's great. It's very powerful. The whole organization can figure out ways to leverage it. But again, if you need API keys to be part of the, you know, in the environment that you're building, or SSH keys, you need to be able to actually

Starting point is 02:43:26 have those things secured. And so our developer tools, they sync with AWS secrets manager at runtime. They allow you to put a pointer to environment file tools that we build so that your credentials are always encrypted, end-to-end, fully traceable, fully revocable. And so that's the beginning of what you'll see. And then, you know, where we're going from here, you know, discovery of agents, governance around it. How does it interact with the observability systems? There's a whole green field of things there that we're really well positioned to participate in and really help define the standards. So, you know, we're going for it. It's a very exciting time. How are you thinking about resiliency?

Starting point is 02:44:03 Do you have any tips for folks building, you know, mission critical systems these days? It feels like we were fighting an outage with ADWS at one point. Cloudflare went down. We don't have nearly the resources that you do, I'm sure. But what tips and strategies have you used? Because I feel like you're definitely going to hear if one password goes down, right? Yeah, you know, look, we are a Cloudflare customer. So we, our marketing site was down.

Starting point is 02:44:33 Our service wasn't down. Sure. So, you know, we are using, you know, obviously do redundancy and failover procedures. And we've got, you know, we come from a security-centric approach. That's probably what's differentiated us from a lot of the other players in the space. Like security is job one and job 10. And like nothing else matters until we're sure that everything we're doing is the utmost security. So we've got a lot of resilience built into the way we do it.

Starting point is 02:44:59 But I think just generally, people need to be aware, like the threat profile is so much more hostile than it's ever been, and it's going to get a lot worse. I'm sure you did a bunch of work on the disclosure that Anthropic put out around that. And it's not just, I mean, Anthropic, you know, as you know, they have like nation state level security, you know, enforcement. They have like bunkers and all kinds of stuff that they've got to worry about, right? Because they're, you know, hugely a big target. But it's not just the biggest guys anymore. It's really, anybody can be a target because you can do really sophisticated social engineering at scale with almost no effort at this point. And it's going to get worse and worse and worse.

Starting point is 02:45:37 So the number one thing that people, you know, will expose is weak credentials. It's another really good spot for us. Like, credential hygiene is more important than ever. So whether it's you're using Atlas and you're a one password customer, great, make sure it's connected. If you're not a one password customer, you should go get it if you want to use Atlas, right? And if you're a developer and you're building an application and you're not and your heart, hard coding credentials or SSH keys, et cetera, into the files, don't do it. Get a one password license, super easy to use, keeps you secure, keep everything ever lacked up,

Starting point is 02:46:07 and maybe we'll get another new customer out of this. I love it. What's your timeline to a point in time where the average American on, let's say, a weekly basis will be sending an agent out to do something for them, and the agent needing to actually utilize the, you know, like actual login infrastructure of the user. I'm talking about imagining the workflow where my agent's doing something and I get maybe a one-password push notification that's, or I don't know exactly how the workflow would work, but saying, hey, I want to do this. Do you approve it or not? And then I hit like, yes, like basically

Starting point is 02:46:47 authorize the agent. So that's exactly what's happening with browser base, for example, right, in the headless browsers that we're doing exactly. It's low friction, user verification. so that allows you to keep things secured. Look, we're seeing tremendous uptick on the, you know, Atlas was a really, was a enlightening to see how many people are actually trying to sort of utilize the browser and have it do things on its behalf. Certainly, the other partnerships we've had, we're so early days on it. But I think when we're going to hit this inflection point where people realize, I mean,

Starting point is 02:47:16 I'm starting to see how valuable in both my professional and personal life the tools can be and just shortening my day. And obviously, you know, I want to make sure I'm doing it securely. I think most people are going to be in that mode before we know it. And so we just want to make sure we're available for our customers when those moments come so that they can do so securely. Last question. What's the biggest fish you've ever caught? Well, I'm more of a fly.

Starting point is 02:47:42 I enjoy fly fishing more than I do like deep sea fishing, you know, getting sick on a big boat and hauling and deep sea fishing is the ultimate is the ultimate hack if you just want to catch the biggest fish, right? It's not, but it's not a pretty simple. And somebody else catches it. Yeah, really did. Yeah, my son, my son prefers fly fishing, so I, I follow him. He likes to catch a big bass. Where is the best fly fishing spot? You know, I like Montana and Utah.

Starting point is 02:48:07 Utah. You like Utah? Utah and Montana. Utah, a fish are a little smaller, Montana a little bit bigger, but, you know, two really good spots. And, you know, just, you know, give me. We're going to become, we're not, we're never going to leave this building, but we're going to become fishing experts. We're never actually going to.

Starting point is 02:48:21 Just through our gas. But just through. asking every guest a little bit more about fishing show. Eventually, this will just be a fishing show. I like it. I like it. Get yourself a guide. Go on a fly fishing trip. Take a weekend. Go to Montana. You'll love it. Thank you so much for coming on the show. Happy Thanksgiving. We'll talk to you soon. Let me tell you about Vanta. Automate compliance and security. Vanta is the leading AI trust management platform. One thing, one thing that's cool, back in 2019, one password raised from Excel. And when they announced it, they said Excel will be investing

Starting point is 02:48:54 $200 million for minority stake in one password. And you don't see that positioning a lot, but I mean, it was a very minority stake, obviously. They put in about 200, I think,

Starting point is 02:49:09 at was it 200 on one or $2 billion. But anyways, absolute beast of a business. Well, up next we have Keller from Zipline. We're very excited to talk him, there's some big news.

Starting point is 02:49:23 The eagle himself. Yeah, I get that ready. First, I'm going to tell you about Figma. Think bigger, build faster. Figma helps design and development teams build great products together. And there's someone who built a great product with lots of talented individuals. Keller from Zipline will bring him in as soon as he's ready. Oh, I don't think he's ready yet.

Starting point is 02:49:43 But we can go over to the timeline. Yeah, we have a post here. Uh, somehow. No, I'm going to jump. Okay. Somehow, David Yulevich, former guest of the show, said, turns out huge swaths of the BS on this platform that claims to be from the U.S. is all foreign.

Starting point is 02:50:01 And then this is Fekiel quoted, and somehow the A16Z account is based in Canada. I really don't. I really don't know how this is actually possible. I want to see if it's been updated. I did check all of her accounts, and I think we've been marked safe. I haven't checked actually most of the team, Tyler.

Starting point is 02:50:18 I don't know. Maybe just secretly. I just know so many I have so many friends at Andresen. I don't know how Why does Tyler's... I don't know, none of them are Canadian. Why does Tyler's account say it's based in North Korea?

Starting point is 02:50:29 I don't understand that. I know. Have you been to Pyongyang recently? Is that where you're going over the holidays? Yeah, you said you're leaving. This is miss info. He goes, oh, I have a dentist appointment. I can't come in tomorrow.

Starting point is 02:50:39 He can't come in. He can't come in. He's just Pyongyang. For a little touch base. Yes, yes. What, uh, what, what, what happened with this, uh, meta, uh, uh, whistleblower sister, uh, situation. Adam Delt says, I'd love to see TV.

Starting point is 02:50:54 Well, you're not going to see it. You're not going to see it. You're not going to see it. I do want to come back to this. But we have our guest here. So we will bring in Keller from Zipline. Welcome to the show from the floor. Fantastic set up.

Starting point is 02:51:08 How are you doing? Best, the team in the game. Good to see you. How are you guys? We're fantastic. Uh, you look great. Uh, give us, give us some updates. Give us, uh, what, uh, what is the, uh, what is the

Starting point is 02:51:20 latest news in your world. Very excited for everything that we've been seeing on the timeline. I want to go into all the jokes about what happens if you shoot one of these down. But first, let's get the serious update. Let's get the serious questions out of the way. I mean, yeah, there's a ton going on. The thing we announced today is that Zipline just signed a $150 million contract with the U.S. State Department. Yeah. You know, to triple the size of our life-saving autonomous delivery network across Africa. We're going to go from serving about 5,000 hospitals and health facilities to over 15,000. We'll add about 130 million people to the

Starting point is 02:52:01 network who don't have access today. And the last cool thing about that is that that 150 million comes with up to $400 million of co-financing commitments from the African governments themselves. Okay. So, you know, this is not, this is not- Just straight on the taxpayer. Yeah, exactly. This is actually, like, encouraging investment in this kind of infrastructure. Yeah, so was this delayed because of the USAID pullback? There's been a lot of, like, back and forth on how much the U.S. would be investing internationally, what would be happening. Has the dust kind of settled there and there's now time to actually go do a partnership like you just did? What's this kind of state of the union?

Starting point is 02:52:39 Yeah, I mean, you know, obviously earlier this year, that was like the first phase of a big shift in terms of how the U.S. is interacting with, developing economies. I think this new vision that's now being talked about by the State Department of commercial diplomacy. The whole idea is let's not, you know, let's not treat these countries as charity cases. Let's actually treat them as allies and trade partners. And the good news is these countries have been saying for a decade that they want trade, not aid. They're sick of low quality services provided for free by NGOs. What they want instead is high-paying jobs, entrepreneurship, technology. And the thing is, like, that's what the U.S. has to offer. You know, like we can, we can deploy AI and robotics infrastructure into these countries in a way that will save lives and kind of turbocharge their economies.

Starting point is 02:53:26 So 150 million from the U.S. State Department matched with 400 million from the partners internationally. 10,000 health facilities. How many actual drones is that? How are you actually thinking about scaling your, like what does this allow you to do? Are you going to be staffing up, hiring tons of people? Is it just build another factory? Is it a dedicated factory? Walk me through how you plan to actually use this money over the however long you're planning to work on this particular growth initiative. Yeah, I mean, look, in a way, this is coming at the worst possible time because, as you guys know, the U.S. business is growing really fast. We're in the middle of, I'm here. Yeah, I'm here on the manufacturing floor. I mean, you can see like dock electronics happening over here, Zip Platform 2, Zip Manufacturing.

Starting point is 02:54:14 And then if we were to go straight behind the camera right now, we're just opening up a new 100,000 square feet of manufacturing facility that will produce both the Platform 2 technology, as well as accelerate manufacturing for this specific contract. The, you know, we're in total, this manufacturing facility is capable of building about 20,000 autonomous aircraft a year, and it's all happening in South San Francisco. So this is kind of, this is the cool part of this,

Starting point is 02:54:44 compact, which is that this is not just a big deal for all these African countries that are now leading the world in terms of deploying autonomous infrastructure to save lives. It's also really good for the U.S. because we are sort of securing U.S. technology and manufacturing leadership for the decade to come. This is creating high-paying jobs in the U.S., and it's accelerating all of our manufacturing efforts here. Okay. Let's shift to what's actually going on here. you know, we've, we've seen the initial partnerships that have rolled out. What have you learned most recently about, or what has surprised you about the actual application of the technology, the adoption?

Starting point is 02:55:22 Like, where are the underrated use cases beyond the meme that I'm sure will follow you forever, which is the private jet for your burrito? So get ready for that one to be around forever. I mean, we just had Sebastian from Klarna on and his entire, you know, business, which serves all sorts of different customer bases has collapsed to buy now, pay later for your burrito because they did a partnership with Tritos are powerful. I think it's actually a bullish indicator if your business is getting burrito memed. It is, it is. But yes, where do you see the shape of the U.S. business going? Where are the exciting developments these days? I mean, look, you know, on, you know,

Starting point is 02:56:03 Saturday, Zipline hit an all-time new record number of deliveries across the U.S. Then on Sunday, day we blew that record out of the water by 10 or 15%. So, you know, it's like every day is basically a new record at this point as a company. I would say the most interesting things are, it feels as though we accidentally like stuck a pipe into the Pacific Ocean. Like demand is so vastly outstripping our ability to build capacity for this kind of service. And keep in mind, I mean, right now, you know, we deliver products in two to three minutes. So even from like when a customer presses a button to has something delivered to their front yard or backyard or front doorstep is typically around 15, 16 minutes. And so it is very magically fast. That results in customers changing their ordering behavior.

Starting point is 02:56:51 Like, you know, a lot of our customers say they'll grocery shop once a week and then they'll order from Zipline three to four times a week. So people are using Zipline more than the average Amazon Prime subscriber uses Amazon Prime. And not only that, but one of the big changes just since, you know, I was last on with you guys is that a lot of these restaurants are really accelerating. We're now doing 15 to 25% of deliveries from these restaurants that we've integrated with. So we are a similar size on a per delivery basis to the big delivery platforms. So I think, you know, maybe people still think, oh, the technology is like sci-fi. It's like years away. And it's funny sitting here thinking like, wow, I mean, this is like completely normal in the neighborhoods that we are, that we're

Starting point is 02:57:35 serving. Talk about the actual experience of getting, let's say, items from a restaurant. We had David Chang on. He was excited about drone delivery and the technology. I tweeted at you guys. Yeah. We're going to, we're going to try to get it. We're going to try to deliver for him in Dallas. Yeah, yeah. And specifically he was saying he didn't think it would solve the, the, like delivering something hot from my point of view. If I'm used to getting something in 20, 25 minutes, and then you deliver it in three minutes, like that feels like it will be a material improvement. So I didn't quite understand what his point was.

Starting point is 02:58:12 But what's the actual experience been like for people and what's kind of the average speed up in terms of delivery times that you're seeing? I didn't really understand what David was saying either. But look, like, our comms team hates when I talk about this. But, you know, like one of the first food deliveries we did, the customer badly burned their tongue on the food.

Starting point is 02:58:35 Okay, yeah, I can see why they don't like this. That's the restaurant's fault. They served food at a dangerous temperature. Yeah, yeah, yeah, yeah, it's kind of a bad thing. I was like, you know, actually, we should make this like a core part of the marketing. Like, that is crazy that customers are so not used. They're so used to getting, like, fries. They're 45 minutes old and, like, ice cold and soggy and gross, and food that

Starting point is 02:58:57 taste bad and milkshakes that are melted. And we just put up with this shit, man. Like, we're completely used to it. And so, you know, I really think this is going to, you know, in the same way that maybe David's expectations are going to get kind of reversed, like it's totally happening with all the customers. When you can get something delivered and there's only, you know, two or three minutes from when the thing comes out of the oven, when it is delivered to your front doorstep, that is a way different customer experience. I think actually people are underestimating the difference and the quality of that experience. Like food tastes really good. It's almost like you're having an in-restraint experience at home.

Starting point is 02:59:34 I wonder, so I remember when the delivery boom happened and all of a sudden people were ordering way more food. And it was really like a Tam Expander. There were obviously ways to get delivery. You could order a pizza 30 years ago on the phone. They would deliver it. But when it became an app and the market expanded so dramatically and people were starting to, you know, DoorDash and Uber eats like every night.

Starting point is 02:59:58 Lots of people did this. the market grew and it actually changed the nature of our food and we got ghost kitchens and we got these funny knockoffs like there's in l.A there's sugarfish and then I saw one that was like sweet it's like they will deliberately try and change the names as close as possible and you get these like kind of restaurants that only exist because of the new technology and I'm wondering what you think of the impact if there will be a change in the nature of the food or the nature of the restaurants just because the delivery time goes down

Starting point is 03:00:31 or because of the shape of the box. Like if you can't do a massive, you know, six-foot pizza, maybe, you know, some smaller products like emerge is like the dominant form factor and the restaurants that conform first. Six-foot pizza, John? I don't know.

Starting point is 03:00:47 The size of this table, I don't know. Yeah, yeah, I, you know, the kids' birthdays. You see those? Yeah, yeah, they're pretty big. It's not six feet. It's not six feet. But you get my point. You know, will there be restaurants that are like, okay, we're going all in on Zipline?

Starting point is 03:01:00 We're thinking about developing meals from First Principles that fit for the best possible experience on the Zipline platform. And then that becomes like a new thing that actually changes our culinary experience. I mean, I, you know, I'm not an expert, but I have a bunch of like hypotheses. And I mean, I would say a couple things. Like, first of all, I actually think we have to do a ton of design, you know, design of the food. now to make it work for a system that, like, is going to deliver the food 45 minutes later and ice cold, you know, and, like, and also there's a lot of innovation that has to go into the packaging, because I don't know if you're familiar, but, like, 60% of food delivery

Starting point is 03:01:37 drivers report eating some of the food that they delivered in the last half the days. So we're, like, doing all the tamper food packaging. 60% over, like, hopefully over the month, not 60% of every time. Or the last month. Not 60% every time. Okay, okay. Over a month. They're like, but only one is the idea when it's really good.

Starting point is 03:01:54 Only when it's really good. Only when it's like my top five favorite restaurants. One fry. One piece of sushi. Yeah, sorry if I'm ruining it for anybody. But like, you know, so we have to innovate on like tamper-proof packaging. We have to figure out like safety and we have to figure out reliability. And like, you know, some percentage of things just get picked up and then never delivered.

Starting point is 03:02:12 And a lot of things actually like the driver will accept the order, but then actually just never show up at the restaurant. Then they have to throw out the food, remake it again. So, you know, I think there's a ton of inefficiency and cost in the system that we will just don't think about. But you do end up paying for in fees and, you know, all the crazy fees that get added on top plus tip. I think another big thing that's going to change is that, you know, this technology is designed to be able to fly out to 10 mile radius. And typically it's not cost effective to deliver something with a car more than like three or four miles. And so that means, you know, it differs on by Metro and where we're talking about. But like that means that these kinds of systems for any given restaurant, you're going to enable about 10 times as many customers.

Starting point is 03:02:53 to receive instant delivery from that location. That's a big deal, and it's a bigger deal actually for, like, mom and pop restaurants or kind of local heroes versus a really big chain like Chipotle that actually has a ton of stores. And so I think, you know, especially when you think about a place like L.A., for example, where you have a lot of these different parts of the city that are hard to reach. Traffic is terrible. You know, it can take an hour plus to get delivery of food. You know, this technology is going to be, like, transformational.

Starting point is 03:03:20 And you can deliver a lot of like the best, most beloved kind of local hero brands and make it universally accessible. By the way, on the university accessible front, you know, some of the cities were serving in Dallas right now. I mean, just one city that I was looking at statistics for yesterday, 46% of homes are ordering from Zipline in that city. And so I think people don't appreciate, yeah, exactly. I thought it was like a map error when I looked. You know, people that talk about startups getting like 2% market share, 5% market share. It's like, Zipline, in a few weeks, Zipline will literally have the majority of homes in our service area ordering from the service. And that is like, that's not a joke.

Starting point is 03:03:57 How are you, so, so clearly, clearly it's working. Clearly customers like it. There's an insane amount of demand. How are you pushing the team to move faster? Because at the same time, yeah, I know, I know you want to, you know, be the practical about the rollout and not get over your skis. But at the same time, I mean, people, people in our check. chat, or anybody that sort of knows that exists, they start to become angry that they don't have it in their area. And I'm sure there's people sending you messages all day long asking

Starting point is 03:04:31 you, when are you going to be here, when are you going to be here? So, yeah, we're going to be announcing something very shortly. We just started talking about it internally, but we'll be announcing the next two metros that we're launching in Q1 and Q2. So look for that in the next week or two. You know, we are then going to start accelerating and adding multiple metros a quarter, starting in the second half of next year. So we are trying to move super, super fast. Hardware companies are hard in that you have to scale all these different parts of the business simultaneously. It's in the name. And also it's like you got to make really sure that you don't accidentally bankrupt the company. And actually, the faster you're growing,

Starting point is 03:05:11 the easier it is to bankrupt the company. Totally. Because like, you know, these numbers, when you're building 20,000 autonomous aircraft a year, the numbers actually start to get really, big. You've got to make sure that like the hardware is validated and like you have, you know, sites that these, that the hardware can go to and begin operating. You got to make sure that aircraft utilization is high. There's, um, there's a lot that has to move in tandem in order for the business to work. Uh, so I mean, to put it into perspective right now, flight volumes have been growing about 15% week over week for the last 30 weeks straight. We are now planning for next year. Yeah. We're planning for next year and we actually

Starting point is 03:05:49 just reforecasted the entire business and we are planning to actually double our growth rate next year relative to what we were planning just at the beginning of this quarter. It's, yeah, and I don't know how much I can, you know, we are, we are definitely anticipating like exponential growth over the coming couple of years, like more than 10x. At this point, it's fantastic. Yeah, exactly. Just send it straight to Wall Street. I appreciate it.

Starting point is 03:06:17 So exciting. Congratulations again. It's always a great time. You guys got to come to Dallas, I think. Yeah. I think it's probably at a point now, you know, when I was in Dallas last week, like, we were just hanging out at all these different, you know, these different neighborhoods that are using Zipline. And there were neighborhoods where I saw more zips than I saw cars. Wow. And so I don't, you know, it's funny how the, you know, the futures here is just not evenly distributed.

Starting point is 03:06:40 I think there are, you know, certain parts of the U.S. people are like, oh, yeah, it's sci-fi. Like, you know, it's totally goofy. And then there's like, yeah, some metros are people like, oh, yeah, I use it like three to four times a day. By the way, these are like predominantly moms and grandmas. There's a lot of AI researchers that say that getting Waymo into DC will significantly update the administration's AGI timelines because they will just see robots on the street and they will recognize that this is a real thing. And until you actually see it diffused in the world, it's very hard. It's very abstract to just see reports. Oh, Waymo has a thousand rides or 10,000 rides or a million rides.

Starting point is 03:07:18 It's like, what does that mean? But when you are walking and it's like, oh, seven waymos in a row and they're all driving fine, not crashing into each other, you get it. Even Tesla FSD. I mean, I was hanging out with my college professor and he, like, had no idea what FSD was or that it was working. We've heard about it for so long, but it's finally here. So, yeah, very exciting. Well, thank you so much for coming on the show. I hope you have a great Thanksgiving.

Starting point is 03:07:42 And we'll talk to you soon. I'm sure you'll be back soon. Yeah, you'll be busy delivering turkeys. Hot. Whole turkey soon. Small ones we can do. Small ones you can do. Okay.

Starting point is 03:07:51 Great. I love it. Well, congratulations. We're awesome. Great to catch up, Kellogg. Have a good one. See you. Cheers.

Starting point is 03:07:56 Before we bring in our next guest, let me tell you about Julius AI, the AI data analyst that works for you. Join millions who use Julius to connect their data, ask questions, and get insights in seconds. And I'm also going to tell you about Privy. Privy makes it easy to build on crypto rails. Securely spin up wide label wallets,

Starting point is 03:08:13 sign transactions, and integrate on-chain infrastructure all through one simple, API. We have our next guest in the Restroom Waiting Room, Royce Branding from ClearSpace. How are you doing, Royce? Welcome to the show. Good to have you. How are you doing? Thanks for having me. I'm doing well. How are you guys doing? I'm doing well. Can you kick us off with just a little bit of intro and background on you and the company? And then there's a whole bunch of questions that I'm sure we're going to get into. Totally excited to dive in. My name is Royce Branding. Co-founder and CEO of ClearSpace. And we help people reduce their screen time. Our company, our mission is to build a missing intentionality layer of the internet. We think there's a fundamental asymmetry in the war for your attention, and we're equipping the individual autonomous consumer in the market with the ability to steer their focus online and protect their family's attention online. What's a perfect amount of screen time?

Starting point is 03:09:05 Have you guys run the numbers? Is there an optimal? Well, you know, Andrew Ross Sorkin asked Peter Thiel this in Sun Valley for his kids. He was like, you're on the board of Facebook. How long, how much screen time do you let you say like zero? He's at 30 minutes a week. Oh, yeah. 30 minutes a week is ambitious.

Starting point is 03:09:24 I think that the ideal amount of screen time is probably a little bit less than whatever you're spending right now. Sure. We say that a calorie is not a calorie is not a calorie. A minute spent on a screen is not equivalent to another minute spent on a screen. Sure. But you want to be increasing the quality of the consumption. And usually overall reduction in quantity ends up meeting an increase in quality.

Starting point is 03:09:45 Yeah, that makes a lot of sense. So how do you actually do it? I mean, there are screen time functionality natively baked into Apple. The iOS, I'm sure Android has it too. Your business is still working, so it's not like you're getting steamrolled, but how do you differentiate?

Starting point is 03:10:02 Yeah, do they have, is there any sort of misalignment to Apple's business with screen time? Because I imagine if they really incentivize people to use their screen less, it's like less apps that I'm subscribing to and less, kind of potentially less, like, they don't have a perfect incentive to make you use your phone less.

Starting point is 03:10:20 They sell you the phone once, but they're like, take more pictures, get those, get those iCloud subscriptions up, buy more apps, spend more in that mobile game, et cetera. 100%. Yeah, yeah, we think Apple has every incentive to have a solution here, but maybe not the ideal incentives to have an effective solution to helping people kind of control their screen time and reduce it. We think that it comes down to both visibility and control. So you have like a feedback loop that's helping people identify how much time they're spending on devices and like where that's actually going. I think the current state of screen time right now is basically like again to use food if you only tracked calories that you consumed, not micro and macro nutrients. So for the last two years coming out of YC, we've been primarily focused on a mobile app that grows people's awareness of how much time they're spending on devices and then allows them to add little bits of cognitive friction into the addictive habit loops that they use on different apps like social media.

Starting point is 03:11:14 And that kind of starts to get them rehabilitating an addictive habit. But what we're actually working on now, and we just announced a few weeks ago, are screen time agents that sit at your network layer and can observe network traffic going to all your devices. And this is particularly useful both for individuals and for their families to think about holistically what it means to have the right type of content, the right time of consumption happening across all of their different devices. And that kind of breaks out of just Apple's ecosystem of what they allow us to see. Yeah. Yeah, that's interesting. How do you think about the responsibilities of different big tech platforms? We were just pulling up this article in CNBC about this whistleblower claiming that meta failed to act to protect teens. Do you have a stance on like, are the big tech platforms being negligent in what they're surfacing to their users? Or do you think it's more just like the natural forces of, you know, the economic incentives. They're trying to sell ads. How do you, how have you grappled with the various, like, dustups in the big tech world over the, like, what happens on these apps? Yeah, yeah. I tend to think about no one is the

Starting point is 03:12:33 boogeyman. I think actors are kind of following the incentives laid out before them. Our mission is really much to meet the opposite side demand. I mean, I don't know about you guys. Everyone I know hates their relationship with their phone. They hate how much time they spend on their device is. And I think what we're seeing is we think that actually equipping them to articulate that preference in terms of how much they're spending time on these platforms with technology that's sophisticatedly protecting their attention is going to actually push back against how big of an intention economy scenario can actually be downstream from people just rampantly consuming content. Yeah. Yeah, that makes sense. So how do you actually

Starting point is 03:13:13 take me through a list of like the walls that you run into. I mean, all the big tech platforms are like famously walled gardens. Somebody in the chat was asking, like, I would love to be able to use YouTube with no shorts, only like long form content. And I actually installed an extension in Safari called Social Focus that allows you to do that. If you use the browser, but if you're in the app, there's nothing you can do. It sounds like you're going higher. at the network traffic level a little bit, but at the same time, I don't even know, can you

Starting point is 03:13:48 at the router level detect if someone's watching a short versus a long form video on YouTube? That feels like that would be obfuscated. Yeah, yeah. That gets a little harder to do, although we're constantly pressing at the edges of like what is possible around that. I mean, we've definitely ran some experiments where you can kind of fingerprint what short form content might look like coming down the wire versus long form content. And so there is like interesting things that you can start doing there, but really robustly for the mobile app, we can basically just like make sure that you're staying within healthy usage limits and that those aren't sliding out of control, which like, when shorts go wrong, what really happens is they

Starting point is 03:14:25 like pull you into a longer session than you want to be in. So there's not an easy way to do that, but different platforms have different rules. Like on obviously Chrome, on the MacBook, there's a way higher ability to detect what type of stuff you're watching. Yeah, yeah. It's, I mean, it's tricky, especially as all the apps, like, collapse with AI. You could imagine, I mean, right now, SORA and chat GPT are separate apps, but in many ways, they're like on the opposite ends of the spectrum.

Starting point is 03:14:54 Like, I want, you know, my son probably spending a bunch of time solving problems in chat GPT and researching things and, like, finding facts and, you know, using it as a helpful assistant to understand the world and probably very little time on SORA. you know, slopping it up. But, yeah, what,

Starting point is 03:15:14 what are, what are parents of teenagers actually doing specifically with clear space? Like, how are they utilizing it? What are best practices? Our, our kids are much,

Starting point is 03:15:24 uh, younger five, five and below. Uh, so we've have probably another, at least five, five to seven years before this is even a conversation. But I'm curious.

Starting point is 03:15:35 Yeah. Some of the cool stuff that we've seen is people inviting their kids into like, challenges as families. So setting up like incentive and reward systems, one of our most popular gambling. Gambling. Gambling on screen time. The most American thing. Fight, fight one addiction with the other. We, uh, we, we hear a lot of parents like set up family groups where they're doing a challenge, like a push-up to scroll challenge where everyone in the family or squat to scroll, you have to earn every single minute you're going to spend on social media that week with a physical exercise that we validate with like a machine learning verified push-up or squat. And then who

Starting point is 03:16:10 whoever's in the lead at the end of the week gets to decide where the family's eating out to dinner on Friday night. I think the most effective solutions we're seeing are kind of these like cultural pushbacks, almost, you know, community type stuff like the practice of Shabbat in like the Jewish community where it's like 24 hours everyone's putting their phone away. And people love it. It's not this like negative you can't use your screens. It's like everyone's kind of buying into a new fun thing. And we think Clear Space is the platform that can empower you to do that like really quickly. You don't have to be a software engineer and know how to hack your router. and know how to program all the phones.

Starting point is 03:16:42 You don't have to buy a dumb phone. You can kind of still participate in the latest and greatest technologies while knowing that they're not completely zapping your entire family. Yeah. Totally. Well, congratulations on the progress. I know we didn't talk too much about the business,

Starting point is 03:16:54 actually, more about the problems that you're solving. But it's, yeah, it's a very interesting industry. Thank you. I'm curious. I'm curious for both of you guys what you're doing about your screen time these days. Most of the time, I mean, I don't know. Because all my screen time is actually really low because I'm live streaming all the time.

Starting point is 03:17:11 And do I count this? Like, I'm looking at a screen. I'm looking at you. The other thing I've always been frustrated with the dedicated. My screen time is actually way lower than it used to be because I'm live, but that's content and I am the content. And so it's a little bit murkier. The other thing is I was always frustrated with screen time.

Starting point is 03:17:26 I'll give clear space a spin. But with the core screen time app, it would track like call hours as part of the screen time, right? Which just makes zero sense to me because I'm like oftentimes, like not using my screen at all. I'm just talking with somebody. And so anyways, I just always felt like,

Starting point is 03:17:47 I don't know, I think I've always been more focused on spending time on, you talked about like all calories are not, you know, created equally, but spending, spending time doing the right things. But yeah,

Starting point is 03:18:01 the best thing I've found is just putting my phone far away and just staying away from it. Yeah, Last week, I spent eight hours in the Maps app or something, which is like, I did to commute last. Map maxing. It was my number one used app last week. I don't know what was going on. I was driving a lot of, I guess.

Starting point is 03:18:22 Putting the phone across the room is what we call a classic low-tech solution to a high-tech problem. We think there's an opportunity to basically deploy super intelligence at the network level, at protecting your attention rather than exploiting. And that's like, we think that's what people want. and that's what we're bringing to them. I'm going to really stick it to Apple and just start breaking my phone into pieces when I want to use it less. And then I got it, because I'm on the whatever,

Starting point is 03:18:46 I get the free replacements, right? So then I just go in the next day, get a new phone. And when I want some less screen time, I just break it into pieces. That's the real low-tech solution. But excited to give clear space a spin and really glad you guys are working on this.

Starting point is 03:19:03 Very important that we have the smartest people in the world trying to capture as much attention is possible, and it's good to have, you know, equally smart people working on, uh, on humanity's side here. So thank you and, uh, great to meet you. Awesome. Great to meet you guys. Cheers. Talk to you soon. Let me tell you about adquick.com. Out of home advertising need easy and measurable. Say goodbye to the headaches at out of home advertising. Plan buy and measure out of home with precision. Uh, oh, this CNBC story, meta failed to act to protect teens. Second whistleblower testified. This is from 2023.

Starting point is 03:19:34 So I don't know why Adam Dell wants us to discuss this. Is there something new here? Because the CNBC story links to a Wall Street Journal article. And I was like, oh, November 2nd, it's a little bit old. But it's November 2nd of 2023. And so I don't know what the news is here. Maybe there's something new. We'll have to dig in.

Starting point is 03:19:56 But we can go and take a tour of the latest in Meta's world. The most recent thing that I saw was that they beat the news. the FTC case on the on the on the on the on the antitrust case that they were fighting so uh things to be seem to be on the up and up um the latest thing from meta on the like is it safe for kids uh stuff was from adamisari at instagram where he was saying he's going to try and make it pg 13 he was using the mpa rating system and i believe that there was like some some pushback from the mpa on whether or not they could i was saying hey matter you're rich enough you should buy the entire mpa i don't even know if you can't probably non-profit but like

Starting point is 03:20:34 Like, I want, I want those ratings, like, in every app, like, just as a consumer, because everyone understands it. You know what R-rated means. You know what PG-13 means. You know what PG means. I can even tell the difference between G and PG now with the kids, because PG, there might be a little bit of, like, a tense moment. It's just, like, emotionally charged.

Starting point is 03:20:54 I used to think PG-G, it's the same stuff, but it's not. And so, quantifying that, using the, using every movie is training data, pipelining that in, And then filtering every message that gets sent into Instagram, every post, hey, how would you rate this? Machine Learning model, AGI God, personal superintelligence. Let's give every Instagram post a rating and then let the user decide, hey, I don't want to see anything above PG-13. I don't want to see anything R-rated. I don't want to see anything G-rated or PG-rated or anything above that. If you could set your threshold, I think that would be very useful for parents who want parental.

Starting point is 03:21:34 guidance on the meta platforms. Okay. I don't know. We've had a bunch of guests back to back. We've also had some ads back to back. Getbezzle.com. Shop over 26,500 luxury watches, fully authenticated in-house by Bezels, team of experts.

Starting point is 03:21:48 Okay. We got to talk about the Kermew. Yes. Keyon Exchange. Yes, yes, yes. I thought that Kermu was pretty, pretty fair. Sure. I didn't, I think he could have gone harder.

Starting point is 03:22:04 Yeah. I Kian said he didn't watch Kermew's interview that he was with a patient I wish that he had because it seems like this is

Starting point is 03:22:17 one of the most important things on his plate right now is kind of getting beyond this and ultimately we tried to bring up as many of kind of the key concerns as possible to be honest it's so it's technical enough that I don't

Starting point is 03:22:34 that it's, it's, we don't have the same power level in terms of kind of like pushing Keon on some of these issues as Kermu and Sichuan. Which is why I think, like, a lot of the debates will be done around the data in blog posts by scientists, but... Yeah, and Keon's main point that I took away was... Sorry, laughing at the chat. But it was like, people are angry that we have the best marketing and the best science, and I believe that they have the most effective marketing in terms of capital. attention right now but I don't have the confidence after that conversation that the marketing should be happening at the scale that it is like I just don't I didn't I didn't

Starting point is 03:23:25 come away from that super confident that that that Keon felt like they've done anything anything at all that's even even even if you just narrow it to what there is full agreement in amongst all participants on which is that there were reviews that were anonymized and then not disclosed to be anonymized even if it's just that like that is enough to you know if you're a customer be like oh like this I don't trust this anymore and it's a very very high trust environment it's a very trust critical environment it's not okay yeah I'm buying a phone case and like they AI generated the like an example of a person holding up the

Starting point is 03:24:10 phone case like that's not what this is this is like life It's bio. It's really important. People are selecting, you know, impacting their future child's life. Yeah, we've talked about this before. And I think that whatever happens, it's really important to get it right. It's really important to get it right. And ultimately, people that use the service, if they have an adverse experience, they feel like the service didn't deliver.

Starting point is 03:24:34 It went wrong. They're going to come back to this stuff. So are there any posts that we need to read from the time? The fundamental problem with Keon's deflection is that even. if the origin polygenic weights can be independently validated to have the predictive accuracy's claimed in their white paper, none of that changes the fact that they completely plagiarized Harrisite, falsely claim that origin is novel and has superior performance, made numerous errors from typos to substantive mistakes in their white paper, have a history of making literally

Starting point is 03:25:06 impossible accuracy claims, in some cases inflated 2000X to independent replication. Uh, these, these are not just, what's up, Tyler? Oh, I just, uh, there's, you can continue. I have another post. I want to talk about it. Okay. Yeah. These are not just restricted to claims about technical accuracy. These are claims about ethics and morality for the vast majority of businesses. Maybe you don't really care that that much about unethical practices. Typically that's fine. I mean, maybe meta does shady stuff sometimes, but I still use messenger. Maybe my local grocery store manager is a wife beater. Who knows? I still have to buy my groceries somewhere. That's a, uh, uh, uh, This is the example that we keep giving of, like, you know, will you go change your tires if you find out that the Bridgestone tire company is spying on, you know, a rival tire company? There are certain things where it's just like if the tires are working, like, you're okay with it. A lot of it depends on like, is the criticism directly in line with the company's product? like if the if the CEO flubs align and they miss earnings but the product still delivers what it is it's like they sell apples and the apples taste great like you're fine but when but when they're when they're making a claim about the actual product and then it doesn't align that's where

Starting point is 03:26:30 people really really start to ask questions so closing out here it says but we are talking about a company that will select your future child how can anyone trust a company where the CEO's first instinct when confronted with evidence-based criticism is to make false accusation of shady conspiracy theories. And Mew follows up and says, if a company engages in malpractice, e.g. plagiarism, providing products, they should know our bad two customers, et cetera. Is it water under the bridge if they can clean up? That's obviously a reaction to my question was, you know, is there a redemption arc in his mind?

Starting point is 03:27:05 Somebody says Volkswagen can answer this question really well. is because Dieselgate is what's going on with uh is is uh is that in reference to yeah anyway uh kermew also says keon alleges that he's been working on embryo services for a long time again that doesn't seem surprising i don't i think this was always the the longer term vision of the company uh but uh kermew says he hasn't been uh he hasn't been doing this with genomic prediction according to their lawsuit which says in 2025 nucleus sought to offer ivf products involving embryonic DNA testing. Because nucleus could not do that work itself, it contracted with GP for GP to use its own embryonic genotyping products to provide test results for patients.

Starting point is 03:27:50 Nucleus made overtures about acquiring GP, but soon it became apparent that nucleus was looking for inroads to misappropriate. Again, these are just allegations that we can't really validate ourselves. But again, didn't come away from that kind of more confident in anything, really. Yeah, the type form, a lot of people are saying that the type form is down. Kian was, of course, saying that, like, he will release the models and that you can go and get the, you can get the models and the data from the type form. It was down at the moment he said that, but Max Glick, thank you for your service doing that

Starting point is 03:28:35 reporting in the chat while we were talking about that exact thing. Max says, it's back up now for what it's worth. And so Kermew says that's good. And I certainly hope, I just feel like the next, the next turn of discussion needs to be, okay, we tested the models, we tested the data, we tested the claims at a lower, at a higher level of rigor, I guess. Tyler, which post did you want to run to? Oh, it's at the bottom of the timeline. Cedgwan Mala is also accusing Kian of using a Chad filter. This has happened before. This happened before.

Starting point is 03:29:11 So when Keon came on the show, maybe six months ago, growing Daniel accused him of using a Chad filter while on the stream and went super viral. And I was kind of like, oh, like, Like, that's, I don't know. I don't know how to, you know, even respond to that. That's a very silly claim. I have no idea if this is real.

Starting point is 03:29:42 I can't, I can't tell at this point on a Zoom call at this resolution. What do you think? Do you think this is real? Are you guys just cracking up because everyone, does everyone think it's real? I don't think that he used a filter. I don't think he used a filter. I don't think he used a filter.

Starting point is 03:29:57 I don't think he just grew a beard. I think he's just been mewing, maybe. Maybe he's just photogenic. Yeah, it is possible that he just, he just, you know, flexed his jaw muscles and, like, you know, it has low body fat. I don't know. I don't know. I feel like it would be extremely high risk to run a chin augmentation filter. goes down for a second. I mean... Because you know that's what happens, right? When you're using like the Snapchat filter or like the TikTok filters, like sometimes

Starting point is 03:30:25 they pop in and out. And if they pop out, like you're done. Like people are going to be immediately, you know, let's see, compare the first two images with the X.com profile. I mean, the X.com profile, that picture looks three years old, four years old. I think on this one, Sishwan, like you might be over your skis. I don't know. We need to prove it.

Starting point is 03:30:46 and then yeah people are people are really going back and forth it's got to get the nucleus test for the gigacad test and publicize the results everyone's cracking up in the studio we're having a wild time anyway in other news this is actually insane apparently according to x i don't know if this is true but the robbery that took place yesterday in which an armed thief posed as a a delivery driver and robbed somebody for $11 million of Ethereum and Bitcoin was a Locky Groom that was targeted. Whoa, what? Which I had...

Starting point is 03:31:30 No way. I didn't realize that it was him, but absolutely terrible and traumatic. And, yeah, when self-custody goes wrong. But I'm glad he's okay and safe. Wow. An armed thief posing his delivery guy finessed his way into the $4.4 million mission district home shared by investor Lockheed Groom. Yes, Sam Altman's ex-boyfriend and another tech investor named Joshua.

Starting point is 03:31:58 Okay, so it was not Locky, but Joshua? Gary Tan posted the footage, panicked enough to delete it minutes later. Crypto security experts are now saying what everyone thinks. Self-custody is great until someone shows up your door with a fake UPS label and a Glock. San Francisco's tech leader about to hard pivot into vault custody, private security, zero public flexing because this heist wasn't random. It was a warning shot. Very chat, TBT, writ. Mario Nafal. But, uh, anyways, very sad.

Starting point is 03:32:29 Yeah. But glad Joshua is okay. I was confused for a second. That is terrible. Anyways, um, thank you for tuning in today, folks. Before we head out, I got to tell you about eightsleep.com. Exceptional sleep without exception. fall asleep faster way sleep deeper and wake up energized and i'm also going to tell you about wander dot com book a wander with inspiring views hotel great amenities dreamy beds top tier cleaning in 24 7 concierge service it's a vacation home but better folks thank you for taking the time to listen to the

Starting point is 03:33:01 show thanks for dealing with our stream issues up and down we sorted it out the full ever happened before posted uh yeah the full interviews will be posted obviously rsss feed youtube uh And the full Keon interview has made it sway to the timeline. Obviously, there's a lot of debate. We will continue covering it, but not tomorrow because we're off and not Thursday because it's Thanksgiving. We will be back on Friday for Black Friday. We have a fantastic lineup of a bunch of different entrepreneurs,

Starting point is 03:33:33 e-commerce, brand builders. Some of the most savage. Operators. It will be a lot of fun. Cannot wait. It's going to be a great time. A lot of friends. Have a wonderful Thanksgiving.

Starting point is 03:33:44 we are thankful for each and every one of you. Thank you for being a part of this, and we'll see you Friday. Goodbye. Cheers.

TBPN Live - Crémieux & Kian on Nucleus Genomics, Michael Kratsios on Genesis, Nvidia Responds to TPU Progress | Kian Sadeghi, Joe Weisenthal, Sebastian Siemiatkowski, David Faugno, Keller Rinaudo, Royce Branning

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.