TBPN - Figure AI in WSJ, Russell d'Sa, Quaid Walker, Alex Dhillon, Robin Langtry, Everett Randle, Keller Rinaudo Cliffton

Episode Date: April 10, 2025

TBPN.com is made possible by:Ramp - https://ramp.comEight Sleep - https://eightsleep.com/tbpnWander - https://wander.com/tbpnPublic - https://public.comAdQuick - https://adquick.comBezel - https://getbezel.com Numeral - https://www.numeralhq.comPolymarket - https://polymarket.comFollow TBPN: https://TBPN.comhttps://x.com/tbpnhttps://open.spotify.com/show/2L6WMqY3GUPCGBD0dX6p00?si=674252d53acf4231https://podcasts.apple.com/us/podcast/technology-brothers/id1772360235https://youtube.com/@technologybrotherspod?si=lpk53xTE9WBEcIjV(39:54) - Alex Dhillon (01:00:21) - Quaid Walker (01:21:53) - Robin Langtry (01:42:14) - Russell d'Sa (02:00:56) - Everett Randle (02:21:09) - Keller Rinaudo Cliffton

You're watching TVPN. It is Thursday, April 10th, 2025. We are live from the Temple of Technology. The Fortress of Finance. The Capital of Capital. We faced a terrible DDoS attack, likely from a nation state, but we did sort it out. Sorry for starting a little bit late, but we got a great show for you. Thank you, folks. Ben and our team of engineers. Yeah, this is why. People always ask us, why is your team so big? Why do you have just shy of 75 people working on the show? This is why? Because it's real. business it's it's serious and you need you need you need you need the neckbeard site reliability engineers really that's right you need full stack engineers you need back in front of everything i'm looking at uh ben right now and his neck beard has actually grown out meaningfully in the last hour yeah yeah you think you oh a couple prompts couple prompts will do it not yet a GI 2027 maybe we will have reliable live streaming by then i don't know about paper clipping but we'll figure it out

anyway well it's good to be here john i'm excited for today's show Let's run through some news. We want to start with this Wall Street Journal article about figure AI. And interestingly, about the Wall Street Journal, two different headlines. On the internet version of the Wall Street Journal, they say the $40 billion startup mystery shaking up Silicon Valley. And in the print edition, the headline is, hottest AI startup makes bold claims but earns little. So aggressive, a little bit, a little bit shamed. nadier on the...

A little more mysterious. A little more shade coming from the print edition. But we'll see. We'll review this. There's some information about figure where they are, what progress they've made, obviously about how they're running their fundraising process. $40 billion's big number. Got to back it up.

So we'll see. Let's do it. I'll dive in. So in February, a little-known startup promising to build futuristic robots set out to raise new cash at a nearly $40 billion valuation. The pitch. Figure AI would put more than 200,000 robots across assembly lines and homes by 2029, solving an engineering challenge that has eluded hardware developers for decades.

It has a long way to go. Figure had no revenue last year and just a few dozen robots in production, according to documents shared with investors in recent weeks. The documents show Figure has signed BMW as its first commercial customer and predict it will generate $9 billion in revenue by 2020. Now, is that $9 billion from BMW? Or is that just they figure generally predicts that they will generate? The figure is predicting this and they're putting it in their investor material.

Got it. Okay. Yeah. Big number. But I mean, to justify a $40 billion valuation, that makes sense. Also, I mean, the crazy thing is paying a little over 4x, you know, 2029 revenue here. Yeah, yeah, 4x. Which feels cheap.

Forward, four year forward revenue. of the year 2029. Yeah, yeah, yeah. So it's because forecast it. Forward, forward revenue would typically mean like the year ahead. Yes. Forward revenue would be, yeah, I'm paying a 5x multiple on this year's full year revenue. This is forward, forward, forward, forward revenue.

Yes. But who knows? Hopefully it's extremely high margin revenue when it comes in because. That would be great. That would be fantastic. I mean, yeah. Even if you, if the margins. less than $2 billion in cash flow.

So here's where it gets interesting. On March 24th, Figures founder, Brett Adcock, wrote that his startup was the number one most sought after private stock in the secondary market, which we talked about on the show. And he shared a list that put figure above SpaceX and Open AI.

We were a little bit taken aback by that. Yep. We talked to Christian Garrett about that, about how those lists actually get generated and whether or not, like, how much faith you should put in a list like that. It turns out, that I guess with like modern word processing,

kind of anyone can put together a list and screenshot it. And so I think there was some questions about, you know, was this from just a word processor? Or maybe it was generated with AI. But I would hope that it was generated with AI. Yeah, but certainly. Make a list where we're at the top. Yeah, exactly.

Pretty effective. Well, it makes sense. AI might be biased towards AI companies and secondary. So it pushes that up to the top. Yeah. How much a startup decided it could raise money at a price tag that, how such a startup decided it could raise money at a price tag that would make it among America's most valuable private companies

as confounding investors across Silicon Valley? Had Adcock leaped frog the likes of Tesla and Google in developing autonomous robots? Or they wondered, was this a sign that the AI bubble was hitting its peak? And so there's been a lot of chatter in Silicon Valley around this company. Adcock, and they give some background. Adcock, a serial entrepreneur has been posting frequently on social media about how much interest there has been in figure shares and touting the BMW partnership as proof of the

three-year-old company's rapid progress. Adcock didn't respond to request for comment. In a March 31st post where he shared a video of the slender humanoids working on assembly tasks for BMW, Adcock wrote, this isn't a test, this is what autonomous robots in production operations look like. Turn up the music. Turn it up. Put it to 11. And yes, it does.

So they dig into the BMW deal. And this is a big question of, you know, if you're a hard tech company going and getting a partnership with a big company is often a huge milestone. Huge milestone. It is important. And so guess what? The journal actually reached out to BMW. Yes. And they confirmed some details.

So on April 1st, a spokesperson for BMW actually said that three of the robots. were at its facility for technical evaluation. Only one is used at a time, but the robot has practiced picking up and grasping parts during non-production hours. In our body shop. But the thing is is that because this came out, this statement was made on April 1st,

it could have been April Fool's Day and they could have been joking. They could have thousands. But it's hilarious to April Fool's on the journal, let's tell the Wall Street Journal that we only have three that are just kind of moving some stuff around and actually the entire car is built by figure we fired all the employees but yeah you know

since you called us in april first we're just joking around about it it's possible that's very yeah it is it is that's the nature of april fools i got fooled tris stifes food me i felt fun it was terrible i got fooled multiple times i yeah there was some you know it's april fools but they still get you there was some founder who tweeted like uh oh there's a there's a there's a there's a hit piece and tech crunch coming out of my company. It's dropping tomorrow. And I made a note of it. And then the next day I was like, hey, like, I didn't see that piece drop. But like if it comes out, like you should come on TPP and we should talk about it. And I, you know, I'd love to dig into it as long as it's not like a complete smoking gun on you.

And he was like, it was April Fool's John. I was like, yeah, okay, I fell for that. I'm sorry. Anyway, so the following week, I guess this is post April Fool's, the BMW spokesman said that he had received an update from colleagues at the plant that there were now more than There were now more than three robots on site. They were being used in non-production and live production situation. That's great. I mean, that does seem like a huge milestone for humanoid robotics that, you know, it can be used. Like, again, all this stuff is like, I'm fine with a smooth gradient of progress.

I'm fine with a, with just a robotic arm doing stuff. I'm fine with teleoperation. I think it's fine to go up the learning curve or down the learning curve, whatever. Yeah, a lot of the big questions here are not around the potential of humanoids. although there are some question around the form factor. The questions are around the price tag, being at $3.9 billion, totally,

$39.5 billion, which at the time when it was announced was more than Ford Motors, which had, I believe they had roughly $180 billion in 2024 revenue. Yeah, yeah, yeah, yeah. I mean, you see what OneX is doing, you see what that creepy European humanoid company is doing.

And they're both very modestly valued. And so that puts a level of scrutiny on those companies. That's just wildly different. Like the Wall Street Journal isn't even asking questions of investors about that because it's like, I don't know what 1X's valuation is, but I imagine it's much lower. And so when 1X, they did a video with Jason Carmen, where it was folding laundry, it was teleoperated. They were pretty transparent about that. That came out that, oh, they're doing teleoperation on humanoid robotics.

Which is still very exciting. It's awesome. Yeah, it's great. We love teleoperation. You know, who doesn't want to be driven around in a teleoperated, you know, Waymo? That's fine. It doesn't matter. Sure, there's questions about how does that scale?

What are the economics of that over a long period of time? But it's fine, in my opinion, to gather training data that way, scale down the learning curve, all these different things. Yeah. And so one of the big questions and one of the issues and potentially one of the reasons why the journal would pay attention to this round is that the investors investing in the round are not. not the typical big names that you see in these multi-billion dollar round. So one of the funding rounds biggest investors, Align Ventures, has spent weeks marketing the round and looking for smaller investors to buy in at the startup's valuation. According to a term sheet and other documents, the smaller investors would pool their money into an SPV, reducing the amount that Align itself

has to put up for the latest round. So Align to our knowledge doesn't have, you know, anywhere near the capital just on standby to lead around. like this, but they are doing an SPV into the round. Yeah, it's also an interesting valuation for the amount that's being raised. You know, the standard round, 20% dilution, 15% dilution. Miramorati, 10% dilution. She's raising a billion dollars at 10 billion. Two on 10.

Two on 10. Yeah, so 20%. He's raising 1.5. You would expect this to be, you know, 10 billion, 15 billion, maybe 7 billion. Even Anderrol's most recent round, I believe, was like roughly a 10%. Yeah, I think it was like two something on 20 something. And so, yeah, like 10% just seems to be like in the ballpark. This is much less than that.

And so that takes some pressure off the fund raising that a line has to do, but it's still a lot of money to raise if you're not managing a $20 billion venture fund or a growth fund or something like that. So in many ways, a bet on figure AI is a bet on its founder. Adcock launched a series of companies since he graduated with a business degree from the University of Florida in 2008. He sold Vetteri an online hiring platform. He co-founded in 2018.

Then he moved to California and co-founded Archer Aviation, a maker of electric-powered air taxis. Archer went public in 2021 via SPAC. That company is also developing futuristic technology and he's yet to generate meaningful revenue. Adcock left the company in 2022. And I'm interested we're having the founder of Zipline on to talk about drone delivery.

And I'm interested to hear how the fly flying stuff market, flying car market, flying drone market, these clearly bump into each other. I'm interested to see how he landed on zipline strategy and how it differs from what Archer is doing. That was the same year that Adcock launched Figure AI. In the early days, Adcock took online AI courses and had books about robots scattered about his desk. Former employees said he hired robotic experts, raised 70 million in venture capital, and unveiled its first humanoid robot. in 2023. And so getting up to speed about the industry that you're going into.

Just a little over a year ago, they raised $675 million in funding at a $2.6 billion valuation in which they raised from Open AIs fund. They raised from Jeff Bezos's family office as well as NVIDIA. So really a who's who of hyperscalers. Yeah. You know, that online course thing sticks out, you know, he's sitting there taking online AI courses. And people would say, oh, like, he should have already known this stuff.

If he's building in this industry, he's taking this, he's doing this company. But like, that's pretty common. Like when we, when we decided to start a podcast, we, we took a ton of online courses about podcasting, how to use a microphone, how to talk, how to make jokes, how to do all these different things. And so I think that's, I think that's a little bit of narrative violation that you have to know what you're doing before you go into the industry. That's right.

It's very good. So anyways, Bezos, after investing, actually visited the company's facility and Figure was in talks with Amazon on a partnership. Amazon obviously employs a lot of people. Over time, you can imagine they would employ some humanoids. Employees, Figure worked on a demonstration where the robot could lift heavy objects. A few months later, Adcock told staff at an all-hands meeting that Figure and Amazon had decided not to move forward. And Amazon had bought Kiva robotics, which is a robotics company, but it's like these, they look like Rumba's basically. Then they kind of slide around on the floor.

You've probably seen these videos. It's really cool. And it makes a lot of sense. And Amazon's obviously super dedicated to robotic automation. And interestingly, they've been in this Centaur model where they're both hiring more and more people every single year and also employing more and more robots as they've scaled the business. Yeah.

So. Yeah, one of the reasons that figure has been top of mind is some of the claims they've been making around their end-to-end robot AI, which they built entirely in-house. They had a partnership with Open AI. Open AI invested in figure. They broke off that collaboration earlier this year and then announced their own. Yes.

Do you have any sense of what they're kind of claiming around? Yeah. So when I heard about the Open AI partnership, I thought it made a ton of sense because clearly Open AI passed the touring test. It's great chatbot. They have great whisper models for voice to text, basically. And so when I think about I want to communicate with a humanoid robot, I want to to be able to talk to it. I want it to be able to understand me at a level much more, much better than what Siri can do. And I wanted to take that text and be able to throw it in an LLM and then look up

stuff for me, talk to me, but then also trigger different actions for the humanoid. So if I say walk over there, I'm fine with an open AI partnership transcribing what I said, sending that to an LLM, and then triggering the, the robot to do exactly what I said. Now this phrase end to end, I think is going to become something of a hot button issue in robotics and humanoid AI, just like teleoperation has been, where, you know, we talked about this with 1X. We talked about this with the Tesla cyber cab and humanoid robots, the Optimus event, where the robots that were serving beers at the Tesla event were teleoperated. And so you would be talking to a real human who would be moving, triggering this stuff.

After that, there's actually an interim level where some of the robotic decision making and actions are dealt with through AI. So the best way to think about this is to go back to how self-driving cars evolve. So self-driving cars, the first step was let's use AI just to understand the world and create a 3D representation of the world as quickly as possible. So you're just doing image recognition. one frame at a time and seeing, okay, there's a cone there, there's a stop sign there. But once you have all that data, you basically import that into a video game, and then

you're writing deterministic code for how to actually plan the path, how much acceleration you need, how much braking you need, and that's usually written in C++. Just for example, I know it was at Cruise, but essentially that's written as like business logic. If stop sign, apply pressure to the brake. Now, Tesla and what George Hott's building at comma, they have all wanted to go, quote unquote, end to end, which means that the only data that goes into the self-driving car is the camera feed, and the output is brake pressure

and you take in all the sensors, and you just output the steering and all of that. And so Elon made a big splash a couple years ago saying that Tesla was going fully end-to-end. end. And in that presentation, he was like, yes, it's all machine learning code now. And one of his engineers was like, well, look, like, there's still a lot of C++ involved, Elon. Like, let's not go that far and make that claim because that's not quite right. But it is awesome that both in the, in the world model, like, understanding the world and then also the planning. There are two different

teams at most of these companies. The planning process of understanding, okay, there's a Park car here, I need to go around it. Like, that is done with AI as well. And so that is basically Brett Adcock's claim when I hear he's, he has built an end-to-end robotic AI system. It's that there is no C++ code or Python code sitting somewhere that says, if beverage, grab with hand action number seven. Yeah. So the big question here, the thing that's exciting has is has Brett and the figure AI team? had a breakthrough that Elon and Waymo and many of the other companies

haven't been able to achieve with billions and billions of dollars. Because if Adcock and the team had, then it's very possible it would be worth this crazy price. And that's exactly what happened with OpenAI when the transformer architecture and GPT4 got so good that it passed the touring test. Before that, we did have this kind of, it wasn't end to end, but you could have chatbots, right? but what would the chatbots do? They would take in your text and say, okay, he's asking about booking a flight.

Let me run the booking flight code and give you, you know, when you chat with like American Airlines, you can tell that you're not talking to an LLM. You're talking to a if this, then that statement. And that transition is what allowed Open AI to like kind of bust the world open to LLMs. And if they did this, it is groundbreaking. And it is, it's, it's incredible.

I mean, it really, it really is. No, it really is. And it, and it is. And I think the more important thing is that most people who are really deep in AI believe that end-to-end systems that are reliant on insane scale and just hoovering up tons of data and then crunching it down. This is the bitter lesson that we talk about with Rich Sutton. The bitter lesson is, look, you can do all this crazy stuff where you, where you, like,

there was a critique of Waymo that they had a cone guy for a while and his whole job was just to write algorithms to detect cones. But if you just scale up all the Tesla data and say, oh yeah, we have a million cars feeding us hours and hours of video data and just train it all on a model. The model, the deep learning will learn what a cone is better than that algorithm ever could. And so you can pull that out. And so getting to end to end on any AI driven system is almost always the goal. And so Brett understands that that's where that's, that's where he needs to go and that's where he needs to build. And if they've done it, it's remarkable. Yeah, totally remarkable. So anyways, so in recent months, unsolicited emails from investors claiming to have access to figures

funding around Silicon Valley have been popping up in inboxes around Silicon Valley. They all offered a chance to grab a stake in a pre-IPO AI robotics company. People investing as little as 100,000, thousand dollars could participate one of the offers stated one email pitched an investment through parkway venture capital one of figures main backers it said there was an effort to raise more than 80 million dollars for a special purpose vehicle that would get to own figure shares in the 39 and a half billion dollar funding round with figure robots on the production line at customer number one bmd bmd and given the valuation being placed on tesla's rival optimist humanoid prototype this valuation is not as crazy as it seems as face value. Yeah, there was always this thesis of like,

if you believe in humanoid, and I think a lot of people do, how do you get exposure to that? Well, the only game in town right now is Tesla and that's this crazy business with so many cars and tariffs and China and all these different things and batteries and so much stuff. So yes, there wasn't a pure play product for this. And that could be part of it. And that could be part what's driving like the narrative around this stock essentially. So one investor received a notice in January that he could acquire figure shares from a former employee at a steep discount. He reached out to Adcock who responded that the proposed sale was fraudulent and that he could invest in a future fundraising round, according to messages

reviewed by the journal. Soon after, a representative from figure messaged the investor and asked how much he wanted to pitch into the series C round. The investor asked for financial information that could help them make that decision, the company provided access to a data room that contained videos of ad cop, adcock, talking up the company, an investor presentation showcased images of robots doing various activities,

including working on a car assembly and pouring a glass of milk. What the presentation didn't include was audited financials or projections. This is an interesting strategy. I was thinking about this. Like a lot of founders, they get bogged down in the fundraising process

when, oh, I think Sam Lesson was saying, the deal goes to die in the data room. No one's ever built conviction in a data room. So, you know, if you don't want to bother with the data room, but you still need a link to send someone, maybe what you want to do is just like go to your Twitter feed, your X feed, take all your marketing content, print it, PDF it,

throw that in there. And then when the investor shows up, they just get to see your bangers. Yeah, one thing is clear. Adcock is an incredible marketer. Yep. And some of the footage just coming out of figure, broadly has been really astonishing and powerful.

Yeah, I think that's the- So I can see why he would lead with that. Yeah, I think that's the lesson for anyone who's working over there at figure. You know, we talked to Scott Wu about this. Like, things are moving so fast. The company's growing at an incredible rate. They have a new office, new campus every few weeks, it seems, new videos, new breakthroughs.

And so, you know, to those folks who are working there, I would just say, like, take pictures, you know, you're going to want to remember this. They're going to write books about this company. going to make documentaries about this company. And so you're going to want to remember that. So take pictures. Look at the state of things and just witness. Just witness everything. Yeah. I think that's the goal. Yeah. Yeah. You really, yeah, this is, this is history. And there's going to be books written. So you're going to want to, you're going to want to witness this stuff. So anyway, that's a, that's story. I'm sure we'll be

digging in more. We'll see what other people have to say and what else is going on in the humanoid robotics. You know, we got to have the 1X founders on and we got to get someone from the optimist team on. We should do a humanoid day. We should. We should have everyone on. All of them. They'd be very interesting. Back to back to back to back. Yeah. They all have different angles. Yep. Like I think it's powerful if you just pick a lane and not say, hey, we're going to replace all human labor in the world right away. Sometimes you can start and just aim to do laundry, just aim to do dishes. Some of these, some of these tasks that are maybe less exciting, but still equally powerful and have tremendous potential.

And you know, we also got to get some humanoid benchmarks on polymarket for sure. Yep. I mean, the logical one would just be, you know, Brett Adcock spacked his last company. He took Archer Aviation Public. I'd love to see, you know, figure IPO percentage on polymarket. I'm sure there's a lot of other, even just like those, the amount of humanoid's deployed at BMW. I don't know if that would be auditable in a way. that Polymarket could work against and resolve the market,

but having some sort of humanoid benchmark on Polymarket would be fascinating to track, at least in terms of like, just sentiment in the market would be very cool. Anyway, should we move on to, we get a lot of questions from folks about how to encourage other potential listeners to tune into TBPN live.

And we have a little script for everyone. Yeah, so this is like something, It's not quite a battle card, right? It's not super condensed, but yeah, if you're out there and you're talking to friends, if you enjoy the show and you want to kind of share it with a friend, you might get some pushback, right? You might get some people saying, well, I already listened to some technology podcasts or I already listened to business podcasts.

So why don't we run through this live? Do you want to kick it off, John? Yeah, why don't you play the prospect? I'll play the sales rep and I'll try and get your cross the finish line. Exactly, yeah. So kick it off. No, you're the sales rep. So just so you're aware, a lot of listeners ask how we differ from other podcasts.

One key area is integration. Unlike other podcasts, TBPN Live connects all business and technology news together into one seamless three-hour daily stream. Does that sound like something you'd find useful? Possibly, but I already have podcasts I regularly listen to. I'm not sure I need another one. Okay, so that's understandable. but before we dive deeper into TBPN Live, could you tell me a little bit more about your current listening habits? Specifically, where do you find the biggest challenges or bottlenecks and staying informed about tech and business news?

Well, honestly, I spend a lot of time scrolling through the X timeline and other social platforms just to catch what my current podcast might miss. It can be pretty time consuming. Exactly. That's a pain. That's a pain point we've frequently heard. It sounds like your current podcasts aren't quite keeping pace with your needs. forcing you to supplement with fragmented sources. Roughly how many hours per week do you spend doing that additional research? I don't know. I estimate about five or six hours each week, John. Oh, okay. Well, TBPN Live is designed specifically to reduce that research workload by at least 50%,

potentially saving you hours each week. How would you feel about having that extra time back for more strategic activities or even engaging more deeply with our insightful content? That sounds good, but I'm a bit concerned about the amount of advertising. I'm currently subscribed to premium podcasts behind paywalls, and they have fewer interruptions. I completely understand that concern. Many listeners initially feel that way until they realize that the quality and relevance of the companies we partner with for advertising. Could I quickly share an example of how these partnerships have provided value to listeners similar to you? Sure, I'd be interested in seeing that.

But again, my current podcast lineup already seems pretty good. Absolutely, absolutely. Your current podcast sounds solid. Many listeners who've switched over were initially satisfied, too, until they realized TBPN Live offers unique features. Would it be helpful if we briefly compared the scrolling news ticker we have at the bottom of our live stream, a feature that's exclusive to our podcast? Actually, that might be interesting.

I'd like to see how it works, at least. Great. How about we set up a quick demo session tuning into the TBPN Live? stream together. I'll tailor the experience specifically around your concerns about fragmented research and show exactly how the integrated stream and ticker could streamline your daily news consumption. How does that sound? Sounds good. Let's schedule that. So that's just a great way to get someone on board. If they're on the fence, they're not ready to take the leap. Yeah, just right way to at least set up a demo. Yeah, yeah, for sure. Set up a demo for sure. Yeah, you really

want to go full, full SDR mode for us. You want to have the discovery call with them, handle the objections, create some counter arguments. Ideally, if you know what they're listening to, create a battle card. Yeah, and a lot of the techniques in here, you can actually leverage in your own business, in your own sort of day-to-day life in terms of winning and, you know, overcoming those objections, like you said. So really great work here, John, and hopefully. I think it's going to convert a lot of prospects for us. Yeah, convert some prospects and help grow, you know, the listener base. Yeah. Let's run through how Elon Musk rescued X from the brink. new report from the Wall Street Journal, diving deeper into the X and XAI merger.

A crowd of investors gathered at Morgan Stanley's New York office to hear X's sales pitch, eager to get a piece of debt that Wall Street had once shunned. Cell phones were a no-go at the January event, and the audience was told to stay seated until X, Chief Executive Linda Yucarino and others had left the room after brief remarks, and without taking audience questions. Banks had planned to sell $3 billion in bonds at $0.95 on the dollar, but ended up selling more than $10 billion at even higher prices.

It was a testament to X's ability to bring advertisers back to the platform, helped in no small part by TBPN, no, by Elon Musk's proximity to President Trump. Obviously, we can't claim victory 100% in the time, but I do feel like we are shifting the timeline and making X a better place. Hopefully. I mean, that is kind of the goal is to bring, to shift X back towards more tech and business content, really pull people back in. And fun fact, we had a listener or we had a guest on the show who didn't have an account, had never used X, signed up, posting, bangers. Yep. Great content. Yep. Fantastic. So also underpinning the debt sale was a possibility that X would one

a merged with a hotter ascendant company, Musk's X AI. In private meetings with Wall Street, ex-executive said there was a good chance that the social media platform might eventually merge with Musk's artificial intelligence company, which makes the Grock Chatbot. The billionaire has said he never lost money for investors, but for a long time, he looked like he was going to with X. After Musk bought it in 2022, advertisers fled over content moderation concerns, and its loans soured as revenue fell. One month after he took over, over, Musk said the company, formerly known as Twitter, was on the verge of bankruptcy. I don't know how this was true. I think it was, you know, in part said to, I don't think he

obviously would have ever let it go bankrupt. I think it was mostly meant to inspire the team to grind harder and grind harder they did. They did. So late last month, Musk posted on X that he was merging the company with XAI and a deal that valued the newly combined company at more than 100 billion, folding X into a larger company competing in a global race to develop sophisticated, generative AI tools could open the door to raising money at a valuation considered impossible just a few years ago. The merger caps a string of events, some strategic, some fortuitous that helped Musk announce a deal before Trump's tariffs effectively closed the market for deals. It is crazy that he's never taken a downround, but it's kind of underrated that once...

I don't actually... Is it downround or that he just always eventually got higher? I mean, I guess like downround in terms of... did raise a down round, right? No. Or no, I guess it got marked down internally, which is not his doing. Yeah, like anyone could do that at any time. Internally, you can mark it down. People have definitely panic sold and lost money, but he's never done, you know,

a primary round and ended up not returning. But I think the lesson for like actual entrepreneurs that aren't maybe issues is that if your business is at all humming along and doing something, like even though Twitter rebranded as X and went through like a dark time and really was facing a lot of pressure on the revenue side, Elon was extremely quick to right size the business and get to a place where it was not burning as much money. Like that seems clear. Like we never heard rumors about like, oh, like Elon had to put in a billion of his own money, right? And I don't think that happened. And so what does that mean? It's like, yeah, the whole business shrunk, but it continued to exist. And so there's a whole world where

if you move quickly, whether it's post-SBB crisis and you know you're not going to be able to raise up-round. And maybe your own VCs have essentially marked you down if you're at a big enough scale. They might have actually done it. If you're probably just some seed or Series A bet, they probably didn't remark you. Yeah, Fidelity was the one that marked it down. Yeah. Fidelity has to do their traditional investors avoided the markdown because they never lost faith. No, no, no.

I'm almost certain that the big venture capital firms do not do markdowns in that way, the same way is fidelity. Yeah. Although if Fidelity puts something out and says, hey, we have marked it down and we're trading at this level or something in some secondary market, then there might be pressure from the, from the VC firms to pull that valuation into their own marks. But it didn't matter because they're up again, baby. A spokesperson declined.

No one's talking to the journal about this, but we still got some interesting facts here. Musk borrowed $13 billion to complete his 2020 to take private of Twitter. And the loans quickly went bad when out of the... advertisers paused their spending. He tried to persuade concern brands to return, at one point offering steep discounts and threatening advertisers that they would lose verification if they didn't spend enough. The company's revenue fell to $3 billion in 2023 from about $4.6 billion in the previous year.

And the thing is that, like, yeah, that's a huge drop in revenue. But when you think about, like, most of the R&D has been done on X, right? Like the apps, you have to keep them alive and you do have to launch some new features, but like they're kind of humming along. You don't need to do a lot on user acquisition. Like the people that are addicted are addicted and stay. I'm never leaving. And so you really can run it with a little with a smaller team. And when you when you have three billion in revenue, which is pretty high margin because it just ads, you have three billion to kind of slosh around and keep this thing going. Like that's a lot of money to keep a website

going in an app. Like that that is what it is. Like it's a big one, but it's not, it doesn't have any like crazy unexpected massive costs if you run it effectively. And these companies again were always so heavily intertwined. It was X had a big position in XAI. Yep, yep, yep. And XAI paid X for data. Yeah. I would have employees flop back and forth.

So and then staying on the revenue thing, X's revenue dropped again in 2024 to about 2.6 billion. It ticked up in the final quarter of the year, according to people familiar with the matter. So they, so it might be turning around. But it's going to take a while to build that back up because a lot of people have left to go to other platforms. But it does seem like, you know, even like Twitter and X have reached their Nadeer, in my opinion, like in terms of like sentiment. And like the people who wanted to leave have left. The people who have stayed are staying.

The advertisers who have wanted to leave have left. And so, yeah, it's still is still producing single digit billions in ad revenue and monetization. And that's great. Musk and his advisors had long thought about bringing X and X AI together, but after the election, those plans accelerated, according to people familiar with the matter. To do that, they knew they would have to successfully execute several transactions in the right order and get a little lucky, but Musk is a pretty lucky guy. With Musk's, with Musk ascendant after the election, X took a new approach to ginning up the new, new ad spending. In December, a lawyer from X called

a lawyer at advertising conglomer interpublic group, hinting that it's recently announced $13 billion deal to merge with rival Omnicom would face trouble from the Trump administration. So he's putting the pressure on him. Other advertisers started to raise their spending, including Amazon.com, in January. Amazon's an interesting one because, I mean, obviously a huge, huge advertiser, but very direct response and very product-driven. And that's a little bit hard for the mindset that you come to X with, because it's a reading platform. It's almost like an education platform. It's not the same experiences as Instagram,

where you're just browsing different visual elements. You see a product. It makes a ton of sense. And I think that's why Ben Thompson's always argued that your explore feed now is just probably the Hennessy. What's it called? The Cadillac Escalade, the Hennessy edition, which has a thousand horsepower to explore feed of just like the reels is just different videos of that. You can't get it off your mind. No, no. Once you figured out it exists.

It has been interesting. Like I'll, I find that my YouTube feed moves much faster than my Instagram feed, but my Instagram feed eventually comes, catches up. So I'll get into cars. I'll be watching Doug Dumiro on YouTube. And my YouTube feed will be flooded with that. And it'll show me all the different, like, oh, here's Vinwicky.

Here's, you know, uh, hoovy. All the different car YouTubers will kind of flood in because Google picks up on it. Instagram I don't use as much, but eventually we'll learn, hey, We showed him one car video and he seems to be more into it than before. That's right. Let's keep it going. But again, on X, even though I do post about cars every once in a while in kind of a joking fashion, X hasn't figured out how to serve me ads.

Plus, I pay for like the most premium version that gets rid of all the ads. But even when I didn't, I never saw that. I thought about turning the premium off, turning the verification off just so I can get the full flood of ads. I wish you could be on the highest tier so you could pay X as much as possible yet still see ads. to get them that sort of incremental revenue. Yeah, I liked, I mean, for a while you could like pay for the checkmark, but then turn it off. You should be able to do the same thing with ads.

Pay to remove the ads, but then checkmark put them back. Yeah. But maybe you need a non-account just for that. Anyways, I'm excited to see what this combined entity does. Yeah. A lot of work to do still. Yeah, I mean, we've talked about it before. It makes a lot of sense in a bunch of different ways from,

from real-time data. When you go to GROC, you're going to want to know about things. All the data gets sucked into X anyway. And so it makes sense from a data strength perspective. And also just as a distribution vector for GROC, it seems like every major foundation model company has found a dance partner. And Google has Google.com. And they're searching and they're sending generative responses

into the Google search bar. Open AI has been able to get everyone to download the OpenAI chat GPT app. It's in my home row, and they've done a great job at that. There's been talks about, oh, perplexity might merge with someone, or Anthropic might merge with someone or partner up with someone. So the XAI, X and XAI deal makes a lot of sense in that context. But if your business is collapsing and advertisers are leaving and you need to cut costs, get on ramp, time is money, save both, easy to use corporate cards,

bill payments accounting and a whole lot more all in one place ramp ramp and we have our first guest of the show Alex welcome to the studio how are you doing today one second we are bringing him in from outtake AI we will give you a little background how you doing big news this week what's going on thanks for having me guys I'm pumped to be here thanks so much uh thanks for coming on busy week for you guys uh hyper scaling and announcing the raise all at the same time yeah but before we dive in introduce yourself and the company and yeah whatever backstory you want to give for the audience yeah of course well first off thanks for having me guys backstory I spent five years working at Palantir I worked across a lot of crazy different stuff as as

one does at that company worked abroad I was in South America I was in Eastern Europe eventually started to work on really deep poor product things towards the end spent a lot of time a week effectively our experimental product team so zero to one new product bets, work directly for the Pounder's CTO, Sean Sunker, absolute goat, a shout out to him, also one of our angel investors. Amazing. And then, yeah, and then, yeah, towards the end, you know, as everyone sort of became obsessed with AI, it was, it was really obvious to me that like,

we're going to live in an increasingly like schizophrenic internet. Like, you know, what you see is not necessarily what you get online anymore. especially the way I think about it is the cost of your shitty person online went to kind of zero dollars over the last like three years. You got infinite text, infinite images. That stuff is awesome, super creative. Just obviously there's bad people out there. They're going to misuse it. And I wanted to do something about that. And that's really what outtake is focused on. We, we try to preserve authenticity identity online. Go deeper there. Who are the kind of companies that you're working with today and we'll be working

with, you know, I'm sure some you can share, some you can't, et cetera. Yeah, it's, it's funny. In cybersecurity, there's this like, there's a, there's a, there's a lot of sensitivity around like what you can, cannot share in us. But we've been lucky to work with phenomenal places like, you know, open AI is one of our, one of our favorite customers really pushed us when you think about the kinds of cybersecurity threats that they might be dealing with. Really, anyone that has a brand big enough to really matter is a brand that can be

misused, right? So a fun side effect for us as a company is like our customers are intrinsically people that matter because they're the ones whose names and brands are being misused to manipulate people, right? Yeah, can you talk about so, so I feel like everybody's been aware of this threat around AI impersonation, you know, some of these like voice models, even video models at some point are getting so advanced that it's becoming a big threat. Can you talk about it feels like we haven't um we haven't faced like armageddon yet but i'm sure like uh i'm sure part of that is the work that you guys are doing and the work that you know these foundation models are doing so can you talk about like what's app actually happening in the market because i have the sense

that like you see this like tip of the iceberg where you're getting like spam on i message or you're getting email spam or you're getting you know some fishing attempt etc um but what what is the What is all the work that's going into this to make sure that you're only seeing that tip of the iceberg, right? It doesn't feel like we're completely flooded yet. But this thing, you know, this kind of phenomena is scaling. Yeah. To your point, I hope we never see Arbageddon. I think there's this like gradient of sophistication, as you pointed out, where like, you know, the average consumer deals with spam texts, maybe some shitty emails. And certainly, hopefully not like, you know, people have talked about the idea that your dad is like giving you a deep fake voice call and what

But we haven't necessarily seen that, right? Interestingly, a lot of the attacks, you know, on the other end of sophistication are actually really targeting enterprises, right? And it kind of makes sense. It's like that's where the money, that's where the data is. And so what we are seeing is like the quality and volume of, let's say like inbound email fishing attacks has like gone up considerably. And then this just goes back to the idea of like the cost to do that dropped a lot, right? And so enterprises are like dealing with an absolute flood of attacks at the moment. And there's a variety of attacks, right?

So there's like, you know, there's the good old email security stuff we just mentioned. There's just a lot of stuff around social engineering generally, right? It's like, here, let me throw it together a fake website and then use that to like DM people, to send emails by that. Or let me just get the customers of that company. Let's say, I'm going to use a random name that's not connected to us. Let's say it's like, I don't know, the bank of a bank of Canada. in some way, right? It has a website. That's going to now be used to scam Canadian citizens in some way.

Yeah, the sophistication is like really one phrase that I always come back to is like, it's not only that grandmothers are the ones being scammed. It's actually grandmothers can become scammers now and it's because it's like it's just easier than ever. And that's actually the real issue. It's like sometimes people talk about large language models and they're like, oh, like it could teach someone how to like clone smallpox. And it's like, yeah, dude, sure, it may be, but like, it could immediately today tell you how to, you know, pen test someone's website and try to get into it. And so it's just like the barrier to entry is just so much lower than it used to be for all sorts of cybersecurity attacks. Yeah. Can you talk about like the scale of, you know, some of these like fishing things? Like when I think about like, you know, everybody's in tech is used to getting like a coin base style fishing like, hey, you know, give us a call.

Like you had some issue. And I just think about like, okay, if that works one out of a thousand times, it's the best business ever. It's the best business ever. Like it is part of the reason that like, you know, this sort of these issues are scaling so rapidly is just because it's super profitable. Yeah. Yeah. I mean the like one way to think about it is we've all seen the tremendous growth of like AI outbound companies. Right. They've done really well. It doesn't take a leap of the imagination to understand that. like that same sort of like well intended outbound can be misused right um and so maybe one way to like paint an idea of volume is like just as the volume of AI outbound has increased on like legitimate use cases where i'm sure you and i both deal with a ton of LinkedIn inbound or email it bad um it's basically been one for one on sort of the darker side of it as well um what about the stuff you guys are doing with creators, I know in the announcement, you said that you guys are working with SciCom, the team behind Heurman Lab. What are you seeing creators dealing with in terms of

impersonation as these models have advanced? I mean, creators were some of our first customers, right? And it's actually because they were the first ones to feel the pain. I mean, to some degree, maybe you guys deal with this, right? Someone might make a fake handle of your account. That's happened, yeah. Yeah, fake versions of your profiles. There's a lot of value in impersonating you guys, whether it's a financial endorsement, a startup endorsement, just pretending to be you for recruiting and something. Basically, the creators were the first ones to hit because they were the most chronically online, right? Is the short answer there? And in the early days, when the models needed more data to train on, they were also the ones that had the most data available.

I think what was really fascinating for us as a company is, especially since almost the whole team came, from Palantir, we had this, we had a lot of reps that selling to large enterprises, government, and incredibly secure environments. And so creators ended up being this great launching pad for us, where we like built a product really, really fast and then grew, went upmarket really, really quickly. Can you talk a little bit about recent AI model progress? I don't know if you read this article on Less Wrong from Zero Path, but just to summarize it, it said, recent AI model progress feels mostly like BS. This was from a founder who actually built an AI-driven pen-testing company, and they said

they kept upgrading to the latest and greatest model. It would completely destroy the benchmarks, but then they weren't getting real-world results, and this was a question about benchmark saturation. I could imagine that, to your point about, like, the script kitties are better than ever, basically, because pen-testing is democratized now, or hacking or scripting is democratized. But in terms of the work that you're doing, how important is AI model progress to all that? And what is the shape of the curve that we're looking at? Is it sigmoid or are we going acceleratory to the moon?

That's such a good question. It goes back to my sort of like grading of sophistication response, which is like, you know, surprisingly, as we talked about, we're all dealing with that flood of text. Clearly it's working, but people keep doing it. And so we all agree that like even if we paused all model progress, that would continue. Totally. Yeah. And so my general take is like the models are already at a point where they're having serious social engineering attack effects.

Now, as you go up that gradient of sophistication, for example, the pen testing founder, he might have a very legitimate argument where it's like, hey, sure, there's all this hype around progress, but I don't actually see a model being able to autonomously hack a website. Because maybe that's at the further end of the investigation. But I guess my claim is that like, you know, it's something like 85% of critical data breaches happen because of social engineering attacks. It's almost like machines are easy to secure these days. Like, you know, there's a massive multibillion dollar industry that is all of cybersecurity that focuses on that. I think one of the big contrarian things about outtake was, it was really two things. It was like, one, hey, if machines are pretty secure, let's actually focus on securing the humans. And then two, traditionally enterprises will like stop at securing themselves, right?

So they'll say, hey, I have my corporate perimeter and then my VPN and everything. And there is really buttoned up and that's great. Historically, cybersecurity companies have been kind of, I think, gun shy of stepping outside their perimeter and saying, hey, no, I'm going to go search for threats out in the world, like arbitrary websites, social media, wherever it is. And we're going to go look for those threats and proactively tackle that. That's intrinsically difficult because you're trying to influence the thing that you did, you know, it's not part of your org. Our ability to do that and actually do the really hard work to get there is really what helped us stand out. Is the only answer for a bad guy with an LLM, a good guy with an LLM?

And what I mean by that is that, you know, CAPTCHAs are not, they're, they're, they're deterministic, essentially. And yet they can thwart a non-deterministic attack from an AI. You can imagine arc AGI puzzles becoming the next CAPTCHA or or in your view do we need just to stay is it is it a probabilistic on probabilistic combat from here on now I love this question okay so yeah one of my biggest anxieties is that like generally to your point captcha's are broken right which like hey we passed the touring test a while ago for some reason we just blew past that and don't talk about it enough we all are blowing past what effectively was the the like modern version of the touring test which is just captures us.

And like, there is clearly, in my opinion, not enough panic about this, which is why I was so pumped to tackle this two years ago. And we spent a lot of time thinking about this. Anyway, to get to your actual question of like probabilistic versus probabilistic, I think there's a few answers. I'll hint at the future of outtake a little bit.

Some of it's still, parts of it are still in stealth. But I think a lot of what we do today is what you're talking about. It's probabilistic versus probabilistic. And we're thinking about, hey, here's the harm we're seeing. And like, can we go practically discover and remove it, right? The other half of it, though, candidly is like, there's probably...

Briefly, can you talk about the agentic approach there, which is basically like, you're sending good bots around the internet, finding these sort of social engineering impersonation attempts, and then basically, like, acting like a human would in terms of being, like, reporting this content, you know, basically flagging it, being very specific about it. But maybe could you dive in there around, like, yeah, why... Yeah. Yeah, why this sort of agentic experience is important when you're... you're not building, when you're building out in the sort of like dealing with threats in the real

world out on the internet versus internal. Yeah. Excellent point. I think like the one of the other reasons that I was so excited to take a bet on this space is, candidly this whole space is full of a graveyard of companies that were services companies. Right. So like it intrinsically was a task so difficult that only humans can do it. Meaning, you know, let's say we're protecting again, the Canadian bank. Um, you know, you know, need to have an ability to go proactively search for every place that you think the threat might emerge. You need to be able to adapt to the idea, like, let's say that bank sells mortgages in the summer, right? The search terms that you use to find the scams that might occur

because now they might be mortgage scams need to adapt for the season, right? Now, you and I as humans can get to that pretty quickly. Historically, models were not going to just like infer the season and then figure out the like latest trending scam for this particular customer, right? we are able to do that, right? And so it's like having agents that are thoughtful about how search occurs from period to period really, really matters. The other thing is deciding what is considered harmful for that particular customer. Like there's a lot of nuance there by company. I mean, celebrities is a great example, right?

Where it's like there's fan pages. You don't want to remove that stuff. And so there's a lot of nuance in how you determine harm. Again, a reason why you would train up, you know, a team of people to be like, hey, this is what we consider harmful. There's a lot of moat and frankly figuring that out. When you use outtakes product, there's an aspect where our users effectively train our models to explain what they consider risky. And then the final bit is taking over the actual remediation steps, right? Like everyone talks about legal AI.

I think we've built one of the most effective legalized workflows, right, where we like go out and like interact with third parties to resolve external issues. Can you talk a little bit about the actual monetization or value capture here? I could imagine if I'm a bank and my customers are getting scammed, it's not exactly a bug bounty that you're, but you are creating value for me because my customers are going to be happy, then they're going to be unhappy with me if they get scammed, even if I had nothing to do with it. If you go out and shut one of these rings down, how are you getting compensated? It's a great point. The way we think about it is, so I guess good for us. The customers were thinking about these problems, right? They were just a relatively lower voles. volume historically, right? So like, if you look at the chart of like sort of digital attacks happening on, let's say social media websites, etc., it's sort of like is flat, flat, flat. And then like 22, 2023, it like starts to go exponential, right? We were, you know, positioned ourselves well, we saw where the puck was going and said, okay, great, all the people that

were already managing that low level of attacks. I mean, they existed. They just were a cobbled together team of like cybersecurity analysts and occasionally lawyers at every large organization had someone thinking about it, but they could sort of manage the one to 10 attacks a week, right? They were expensive though because they're, you know, yeah, very, very high, like hourly wages for these folks. And so when the volume spikes, it becomes untenable to use that old system, and then it became really clear that outtake could step in. And so to your point about value capture, it really became a question of like, hey, you would be trying to manage this, you would be trying to have these folks work like 24-7,

They obviously can't do that nor want to candidly. And slash, you wouldn't be able to afford it. So let Outtake come in, do a 10x better solution at one-tenth the cost. And this is a really good deal for everyone involved. Can you talk about, we had Ashkei on, Chief Architect at Palantir. Yesterday, we were talking about the sort of forward-deployed meme. Have you taken anything from that approach? You obviously were a forward-deployed engineer all over the world.

Do you end up talking to these big platforms? and they say, like, can you guys please come and just post up in our office for a week and figure this out? Or is it, you know, less intensive than that? No, I think the forward-deployed meme is there's a lot of truth to it. The reality is if you want to, if you want to do something meaningful, you need to go sit by the customer and like make it happen, right? I mean, a lot of our best product insights have happened because like me or another engineer were actually on-site and said, okay, like, here's how we thought you were doing it. but in actuality, you need to do it this way.

So yeah, forward to play engineering, huge fan. Love it. You got anything else? This is great. Yeah, I wanted to know a little bit about, like, mechanically, how do you shut down a ring of scammers? I remember reading this book about Paul Leroux. I don't know if you're familiar with this guy. It's called The Mastermind.

People think he might have created Bitcoin. And back in, like, 2004 or something, he would. was, you know those spam emails you get that are like, oh, magical pill, you know, pill mill stuff? He was the one sending all of those. And he leveled up so high that he would get a website shut down, instantly spin up a new website. But then he went even further and bought a registrar so that he could register like new domains for free.

And you couldn't shut him down at the registrar level, not just like, he had his own like TLD, basically. It was crazy. And so some of these, some of these like rings are extremely sophisticated. Is there a moment when you transition from, okay, my AI agent is just like sending a little cease and desist to, okay, we're getting in touch with like the authorities on this one because we discovered like, you know, a major, major organization here. Yeah. Okay. I have to be careful about what I can share publicly.

Sure, sure, sure. But yes. Yeah, we saw it with the Mark Rober. Mark Rober like busted some fraud scheme in India, right? I don't know if you're familiar with that one. Maybe just tell us some historical stories of how this works. Yeah, yeah. So to your point on the mechanics, like, the traditional way to like deal with all this is is purely legal, right? You put together, let's say, or it depends on the exact type of attack, but generally everything you said is spot on, which is like cease and desist letter, maybe a complaint of the domain host, the registrar, etc. That was a key ingredient in the early parts of outtake. I think the big differentiator, you know, in addition to all the agentic stuff is like we've thought really deeply about how do we become, um, How do we become the platform that is, like, trusted by domain registrars, hosts, social platforms, etc., to be a high-quality source of reports, right?

Like, that is incredibly, incredibly important. And I can't say too much, but, like, yeah, we've invested a lot of engineering resources in that direction as well. Yeah, that makes a ton of sense. Well, thanks so much for joining. This is a fantastic conversation. And good luck to you. Hopefully, you bust, you know, some massive scamming ring soon. Yeah.

Every day, I'm sure. Yeah, your new cybersecurity correspondent. Thank you. Thank you for all the helps. He's already been finding, you know, people impersonating us. Oh, yeah. So I appreciate your, your hard work and the whole team. And congrats on the new round.

Yeah. Thanks a lot. Thanks, guys. Talk to you soon. Bye. Talk soon. Yeah, I actually have gotten multiple times people set up fake John Coogan accounts, copy everything, copy every tweet, and then block me so that I can't see it.

And then I get, and the worst part about it is that I don't think that many people are falling for it. but I get so many DMs from people being like, there's a scammer that's impersonating you. I'm sure you've seen this. And it's like, yes, I have seen it. And then I have to go hunt it down and report it and stuff. And so it's just like annoying getting like a flood of text being like you're being

impersonated. You got good friends. But yeah, yeah, I appreciate it. Thank you. If you've let people know. But in the future, yeah, I mean, still send it to me. But definitely just report it and try and get the account taken down if it's out there floating around.

Next up, we got Quaid coming in the bill. watching. Watches and Wonders is done and it's been a busy week with all the tariffs and so we want to get a breakdown from him about how the watch world is reacting to all the tariff chaos. He's been in Switzerland called in from Watches and Wonders late night last week and now that the show is over we're going to get a full deep dive. Also want to talk to him about how how the news breaks around different watch news and how the how the media cycle works. So Quaid, welcome to the show. go. Thank you for having you guys. Excited to be back. Are you back in L.A.? I am. It's funny. The last time I dialed in, the tariffs dropped, I think, the minute I hung up with you guys. Yeah. Yeah, it was 4 p.m.

that day, our time. I was in such a haze. I think I texted you quite. I was like, what's the reaction over in Switzerland? And you're like, not good. Not good. But now, you know, we've seen things, there's been a pause, there's been ups and downs. Well, there's still the 10% base tariff. Yes. Correct. So what was the whole journey throughout watches and wonders? Very funny timing.

Good to be there on the ground. Take us through the reaction as things developed and where the heartbeat of the industry is right now. Yeah. I think you can split the conversation kind of between the primary market and the secondary market. And so I'll intentionally walk through that. On the primary side, I think all the meetings I have.

had the day after chatting with you guys, the tariffs dropping, there was a lot of like the Spider-Man meme pointing at each other, like not really knowing what to do right now. And I think you can split the reaction from brands kind of by the type of brand they are. You have like the giant massive brands that everyone knows like the Rolexes and the Pattex and the APs of the world. And their reaction is decidedly different than some of the brands that sit kind of in the middle, the market. For every brand that isn't these like very, very, very top tier brands, their focus was how quickly can I get product to the US? And whether it was

functional products, watches that are just parts, like they just wanted to move everything to the US as possible from most of the meetings that I was having. On the larger brand side, I think they had the luxury of sitting and waiting and seeing what was going to happen. You saw the Rolex of the world and things like that. not take their time. And the initial answer that I got from secondary feedback from, from authorized dealers meeting with Rolex that week was the large feedback that Rolex tended to give in that moment was like,

we don't want to pass this on to the customer, but we're not going to eat the cost of these tariffs either. So we're just going to hold off. And the main focus point was we'll make it harder to buy outside of your local markets. Because I think a lot of the authorized dealers in the U.S. are all of a sudden saying what happens when all of the first, my core customers decide that it's actually cost efficient to prioritize that flight to Paris. Oh, interesting.

Buy that watch there. So the thinking was how do we keep it in the local markets, unclear how they hold that out, right? That's really tricky, yeah. That's what was happening in the primary market in the moment on the secondary market in the U.S. particularly, I think you saw a lot of dealers think it was an amazing moment for a second, where all of a sudden the inventory that was just being in their lots in the U.S. was much more valuable than it was, you know, a few minutes before the terrorists were alive. And so on the secondary side, you saw it really came down to like what type of dealer and what

your economics are. If you're a dealer that happened to have a lot of inventory and you don't have a hard time stock in that inventory again, you have great sources. You had a lot of folks that just saw deals come through in an increase and they just sold them. Like they moved product and that was the name of the game. And then you saw a lot of other dealers try to opportunistically drum their prices, you know, 10 plus percent to all of a sudden increase that value. So we saw the best weekend we've ever seen after the tariffs in the sense that there was a huge increase in demand and everyone was out trying to buy those watches. And then the prices, I would say a lot of them are 10, 15 percent more than they typically would. We haven't seen a lot of like the craziness,

like no one's pricing in selling at 30 plus percent to try to match the 31 percent. move. And then since they announced the 90 day moment, I think things have chilled out a little bit. We're still seeing pricing being up 10-ish percent on a lot of the sales that we're seeing come through. But it's really interesting to see what dealers are going to do. I think it's a question of if you believe the 31% it's going to hold. And if you do, I think you're starting to stock up right now. If you don't believe it's going to hold and it's purely a bargaining chip, then I think you're kind of sitting and waiting out and seeing what happens next. American watch manufacturers there just like popping champagne like the Hamilton's or the

RGMs just being like I've been waiting for this I'm finally going to catch up to pot tech not a ton it's it's very it's very Swiss centric while you're there but uh it certainly would be a good moment to be but it was like a lot of these brands were lamenting because especially on the independent side like they had just opened up New York boutiques or they had just signed the lease in LA or whatever it is. And so the question was, how do we get product there? And a lot of the independent and smaller brands, the answer kind of became like, I'm willing to forget the U.S. market for a second. Like, can I come up with other skews in Europe that make up the revenue or lose in the U.S.? Or can I just incentivize, especially on the six figure plus pieces,

can I incentivize an experiential moment where I take my best collectors and I organize a trip where they come to Geneva and hang out with me as the CEO and they buy their pieces there. Yeah. And is that, I mean, obviously that's like probably not completely kosher with the tariffs, but watches are unique and that if you buy one and you throw it in your suitcase and you're wearing it as you come across, like how would they ever know that you bought it while you were on that trip, right? That's just the nature of the game. I think both people they buy it and shop it on and they walk out. Yeah. What about, so one of the big concerns of the last week is that even, for example, auto manufacturers that make cars in the U.S., a lot of those parts

come from overseas and specifically China. For the Holy Trinity brands, are they making every single individual part in Switzerland? I've seen those claims thrown around. It almost is inconceivable, but it seems to be the case that they're not dealing with, you know, the same reciprocal tariffs from China or anything like that. but I'd love to get your point of view there. Yeah, I think for most of the brands that we're talking about, everything's made in-house, and even a lot of the brands that are using generic movements,

like they're buying generic Swiss movements for the majority case. And like these, you know, $10,000 AOV plus. There are a number of brands, though, that will manufacture or use parts from other places. And then, yeah, they'll behave with those tariffs 100%, especially on like the, I hate to say the word cheaper, but on the kind of cheaper model. expensive. Plus expensive is the right word. You're right. What about any any further thoughts on the land dweller? Were you surprised at where it's being priced on on the secondary market at all? Where is it being priced? I think the steel is roughly sitting around high, high 30s, low 40s.

Is that correct? Yeah, we actually, we haven't seen any come through yet. So I'm yet to see like a transaction or a listing on bezel for it. So kind of waiting to see where that lands. But yeah, I mean, double retail, I said, I said that last week. I think that wouldn't surprise me at all. Since I spoke with you, I had a chance to, like, see it in the metal and put it on my wrists and play with it a little bit. And I liked it a lot. I didn't think I was going to like it as much as I did.

Frankly, I didn't want to like it. Not a fator at all, but it's just Rolex has this thing where they drop a model and it's just always good. Like, there's an executed finesse in everything that they make happen. and sometimes it's fun when it doesn't work out maybe and it always works out. So it's quite nice. I like the Rose build example better. Yeah.

But what is the base price at retail for a Rolex land dweller? I think this deal example is 14. 14. I think this word price is that. More than double. And then when the first one comes through, can you talk to me a little about like how you think about that deal on Bezell? Are you going to push that person to do an auction? Is that a better move for someone who was sitting on a little bit?

a landweller who's thinking about selling it or should they just try and price it through the normal marketplace? I think like we'll leverage the auction probably to do it. We did that with the Cupidus too. I think we sold if not the first cubits at auction one of the first to go, Cubit I to come out of the auction. And I think we'll do the same thing on the landowner side. The fun thing is like the folks that are buying the initial examples of the land dweller are doing so exclusively to be one of the first owners of the landowner. I think it's like that is the, you'll pay a premium for that certainly. Same thing happened with the Cubitus and any of these,

these like kind of new drops. If you want to be courtside wearing this watch and, and everyone know that you got it first, then obviously you're paying a premium for that. I don't know where it's going to settle, though. Like, uh, I imagine the premium for the initial examples will be certainly over double retail. Um, but then like, is it going to be evaluated like it's in,

in a Royal Oak kind of like very hyped. Is it going to settle to be similar to a day chest where it, you know, it trails. It's a little bit about retail, but it's, you know, not as exciting. I don't know.

It kind of splits the middle there. Obviously, I'm leaning more towards the hype train, but. Yeah, the thing that makes me think it potentially pretty sustainable is just a great looking watch and it seems to wear really well. I think I saw the picture you posted.

And it just looks great. It's hard to, it's hard to not like, even if you're not a Rolex guy traditionally. I like it when I saw. What's the sentiment around Patech right now? I felt like the general react, felt like the reaction to the Cupidus couldn't have been worse online.

I mean, like so many people just didn't like it. I think it looks fine. It wears well. But like, what's the feeling around the brand, you know, last week? Yeah, I think like it was cool to see them double down on a lot. of like traditionally interesting Pat-Tech models like there are not there are deviations from the sports models like you know there was a really badass Kalachava that came out in platinum with the salmon dial and it's incredible and I was my favorite watch and all the watches and wonders

to see in the metal and they came out with a 40 millimeter cubitist which I think is a is a very fresh take on the cubitist the other one is just quite massive and doesn't really feel right in the wrist in in my opinion I think the vibe against the The Cubitus is like similar to Rolex, Pat Tech tends to not miss. Like they, they kind of bat a thousand when they do things. And I think everyone wants these big brands to catch an L so they can talk shit about it and feel like, you know, they're able to say that. So I think the Cupidis like paired with a bit of like an interesting take that felt

maybe like it was overly modernized from a marketing campaign. like didn't sit well with a lot of early, early collectors. But yeah, I don't know. I think like I put the 40 millimeter one on. I liked it. You know, it's like they make great watches. It's a little bit angular and I'm a bigger fan of like a 5711 Nautilus. But, you know, if I was allocated one of those things, I would certainly wear it and I would rock it.

Love it. What, any interesting takeaways around the secondary market, you were on a panel. specifically about the secondary market. Was there anything? I'm curious what your major points were and anything else you kind of learned from other people that may have been on the panel or just that attended the event. Yeah, it's really cool to hear like how the brands are thinking about the secondary market and how the difference from how we're thinking about it and just, I think, where it's going.

I think it's, you know, we entered the space when the secondary market from a brand perception was, was kind of like, oh, don't, don't dip your toe into it by primary only. And now to see that, you know, I was in a panel next to Vashara on Constantine, I was in Sotheby's, and you have, you know, a bunch of representatives from all the top brands at the panel. It's quite cool to see that these primary brands are now embracing the secondary market is something that's quite interesting. And I think it allows customers to feel comfortable kind of interacting with the space as long as they're asking the right questions and they're making sure that they're focusing on authenticity and they're buying from reputable sources. And so it's been awesome for our business because it allows us the avenue to double down on our initial goal, which was just make a really, really safe, comfortable experience to transact in the secondary market, lower the barrier of entry to buying your first watch, but also just generally, if you don't want to wait in a wait list,

you know, come through and transact and buy the things you want. And that's why like the tariff conversation was super interesting for the secondary market, because if the premium from, you know, a lot of these pieces from the secondary market to retail is closed, close to 30% all of a sudden if a primary market watches 31% up from a cost perspective, that Delta is certainly meaningfully decreased. Do I actually want to wait on a wait list any longer? Or should I just buy one that's available in the market in the U.S.? Of course, sellers know that and they're pricing it up 10%.

But it still makes that difference smaller than it was. And so all of a sudden, maybe it's not as appealing to wait two years to get that Royal Oak versus just buying it right now and getting it when I want it. Never wait. Never wait to buy a Royal Oak. Always buy one just immediately. Just do it. What how does I'm curious, uh, how does influencer marketing work in the, uh, you know, for a company like Rolex when, you know, it feels like there's just this permanent stream of images coming out from the most influential celebrities in the world. And whenever I see one of the pictures, because I've been like involved with influencer creator marketing forever,

I'm always looking at the image. I'm like, does, you know, does this athlete always roll out of bed and want to wear Rolex, you know? Like, does Tom Brady always want to roll out of bed and wear like the It Rolex, you know, Submariner of the day? Or is there some type of, you know, sort of transaction happening behind the scenes, whether it's free watches or, you know, it's totally possible that they just love the watches and there's enough of them out there. But, but I'm curious if you have any insight there. Yeah, and it was a big influencer marketing year for Rolex. Like, prior to watches and wonders, they hinted that there'd be like a massive announcement and we were all kind of spiraling it because it, a new model, are they launching a new platform, like what's going on?

And they just announced Leon R. Caprio as an ambassador. And they dropped the whole like Rolex family page where it talks about like all aspects and touch points of creative, entertainment, sports, whatever, the dominance they have on like really influential people. They call their family page a celebration of human achievement. And it's so funny because I didn't see the announcement about Leo, but I just,

Instagram just like serves me pictures of, you know, celebrities wearing watches. And I'm always just like, Leo doesn't go anywhere without a Rolex on. And it seems like now it's like, okay, yeah,

he's contracted basically be like, don't leave the house without a Rolex. And everywhere he goes, he's just getting photographed. And it's just sort of like that natural, marketing. Yeah, and it's funny because I mean, Leo was a tag guy for a really long time. Like there's that story where the, the, um, in Wolf of Wall Street, he's wearing what looks like, uh, like I believe it's like a vintage GMT Rolex, but it's actually a tag for

that they had like intentionally made to look like it was like an old 80s, like, you know, stock markety looking watch, but it was, it was tag. So, um, I think the interesting part, you mentioned Tom, Tom Brady, like Tom Brady was an I WC ambassador for a really long time. I love those brands. I think it's a lot easier for those guys to be a Rolex ambassador than maybe another brand. They don't wear as much when you ask the initial question of like, do I feel comfortable wearing this everywhere? I would love to be a Rolex ambassador. I think they're very lucky because the tap and tapping to the brand that ever wants to wear a lot of the most.

Same with the APs of the world. They have a lot of amazing ambassadors there. But yeah, I think there's undeniably a set of collectors that don't want to be ambassadors because they want to wear a diverse collection, right? Like I, you saw the same thing. I remember sitting down with some early investors in Goat, and they capitalized that in the sneaker space where you had athletes and folks that didn't want to just wear

Nike's when they stepped out. They wanted to wear a more diverse set of streetwear and shoes and whatever it is. And the goat became the business that allowed you to catalyze that and wear multiple things. I think the same thing is happening in the watch space. And that's why I'm so excited to build kind of consumer brand, around the secondary market, if it's like you can get to a place where you can have a watch

sponsorship, but it's like more of a secondary oriented one. So you're making the statement that I like wearing watches, but I have a diverse collection. I'm not necessarily bound to one specific brand. That being said, not a bad gig to be a Rolex ambassador. I think it's pretty good. What are some of the known unknowns in the watch industry for the rest of the year? What are the things that we might expect around, but we don't have the final details on if we want to be completely dialed in going into the next watches and wonders. What should we be tracking today? Yeah, I think like the immediate term known unknowns is like kind of a general like I don't know what's going to happen with tariffs. It's going to change everything from that perspective or is it going to fiddle out and everything's

going to go back to normal. So we might so we might concretize that by saying something like if we see an increase in Rolex's stated prices on their website, that would be a reaction to the tariffs, right? Exactly. Yeah, yeah, yeah. Like, have those prices going to be passed on to the consumer. And I think, like, the consistent known unknowns that are always fun to, to, you know, posit ideas on in the watch community or, like, what's coming next, what's being updated next. We were talking about this via text before the episode, but, like, there's a very cyclical process that exists every year for a lot of these major brands, like, Rolex pretty much exclusively drops models that are in catalog and watches and wonders.

So all year, the watch community is throwing out their ideas, making renderings of what's going to happen next. The introduction of the chat, GPT image product has been funny to see. Just seeing all of a sudden, like anyone can render a green, whatever GMP or whatever it makes happen. And so the way that typically works is a lot of the investigative Instagram journalists in the watch space will look at what Rolex is trademarked or patented. Like Rolex Rolex trademarked Landweller a number of months before dropping Landdweller and everyone knew that, but it was like I didn't know what they were going to do with that. And then you'll get

renderings of the watches that are coming out and you don't know if that came from someone's Photoshop file or if that's real. And then to go tie this home with the influence of marketing, like Federer wore a land dweller somewhere in the Swiss Alps and dropped it on Instagram and it looked unintentional, but I'm sure it was intentionally, you know, very much planned. But so I think a lot of the known unknowns are like what's coming next, what's going to be dropped and then and then what's going to be discontinued and replaced. That changes the market significantly.

Yeah, that's very fun. Well, it's such a fun market to follow. It's so interesting. It has all the drama of like, oh, speculation about the next iPhone with the fake renders, except the iPhone hasn't changed at all. And so no one really cares anymore. But it's like within the watch world, it's like we're going from iPhone 3GS to iPhone 4 like every single year. So thanks for coming on and chatting with us.

Great to see you, Quaid. You too, guys. Thanks. We'll talk to see. Our official watch correspondent. Yeah. I'll be back. Super important.

Next, coming into the studio. we have Robin from Avalanche talking about nuclear energy and fusion in particular. I had a chance to go up to Seattle with Ben, actually, and do a whole tour of his facility and see how he's building a probably, I don't know, maybe the smallest nuclear energy product. It's pretty small. It's almost the size of a battery. You can kind of hold this thing. But fusion, not fission, much safer, a whole bunch of trains.

trade-offs there, which I'm sure we'll get into. And I'd also like to talk to him about some deregulation efforts that are happening. And there have been some announcements and executive orders, a lot of speculation about what might happen. Even Keith R. Boy was talking about long-term looking towards energy deregulation as a potential driver of innovation. We talked to a few people about how energy production is growing 20% a year in China and 0% in the United States. We don't love that. So I'm excited to talk to Robin and see if we can bring him in the studio and chat with him about how everything's going. Robin, are you there? I think we're trying to get audio.

Can you talk again? Can you hear me now? We can hear you. How are you doing? Awesome. I'm doing very good. How are you? I'm great.

Can you start with just quick introduction, the company, and then the announcement? Sure. So we're avalanche fusion. I like to say we're trying to build the world's smallest fusion machines. So we're trying to do a very SpaceX-style test-fail-fix approach to fusion, try to be really capital-efficient. The fusion machines we're building in our lab today are about 12 centimeters in diameter. That's the plasma core. And, yeah, raised a 40 million Series A in, you know, fall of 2022.

That was not an easy time back then. And we've been sort of in the lab grinding and improving these machines ever since. And so we've kind of made it to the point where we have. something interesting and so the announcement that we're doing today is we're announcing our new test facility which is going to be in eastern Washington in the Tri-Cities area called the fusion works and so what the fusion works is is it's basically a facility where you can come test different you know fusion technology so we're going to provide the neutron generators with our

fusion machines we're going to have a broad scope radioactive material license including the most advanced capability to handle Tridium that's available commercially. And so you can kind of think of this like a wind tunnel facility for fusion. So if you have some blanket technology you want to come test and see how it does, you can bring it to the fusion works and we'll hit it with neutrons from our machine and you can go qualify your hardware and then, you know, off you go. So I think I'm really excited. I think it's sort of like the next level and sort of, you know,

expanding the fusion supply chain workforce, course, all that kind of stuff. And then we've got some really great partners we're going to be working with in the eastern Washington as well. I remember talking to AJ at Hermius building hypersonic planes. And one of the main hiccups to their progress is just getting wind tunnel time. And so I haven't heard of anyone building a wind tunnel startup. Maybe that's a good idea.

Can you talk about where you're seeing demand, who you were talking to that made you realize that you needed to build a facility that would service a broader swath of the fusion community? Yeah, so funny, should they mention that. So like AJ and Hermius just recently announced their heat facility, right? Oh, really? Which is like basically, so what it sounds like after that conversation, what they did is they went and built a hypersodontics test facility themselves. And so the Herbius announced with, I think it was a couple weeks ago now, was basically they are opening up their test facility to the rest of the DOD because there's a shortage of hypersonics testing. That's great.

And so it's kind of like, duh. like if you spend all this money standing up this like unique thing um you know if it's operating 24 seven obviously it's going to be a much better facility and you probably don't have enough work to operate this thing at that cadence so why don't you open it up um you can bring in some revenue while you do that and you can make this like really awesome facility that's just sort of like grinding day and day and out getting really good at what it does and so it turns out that idea applies to aerospace and hypersotics. And I think it's also going to apply to fusion at the end of the time.

Yeah, it's commoditizing your compliments. Okay, I got to ask. I got to ask new presidential order this morning. Zero-based regulatory budgeting to unleash American energy. There was a whole bunch of mumbo-jumbo in here, but it seems important. What's your take? What's your take? Yeah, I read that this morning.

I kind of love it. You know, this idea that, every now and then a regulation is going to come up for review and it needs to justify its existence is kind of awesome. I can give you an example of how it applies to fusion. So, you know, Tridium is regulated at the state level if you're an agreement state. So like we're working with Washington Department of Health to get it licensed. But the NRC has this rule that if you can't account for 10 curies of tritium in your system, you have like 30 days to report it as an incident to the NRC that you've like lost 10 curies of tritium and to to give you an idea of how

small that is an exit sign has like 25 curies of tritium so you know here you are running your fusion machine if you can't account for 20 or 10 curie because it got like absorbed into the walls and it's self-shielding so you can't even measure the x-rays from it you just don't know where it went yeah uh uh that's a big deal and you need to report that as an incident but like if someone like throws an sign in the garbage and nobody knows where it went like nobody cares right there's no like what do you mean an exit sign like like an LED exit sign you see in a theater or like what's that what do you mean by exit sign is that something like like an exit sign above the door right like let's say the fire there's a fire and the lights go out okay and you have this glowing exit sign yeah those are powered by 25 curies of tritium

no idea yeah and so like we decided a long time ago that you know, people being able to see how to get out of a birdie building in the dark. Okay. Safety associated with that was worth the very small risk of having Tridium in these exercise. I had no idea. We decided, you know, human safety over like some very, like, unlikely chance. They're going to freak out journey. He's going to avoid every exit sign he sees now because he's super paranoid.

They're made with microplastics too. Yeah. Oh, no. Yeah, that's hilarious. Can you give us a broader picture? Yeah. So there's no agency out there investigating missing. Yeah.

Exocides, whatever one goes missing, right? That would be like silly. Yeah, yeah, yeah. And so, you know, I think this NRC 10 curie missing thing is an example of a regulation that maybe made sense of the time. Yeah. But man, if it had a sunset clause and had to go defend itself, like everyone in fusion hates this one. Sure, sure, sure. Unanimously, like I feel like this was an example of something that probably should die.

Yeah. You know, if this executive order is a way to get rid of some of that stuff, I'm all for it. At the same time, it seems like my kind of novice understanding of it is that on the fission side, we are highly regulatory constrained and there's high regulatory risk and relatively low, high regulatory risk, relatively lower science risk in fusion. We're higher on the science risk and lower on the regulatory risk. Is that the right frame to think about it? And can you just give us a brief overview of how American science,

scientists broadly are thinking about progress in fusion. We hear, you know, papers go viral. Oh, it's solved every couple of years. We're trying to get to, you know, net energy created. Where are we on the path of progress towards, you know, this amazing fusion future? Yeah. So, I mean, we've demonstrated ignition or Q greater than one using lasers, like hitting, you know, a pellet of deuterium tritium with lasers and making more fusion energy in than the laser energy that went out. That's awesome. And so there's a bunch of fusion startups that are working on laser approaches to doing fusion energy. There's another crew that are working on like magnetic confinement or combinations of that. That's like what I would consider what we're doing

essentially. CFS, Helion or ZAP Energy or other companies that are kind of doing that. They have not hit Q greater than one yet, but we're all building machines that have a chance of doing that in the next couple years. And so I actually think it's going to be really exciting. There's a chance that one or multiple magnetic confinement fusion companies are going to cross Q greater than one during this administration. And those machines look a lot closer to what a fusion power plant is going to eventually look like than sort of the laser fusion techniques. And so if that happens, and I think there's a good chance there is, that's going to really change geopolitics around energy. You're going to see this, like, amazing, you know, rush to kind of try and commercialize magnetic fusion confinement, which is

going to be super exciting. To give you an idea of like the regulatory split. So right now, there was a decision about a year ago that fusion machines are going to be licensed like particle accelerators and specifically on the radioactive materials they produce. And so what that does is that pushes the regulator you work with to the state level if you're in agreement state. So in Washington state, it's an agreement state. We work with the Washington Department of Health to license our fusion machines as particle accelerators. The process is essentially we design our radiation vault and do all the calcs with that.

And then we apply for a particle accelerator license. They come in, they review it. They do a tour. We show them what we've built. They have some suggestions on how to improve it. And then a month later, I get like a letter in the in the mail from the Washington Department of Health with my particle. accelerator license, which I hang on the machine and now I'm licensed. So that's like pretty great, especially for like, you know, at the point where we're building

experimental machines. And then you contrast that with what Fission has to deal with. It's like many, many years of regulatory work working directly with the NRC. It's a really, really tough process. And we haven't actually even licensed any of these small SMRs that we're trying to like build here. So I hope that Vision gets some regulatory relief in this administration. They definitely deserve it. We should probably talk about Texas and Utah that are suing the NRC to try and get regulatory control over small modular reactors. Yeah, can you break that down for us? I saw a couple founders tweeting about that, but I didn't really have full context. Yeah, so I went to this investor event in Dallas, Texas

in December that Belerion Ventures put on. And Rick Perry, was the special guest speaker there. And I almost think I was in the room when this got decided. But like Rick Perry's advice to all the vision people in the room was you should join up with Texas. You should sue the NRC because there's some clause in the regulations that say the NRC has the right to regulate large power plants. But it's not clear they have the right to regulate some of these smaller things because the risks associated with them are so much less if something goes wrong. And so I like, you know, Brett Cougolmes was in there from Last Energy and he's like one of the main parties to the lawsuit. So I feel like I was like an observer to history when the fishing community decided we're going to sue the NRC and get them out of Texas.

And if they succeed and Texas becomes the regulator for SMRs, that's going to be really exciting in terms of their ability to innovate and go faster. Yeah. Can you talk about the positive sum nature of sort of nuclear broadly? It feels like it's been so hard to build in the space that everybody just generally wants to see people be successful in the space. So maybe it actually creates a dynamic where people that are- People-E-Compies suing each other. They're suing the government, which is very different than what's happening in AI and every other startup. I don't see a lot of spying going on like we do in B2B SaaS.

Seems like a lot friendlier community. it is i mean the fusion people definitely you know kind of get rid by the fission people's like if only you could solve your science blah blah blah um but you know like i want i want my vision siblings to you know deal awesome i think there's room for both fission and fusion i think my personal belief is that vision is going to go bigger like hundreds of megawatts to gigawatts and fusion is going to be really great from the kilowatts to a few megawatts um and fusion is going to be really great for mobility and stuff. So like autonomous vehicles, fusion powered, like an Andril dive X-L or something like that. Like I think that's going to be a really great application for fusion

in defense and space. I think the Russians are the only people crazy enough to put a vision reactor on like an autonomous vehicle. So I don't think that's a thing that's going to happen with vision. But I think, you know, I think if you have sites where there's people around like guards and stuff. That's a great use case for fission and it's going to scale like nobody's business if they can get the regulators, if they can get some regulatory reef and get some reasonable regulations there. So I'm really excited to see what happens over the next couple years. Can you talk a little bit about lessons from the past two decades in in rocketry and specifically lessons learned from how Elon and SpaceX approached building rocketry?

versus other players in the space and the decision to go, more modular, smaller, higher iteration speed. I've talked to you a lot about that, but I'd love to hear you kind of retell that story. Yeah, yeah. I mean, the idea that, like, so like I was in Blue Origin for six years working on New Glen. It was super exciting to see it fly in January for the first time. And so, like, one of the things we learned there was like the importance of size, right? If your rocket is too big, all of a sudden, there's only like three suppliers in the whole country where you can get castings made. is made and now you're at a nine month, every iteration takes nine months, right?

And it's just too slow. So, you know, that was the thing that we really tried to bring from New Space is like, you don't necessarily know what the final answer is, but you need to be able to try things. You need to be able to try things quickly and have low barriers to build things and test things. And so that's that's sort of the key takeaway I took away from New Space and we're trying to bring that to fusion, right? Like very low barriers to test. we said at the beginning of Avalanche that we want a physicist to be able to go from like,

I have an idea to testing it in a week. And we've actually achieved that a couple times. Like someone had an idea on Monday. They worked with the mechanical team to design it. And then we fathed it on, you know, Wednesday, Thursday, broke vacuum on the machine, implemented it, pumped out over the weekend. And then off we were going testing on Monday. Like we've hit that cadence a couple times.

And it's really kind of exciting to see that happen. And I believe that's how you solve really hard. engineering and technical problems as you try it and you iterate on the thing and eventually you're going to find something that works and then off you go. Yeah, can you talk a little bit about the pipeline of talent in the fusion community because it feels like there's a lot of like thankless PhD tracks that are more theoretical. But again, I've been to the facility. It's a lot of wrenches and hammers. It's a very different culture, I imagine. But what has that been like scaling the team and getting people to adapt to kind of the different style of experimentation and work?

Yeah, that's a really hard one, right? There is not a lot of places you can go to learn how to do fusion. And so what we've taken is a very broad approach. Look, we'll teach you fusion. We'll teach you plasma physics, but you bring your mechanical engineering degree, your electrical engineering degree, right? You know, we love people from Blue Origin, SpaceX, Tesla, who've kind of worked in that kind of an environment that like go fast, try stuff, iterate, stuff like that. And then we will teach you the class of the physics. Don't worry about that. And then we'll turn you into a fusion engineer, if you will. So I think that's kind of how we're going to have to do things for a while until sort of the universities catch up and you can take a degree in

fusioneering, if you will. And, you know, I would love to put a couple fusion reactors on campus at some universities. So like, like programs can get some experience actually like operating them and building them and running them. But that's maybe something for a future. Yeah, there are. We'd like one for the studio. Yeah, yeah, bring it in the studio in case the power goes out. We'd love that. Yeah, I mean, there actually are some small nuclear reactors, some fission reactors at like MIT, I believe, and some. And they work on them and they produce a small amount of radiation and there's a little bit of risk, but it's much lower than Chernobyl, right? Oregon State U has one.

I've been there, like I stood on top and like look down at the division. Yeah. Washington State, WSU has one. Last question. They're called trigger reactors. So they're used basically for nuclear engineering programs to teach.

You have like basically students like undergrads running these things. That's awesome. They're licensed by the NRC as nuclear operators and something like that. I have one last question. I mean, we saw, everyone saw Oppenheimer and everyone knows the progression of the, the, the, the, the, the, the, the, the, the, the, the, the, the, if these, if these, uh, you know, uh, if these, uh, if your devices really get baked into, you know, society and, and autonomous vehicles, is it even possible to turn it into a weapon? I know, obviously, we'd want to be careful, but from a, from a, from a science perspective, like, what, is it just impossible?

No, it's a, it's a vacuum chamber, right? And it needs to have very, very deep vacuum to maintain the plasma and do the fusion. And so we actually had a machine break on us when it was running. So we had this cathode that was held on with the set screws. It was like one of the first janky machines we ever built. And this set screw came loose and the whole rod cathode fell through the bottom, hit the window and like ended up on the floor in the factory somewhere. And the whole thing is just like imploded and spread glass all over the thing. But like nothing bad happened.

Like, it's just like, why is the cathode on the floor? What's going wrong here? I don't understand. Oh, if you should, it's going to be very safe. You know, some of the conversations I've had with the DO and he's like, well, what happens if I shoot it? Like with like with a gun. It's like, well, save thing with the cathode.

Like it'll implode and then it'll stop and that'll be the end of it. Yeah. Makes sense. Yeah, the real problem is just getting it to produce energy and that's, uh, that's the path you're on. It's hard enough, right? Yeah, yeah, it's hard enough. It's hard enough to make a case.

Battery. Can you, before you leave, could you give us like a one minute rant around the, you know, when Fusion is really working and we have these sort of small, I don't know exactly, sort of like cooler sized devices that can create energy? Yeah. Where are all the places that we're going to put that? Black mirror just dropped. I want to hear the white mirror version.

I understand like, you know, the DoD applications. But like, yeah, is it like backup power, you know, when at Hollywood, you know, You know, if they even make movies then in person, is it, you know, for your house, right? Yeah. You know, what are the, what are all the different use cases that you're excited about? The Adam Punk future. I mean, the practical one is they're probably going to be containerized. And you're probably going to just be shipping containers all over the place, like powering a neighborhood. You know, that's like probably what's going to happen.

Everyone wants this containerized one to five megawatt form factor, it seems like. So that's like probably what's going to happen. But like the atom punk like future would be like you're driving around and you've got one in the trunk of your cyber truck. And like you don't have to like, you know, supercharged if you're going to Eastern Washington because it's actually not that many charging stations. You could just like flip on the fusion reactor and go go on fusion power. I've driven around San Francisco in the Dolorian with a plastic 3D printed fusion reactor last year or Deep Tech Week. I'd love to do that with a real one someday. That would be fantastic.

Well, thanks so much for stopping by. We'll talk to you soon. Congratulations on to the launch. Yeah, congrats. Cheers. Bye. Hoping on next, we got Russ. Is that right?

Yes. From LiveKit I.O. That's right. That's right. Coming into the studio. We got our next guest joining. Some big news. Big pivot from watches to fusion reactors.

But that's the nature of the show, folks. We're men of many talents. We keep you on your toes. Keep you on your toes. we'll do dedicated days where there's one theme some days. You just get a tour of the entire technology industry. So welcome to the studio, Russ. How you doing today?

I'm good. I'm good. So you're saying that I'm following fusion reactors? We're following fusion reactors. And immediately before that, we were talking about fine watches made in Switzerland. So we got a little bit of everything today. Taking us on a world win tour. I don't know if I can follow fusion reactors, you guys.

I'll try my best. Well, we're excited to talk about voice AI. Yeah. But why don't you introduce yourself and the company, and then we can talk about the news and go from there. Sure. Yeah. So I'm Russ.

I'm the CEO, co-founder of a company called LiveKit. LiveKit. Yeah. So LiveKit, what we started actually is an open source project. So back during the pandemic, you know, if you think about all the technology that Zoom has underneath, We started an open source project that kind of built something similar to all the technology that underpin Zoom, but as an API for any developer to integrate similar type of technology into their app.

So you could connect any two people or any number of people on the planet, kind of instantly with less than 100 milliseconds of latency over video invoice. Fast forward, like about a year and a half, two years. and we built a demo when ChatGPT first came out where you could talk to it instead of text with it using LiveKit's infrastructure. And a few months later, OpenAI ended up finding it, wanted to build a voice interface to ChatGPT,

and we started to work pretty closely with them on voice mode and then advanced voice mode after that. And kind of almost by accident, I guess, we got pulled into this AI space, Voice AI was not a space back then. Happy accident. Yeah, I remember our Series A, which we closed at the, we closed the first tranche at the end of 2023, 2020,

just at the start of 24. And everybody was just telling me that, hey, this is cool, but voice isn't going to be a thing for three to five years. And then, of course, GPT40 happened. Yep. Can you, yes. Can you steal man at all? Why speech to text is still bad?

on my iPhone. I just really wanted to be good. John's a big voice guy. He's always, he's always, he's always using LiveKit unknowingly. I worked it,

I worked at a voice AI company in 2010 or something. Oh really? Yeah, it was, you know, Dragon naturally speaking. I do, yeah, yeah, yeah. Dragon eventually acquired this company. And it was like,

it was direct competitor to Siri, Apple bought Siri, and Vlingo, who was purchased by Dragon. And their primary, it was such a funny star. Silicon Valley, the TV show. Blingo actually. It was a good company. I think it was a good exit for everyone involved too.

And their primary target was BlackBerry. And their whole goal was let's get pre-installed as BlackBerry. But there was a lot of chaos in the market at the time with, it was a $20 app. Then they had to cut their pricing, do an ad model, all these different things going on. But I'd love to know like just if you were in charge of, you know, Apple, how would you improve speech to text? Speech to text specifically, like transcribing text. Oh, well, anything in voice agents or just just improving the customer experience broadly. Like surprise and delight, right? Well, you know, it's one of those things that I think is funny, like people talk about hallucinations as this bad thing. Yeah. But in a lot of ways for like kind of this speech to speech interaction model where you're talking to an AI, hallucination is a feature. So if you remember like your Alexa or Siri, like when you were talking about this, you know, 2010, 2012, you go and you ask it a question and it's just going through a bunch of like a decision tree. And if it hits, you know, a dead end and it can't answer the question, it just doesn't do anything or it says I can't answer the question. And so the second that it can't do something you expect it to do, all of a sudden you, you know, it's such a punishing user experience. You just don't want to use the thing anymore and you stop using the thing. But with kind of these new models,

that actually hallucinate answers into existence, even if it doesn't know the answer, it tries, you always get a response that comes back from the model. And for voice interfaces that actually can be more of a feature than a bug, it also helps that these models. Yeah, it's funny because it's human, right? So like if you're having a conversation, John asked me a question, I don't know the answer.

I go, well, like, is it like this thing? And you're like, no, it's this thing. But for some reason with like these older generation models, it was like very annoying to go down that wrong sort of. Exactly. And it's always the same answer, right? It's like, I don't know, or I can't help you with that. I can't help you with that. Or I'll search Google for you or whatever. And so I think that like just having these LLMs that have ingested the entire internet and can always generate some plausible answer for everything, whether it's 100% correct or not, I think that that just makes the user experience so much better

with these kind of modern or contemporary voice agency you interact with. So, you know, I think the other thing I'll say related to the Apple stuff, and that's improving very quickly is a latency. So like we have a lot of providers out there now, GROC, Cerebris, folks like that who can run inference much faster than even a year ago for some of the model providers. And now like LLM inference can actually be done in, less time than generating speech with TTS.

And so I think like getting that latency end to end down, you know, to 300 milliseconds or 500 milliseconds on average for like turn latency, that that kind of helps you cross this uncanny valley for voice AI. I mean, should we be thinking like even faster than 300 milliseconds? And is there any efforts that you've seen to bake some of these models down into silicon? We've seen what etched is doing with putting

transformer architecture on silicon. You could imagine that once mid-journey gets good enough or kind of hit some peak that they would just bake it down into silicon, we would just have image generation models, you know, A-6 essentially, like what happened with Bitcoin. Is that the future here? Not that you would pivot into hardware, but maybe you would like vend your software into a hardware provider at that level? Well, so we like, I mean, we partner with like folks like Cerebris and GROC. And so we allow you to kind of plug in their models and plug in effectively their hardware accelerated inference.

And so we're compatible with that world. I think on the inference side, you can definitely, I think it's going to continue to push as low as it can get. There's obviously some limits. You know, it's a tradeoff between kind of capabilities and level of knowledge and how fast you can kind of, you know, run that that pass through the model to get the result out. So there are tradeoffs, of course, that follow the laws of physics. But there's also kind of diminishing returns after a while. To give an example, I once built this, this cerebris demo. I used like a llama

7b or 8B, llama 8b. Have to remember these numbers on the primary accounts. But llama 8B hooked up to cerebris and I got a bunch of feedback on that voice demo that, the model was responding too fast. And can you slow it down? And it's kind of going off the rails a little bit. Yeah, yeah, interesting. I mean, on that note, is there a kind of like, like at inference time fine tuning that needs to happen

on the voice agent's side to give the human listener the appropriate interaction? Like some people want to have, you know, this really fast back and forth conversation. And I'm like, just get to the point. You're, I'm just trying to book a flight. Can I just like, just like, just give me the information as quickly and condensed as possible.

Other people might prefer an agent that speaks slower and really draws things out and gives the full context and then lets them answer and kind of ping pong back and forth that way. Is there any movement towards like that level of fine tuning that you think might happen in the future? Is it already happening? I don't know. Yeah, it's already happening. So I think that there's kind of two different flavors of this and you can kind of combine them together over time. So the first one is that if you've interacted with like the real time models, like the ones that are, so there's open AI real time API, there's Gemini Multimodal Live API.

There's a few others coming out as well. And for all of these models that natively understand audio, you can actually tell them to, you know, whisper or slow down or speed up or act hyper. you can kind of give them an explicit signal of the style or the way that you want them to communicate with you. So that's already available and possible with these models. Then there's another part of it, which I think will come in the next year or two, where the model will implicitly be intelligent enough to pick up on what your pacing is or your state of mind is, just based on the way that you're talking and expressing yourself. And also if we weave computer vision into it, it might see you and understand from visuals,

okay, this person is stressed or this person seems like they're in a hurry or this person is calm and like relaxed and you can kind of tell that by your body language and by the way you're speaking and it can automatically adjust you in the same way a human would be able to. So you guys work with a lot of the biggest companies in voice already. Open AI, speak, character AI. you're working with Tinder, I see like really broad swath of company, Spotify, Oracle, things like that. Are you seeing as many startups as you would like sign up and start building in Voice AI? It feels like a category that for, if you go back like five years ago, people were really excited about Voice,

but then you had these sort of like high profile companies that didn't quite work, but, you know, maybe in many ways they were just too early. So what are you seeing on the on the startup side in terms of new companies being. specifically to leverage voice and LiveKit and the underlying models. Yeah, we're seeing probably about, you know, we're doing like thousands and thousands, many, several thousands of, of like signups to the cloud product, our commercial product. And most of those, the vast majority, are startups and growing companies. And out of those, probably around 75% or so 80% of those signups are our voice AI companies that are building voice agents.

And so the way that I kind of see the market segmented to a degree is that you have the large AI labs and they have popular consumer apps. And a lot of them are building kind of open-ended voice agents that you can talk to about kind of anything, right? Their assistance, they do question answering. They do therapy for mental health, all kinds of language learning in the case of speak. And then on the other side, you have these kind of pockets that are what I call kind of voice native systems. And those are really anything you pick up the telephone to, you know, when you call someone on the other end, call a business, there's someone that answers that line and they're either, you know, doing patient intake at a hospital, or they're doing loan qualification or insurance eligibility checking.

there's a lot of these kind of business process flows where there are pockets that are like really large in nature so customer support is the one that gets talked about a lot right like some of the markets at Sierra Decagon and folks like that are playing in and so for all of those systems we're seeing tons of startups flood into those spaces because it is now viable to take an LLM and have like a voice stack like LiveKit for example that's hooked up to that LLM and you can build like an automated version of whoever is picking up the phone on the other end.

Well, yeah, the thing that I'm excited about, everybody's gotten on one of these like robo-cx calls. And, you know, I've had the experience in the past where I just say talk to a human until I get to a human because I know they're not going to resolve like the issue the way that I want. And I'm like the annoying guy to the robot. But I do feel like there's a point here quickly where the AI can be considered. better than the human on the other end because they're not tired. They're not in a bad mood that day. Like, you know, they're really, you know, they just have like high energy because they're code. He's going to be a flippening and it's going to be like, oh, I'm talking to a human. Robot, robot, please put me on there with the robot. Yeah. I was, I was on a customer support call

with Comcast the other day and I was talking to a human for sure and they were trying to look up something for me. And then they started asking me if I'd ever seen the same. and if I'd ever spent time in the snow and I was like, what? And where do you live? And I'm like California. They're like, so do you experience snow in California? I'm like, well, in Tahoe, yeah, I guess. But it was so weird and I was just like, am I talking to an AI right now or is it a human?

And if it's a human, I kind of want the AI. Yeah, yeah, it's very bizarre. Super awkward. Can you tell me about some of the more like nuts and bolts enterprise customers you're working with? I see this case study from playback. We actually had the CEO of Playback on the show. Live streaming for sports makes a ton of sense. Do you know exactly how they're leveraging live kit

and how you can kind of give us a more concrete example of like their infrastructure basically? Yeah, for sure. So playback, they're also gonna integrate AI into that flow as well, like a voice-based commentator or whatever that can watch the game and provide an overlay for fans. But they use us for a pretty different use case.

In that case, they have these kind of court side cameras like NBA games, MLB games, and all of that. And these cameras are, you know, the way that you watch sports on TV. And so one interesting part about, and this is kind of like a deeper technology thing, but one, if you've ever watched like the Super Bowl or the NBA finals or any sports game on a TV and then like you're texting your friends about it and they're watching the game or there's someone in another part of your house that has a game on on a different TV, you'll notice that you're not synchronized.

You're seeing plays before they are, sometimes up to like 30 seconds or a minute before. And so the technology that we use for live broadcasts of sports games is not true real-time technology. Not everyone is synchronized in the same way that in 1969, when the lunar landing, everybody saw that at the same time because there's a server that is sending all these kind of broadcasts out over TV antennas that are receiving these things. and it is truly a shared experience. And so playback, what they do is they ingest the video feed from these cameras at Courtside or at the MLB game. And they're ingesting it using LiveKit. That goes into a single system. And then LiveKit's cloud network effectively, everybody who's watching that NBA game, for example,

we are effectively shuttling those bites from that camera in a synchronized way to every single. person that's watching that game. So when like Steph sinks a three, everybody who's watching that game through playback is seeing that three get sunk at the same time, able to truly have a shared experience of watching their favorite team play the game together, cheer together, all of that stuff. And so playback is effectively using us for the backbone of all of the audio and video kind of transmission that is going on within their application. Makes sense. Before we let you go, we should probably talk about the news today. Yeah, you guys had a bit of an announcement.

Why don't you share a little bit there? Yeah, for sure. So kind of how I mentioned at the start, right? We started as an open source project, you know, connecting humans to other humans. But now we have found ourselves kind of operating at a very large scale, connecting humans to machines. So using voice and computer vision. And so we closed a series B today that's led by Altimeter Capital. That's right.

Gerson, let's do it. Let's go. Let's go. We love BG. Yeah, yeah. So we close that round, and now we're really going after voice AI. We're building like an all-in-one platform for anybody, any developer to build a voice agent. Give us the stats.

One million. Two million. How much did you raise? 45 million. I got to do 45 million. I got to do 45. I'm not going to submit it to that. Congratulations.

Amazing milestone. I think, you know, clearly like a five-year overnight success. Yes, yes, classic. Classic example. But it's cool. I love these stories where you sort of work on a really hard problem and then discover, like, you know, the most powerful application, you know, years later. So congratulations to you and the team. It's great to have you on. You can be our new official voice AI infrastructure correspondent. You love that line. I'll make a clone of me and then that will be your official. Yeah, voice ad correspondent.

Cool guys. That's perfect. That's perfect. Well, thanks so much for stopping. Yeah, thanks for coming on. We really appreciate it. Thank you, guys. Talk soon.

Talk soon. Cheers. And next up, we got Ev Randall coming in from Kline of Perkins. I love how like there's always like you can always go niche enough to make somebody a correspondent. Yeah, exactly. Hyper, hyper niche. Anyway, $45 million.

That's a great round. There are so many funding rounds happening. Now that we're like in the thick of it, I'm realizing like, wow, the money is flowing in Silicon Valley. It's great. But we will get a full market update from Ev, hear what he's seeing over at Kleiner. Actually, cross paths to them at Founders Fund two years ago, maybe two and a half years ago. Great dude, wrote one of the most probably banger memos during the Zerp era,

all about Tiger Global and crossover investing. And I want to follow up with him on that. But I think we have them here. Ev, are you there? Welcome to the studio. Hey guys, Jordi Coogan. Great to be here. Longtime friend of the pod. Finally, dude. It shouldn't have taken this. I know. You should not have taken this long. You're in our original pitch deck because you posted like, how can I go long, which

hilariously caused a lot of problems for us because someone created a fake pump. Dot fund token about us. And everyone would say, oh, they endorsed this. And we're like, we did not. We just. Yeah, by the way, I didn't get rich off that. So that is, I have a bone to pick with either the creators of the coin or you. I don't know. Don't even joke about it. Just it's bad matter what. If you say anything, they'll make a token. They will turn it into a coin. Anyway.

But it's great to have you. Thank you for the early support. Why don't you, John gave a little introduction, but why don't you introduce yourself? Yeah, yeah. What are you working on most recently? What's interesting? Yeah, of course. Yeah.

So I'm a partner here at Kleiner Perkins. KP, obviously one of the more storied firms, kind of inventor capital history. As a firm, we've led early rounds of, you know, companies like Amazon, Google, Genentex, on microsystems. More recently, companies like Rippling, Figma, Glean, Harvey, and many others. You know, ventures evolved dramatically, obviously over the last five years, but especially over the last decades. And I think what made KP special back then is what makes it special now, is that we're a very small team. There's 10 of us total on the investment team. And we all just care very deeply about the craft of venture, founders, company building, and we want to

keep it a craft. So we, I mean, we're fully multi-stage. So we have an early We have a growth fund. We announced a new set of funds last summer that were $2.1 billion in total size. We can do seed through pre-IPO. I personally help head up our growth fund. So I spend, you know, the vast majority of my time kind of on Series B and beyond investing, but all of us invest across the venture and growth funds. So I've led, you know, our investments in companies like Flock Safety, Huntress, a cybersecurity company,

captions involved in our investment in Rippling and a few others soon to be announced. How would you how are you processing the tariff news where you just texting every portfolio founder like hey have you seen this? Yeah sending them a sending them a screenshot of the NASDAQ 15% What are we doing about this? Hey, hey, hey, what are we doing about this? Hey, hey, what are we doing about this? Is there a memo coming? Is there a black swan memo? Are you going to get in the black swan memo game? Give Sequoia a run for their money. So Koy has pretty good market share of the of the black swan memo. Yep. market and so it feels like you know it feels a little too memetic I'm a little too

kill-pilled still to do something that memetic yeah well but I mean how have you been processing it do you think it affects tech do you think it affects any of the I mean like immediately the reaction was like all the stuff the NASDAQ was way down but at the same time like even semiconductors were you know on day one kind of excluded was there is this something that all founders should just tune out or is there something here to pay attention to I think I think it's probably both, right? Like, I think Mike McDonough from Lightspeed had a good tweet or post today or yesterday where he was like, I'm getting a lot of questions from, maybe it was LPs about the tariffs

and they have to realize, you know, I invest in companies that aren't going to exit for seven to 10 years. And so there's like something about the like duration of the asset class where, you know, something like economic turmoil that happens today if you're investing in the venture asset class isn't going to necessarily impact like your exit because ultimately the founders and the board and the management team have some control over when they exit. And so you can kind of theoretically wait a little bit for a better environment to either IPO or sell or whatever you want to do to get liquidity. I think at the same time, the like underpinning, like the undercurrents of what's

going on, especially politically, are immensely relevant for startups and tech. Like obviously everything thing around Taiwan and just like semiconductors generally. And if that escalated, how bad it would be for not only technology companies, especially in AI, but like just our day-to-day lives. Like I think our day-to-day lives would pretty much come to a halt. And I don't think that people really have an appreciation just for how permeated, you know, semis and other critical materials and inputs from places like Taiwan are in our day-to-day lives. So I think it's of critical importance. But again, it's like, it's one of those things

where it's like, well, what are you going to do about it? You know, unless you, there's very few, you know, there's very few companies or folks that can actually have an impact on this. And so I think for most companies, the best advice is always just to go heads down and. Well, yeah, let's talk about one of the immediate impacts, which is the IPO window, which was so briefly open, feeling like it's slam shut. I love when it's open personally. I'm a huge kind of I wake up, I get out of bed. Huge fan of open windows. Huge fan of open IPO windows.

No, but poor Clarnah and Stubhub, you know, they've been waiting around. Moment of silence for Clarno. Yeah, moment of silence. Brought to you by ramp. No, but yeah, I don't even know if like, you know, Circle gets out at this point. But at some point, you can imagine the companies that end up going out are the ones that have to go out. And then that just becomes a vicious cycle where they maybe underperform because they're not these sort of best in class companies.

And then it's like, well, it's really shut now. We're not going to touch it. But what's your kind of read on that situation? Yeah, I think, I mean, obviously before this all kind of blew up, obviously it was unfortunate that there was like, you know, four or five IPOs kind of on the shelf ready to go right when this blew up. And so it seemed like, ah, you know, we were opening things back up. and then immediately some Black Swan event happened and it kind of shut it back down.

But for, you know, like a full almost nine to 12 months before that, it was a solid environment. It wasn't, you know, IPOing in 2021, where you could go and, you know, spec for 50 times error or something. But it was not a bad environment. And I like, I kind of have a slightly orthogonal take on the IPO window, which is there's this kind of like rock and hard place situation for the IPO market, which sums up to like, one, do the. IPO markets want you, but then also just as important, like, do you want the IPO markets? So on the first one, I went and counted this morning the number of public software companies on, you know, the Meritech kind of software comparables index that they run that currently have over $500 million in ARR, which is like an incredible achievement. And right now, there are over 80 public

SaaS companies that have more than $500 million of ARR. So if you want exposure to some trend or idea and you want it to be a big company that is profitable, there is very likely a public SaaS company that can give you exposure to some theme or trend or idea that you want that already exists. So it is much, much harder today to have an IPO that bankers, long only public funds, like the people that make the IPO machine work, it's much harder to have an IPO that they're going to care about. Like you probably need to be free cash flow positive. It probably helps a lot if you're closer to a billion dollars of revenue scale, then even $500 million, which is already incredible.

And you need to have some unique angle or something that's really cool or unique in the story, like the qualitative kind of framing of your IPO, that's special. So I think, like, one, that's like, that's already hard enough. And then on the other side, in terms of like, do you want the IPO markets? I think what we're seeing out of several of the very, very best tech companies is that they've gone to pretty drastic measures to not IPO. And I think there's like many reasons why you'd not want IPO. So SpaceX obviously is kind of the poster child of this.

But then you saw Stripe obviously do that that really large kind of like RSU catalyzed round at like $55 billion, I believe it was. Databricks obviously in Q4 did a mega raise for the same reasons. And the asset class has grown so much that if you're an amazing company, compounding your intrinsic value at 25, 30% a year at scale, you can raise practically unlimited capital. Like, Databricks raised $10 billion of equity. They did a, you know, multi-billion dollar debt raise, too. But 10 billion of equity in, like, I think, you know, it was probably like three months. Like, it didn't take very long to raise an unbelievable amount of equity. So the fund sizes have gotten so large and the amount of capital that's available for these companies is so large,

as long as you're one of these, like, top 10 companies that if you don't want IPO, you don't have to. And I think there's several reasons why you wouldn't IPO. The main one, or one of the main ones is that you can be a lot more aggressive on M&A. you know, Databricks, for example, you know, bought this company Tabular. I think the reported valuation was $2 billion, and it was also reported that they were either pre-revenue or close to pre-revenue. If you were, you know, Snowflake and you're a public company, you probably can't buy a pre-revenue company for $2 billion.

Like your stock probably goes down 20% the next day. Almost said effing, but I think this is a PG show, so I'm not going to cuss. But, you know, the stock probably goes down 20% if you do something like that. That's an extremely strategic acquisition that Databricks is able to do because it's completely founder controlled. No activist can buy 10 to 15% of your stock. No one can mess with you. So I think there's good reasons to do it.

And the only thing that you have to figure out if you don't want IPO is you have to give employees liquidity. Like you have to give employees regular opportunities. We love when employees get liquidity, but what about if you're a, you know, just a $2 billion fund and, you know, you're trying to show DPI to your LPs. are you seeing these sort of funds that, you know, like KP, that there's now funds that are like effectively like four or five times larger and then these sort of crossover funds that are getting involved. And are you seeing more like do traditional venture funds like KP ever try to sell

into these rounds that are intended more for employee liquidity? Is that a potential future if companies are just staying private? Yeah, I think we're just now in the early innings of this, but we are 100% starting to see it. We certainly have not done like a continuation vehicle is kind of like the most common term for this, like a CV where you can sell essentially like a slice of your fund. Because commonly venture funds have like a 10-year life

and then sometimes you can have an option for like two years of extension. And then after that you're kind of supposed to have the capital returned and like everyone can move on with their lives, hopefully, hopefully much richer for it. And, and, you know, SpaceX has been a private company for, what, 20 years now, you know, over 20 years. And so I think the most common vehicle for this would be a CV where you sell either like a bundle of company, like a slice of your holdings or like a slice of your fund. And you're starting to see that. Like several funds have done that. I think I think NEA has done it in the last year or two and you're seeing more funds do it. We certainly haven't done that before, but I think as an asset class, we're going to have to start coming to grips with the fact that the liquidity

timeline for these amazing companies is very different than it used to be. And it's actually really flipped on its head because it's almost like the better of the company, the longer they wait to IPO, because they can be private, you know, again, compounding their value at 30 plus percent IRAs for much, much longer than they used to. And so obviously you can sell in these rounds. Like, I think like obviously SpaceX is extremely. liquid, stripe is pretty liquid, data bricks I think is very liquid. There's some names that are very liquid, but it's only the best names where that exist. And so if you have a portfolio of names that are good, but not like SpaceX level, call it, then your best bet is probably one of

these CVs, which is very, very common in private equity. But as venture kind of becomes more institutionalized, we're going to see a lot more of them for sure. Speaking of DPI, it seems like you got some mathematical formula behind you. Can you break that down for us? Are we leaking alpha right Now what's going on? Just copy paste that and I get your returns? Yeah, I was trying to figure out the equation for DPI. I've been searching for a long time. That's the first one.

The second one, I was also trying to find the equation for Delian's gross margins on his investments. Oh, yeah. That one actually might be, that one might be harder to find than the equation for DPI. Yes, a little Easter egg for the thing. That's fantastic. How do you and the team, uh, does that? side if something is is in the wheelhouse or not. Obviously 10 people on the investment team. I'm sure you get pitches all the time where you're excited about the entrepreneur and and kind of their vision,

but maybe feel like it's not right in in the wheelhouse. But, you know, clearly, you know, you're doing flock safety to these other sort of like app layer companies, clearly willing to kind of, you know, look everywhere for opportunity. Yeah, 100%. Yeah. I love to say that there's, you know, A lot of people talk about Conway's Law, which is like you ship your org chart and as it relates to kind of like how companies mature and ship product over time. And I love to say that like Conway's Law exists for venture firms for sure, which means that you like you invest your org chart. And it's like very hard to, you know, have an investment strategy or a portfolio construction

strategy that doesn't match the size and kind of scope of your team. So for our team, seven partners, 10 total and the. investment team, what that means is that like if we want to be company builders, if we want to be involved in our companies, we can't do that many investments per investment professional. Like we need to we need to really stay concentrated and stay tight. I think you also see this at my you know, my previous employer founders fund. You see these these headlines where they're doing massive, massive checks into really high quality companies.

And I think it's because it's also a small really lean team that has the trust in a high degree of relationship with really strong founders. And I think we approach, especially growth investing with the same model, each of our growth funds is typically only 10 to 12 core investments. And so every single investment that we're looking at needs to have something very unique or special about it where we can look around the table and say, this is going to be a company that we're talking about on TBPN in 10 years as one of the top five companies in the world. Like it has that potential every single time. Can you take us through a little bit of a retrospective on playing different games now that it's been, I guess, four years in two days. April 12th, 2021, you dropped playing different games or why Tiger is eating your lunch and your deals. This cycle, it feels like we haven't, no crossover funds have really made a name for themselves.

But at the same time, Databricks is doing bigger deals than ever. There must be new pools of capital coming in. what does the late late stage look like and how is it changing yeah so it's funny i actually dropped a sequel to playing different games like very quietly i initially wrote it in 2022 and it was one of those things where like i think it's like solid but it's just nowhere near the banger that playing different games was and so i had a lot of like sequel anxiety about it and so i just like kind of like quietly put it out there because i'm like at my thought should probably be out there but it's called game over question mark and you can go read it on my substack and i think it's held up pretty

well since I wrote it in 2022. And I think the like overarching message from that. And I use a Game of Thrones analogy in playing different games. So I'll use one for this too. There's like, you know, there's a part in Game of Thrones where, you know, spoiler alert, if anyone hasn't watched the show where like Rousse Bolton for a while is this like Lord who's kind of like one of the big bads. He's one of the main bad guys.

And he gets killed by Ramsey Bolton, his son. And then you realize like, oh my God, this guy is like so much worse. And it's like, he has like. The scope of his destruction is like so much greater than like this guy was actually kind of a pedestrian bad guy. And I think the kind of version of that that happened to our asset class is that obviously people saw Tiger kind of fail the marshmallow test of like they went too far too quickly and kind of stretch the bounds of what you could do. And definitely paid the price for it. But instead of Tiger, you know, continuing to be this large platform, you now have like four other platforms that were actually established in those days of 2021. And so it's like, okay, yeah, like you don't have Tiger, but like, how big is Indrisen's last fund and how big is General Countess last fund and how big is light speed's last fund?

There's like four or five of these folks that have like filled that gap and continued what I think is like a secular trend towards platformization in the asset class. I think that the to your specific point around like our crossovers kind of like coming back into venture and playing in the asset class again. I think the most important thing here is that like venture means at least like four or five. different sub-asset classes, right? Like venture includes growth, but then, like, investing in a, you know, entropic at $62 billion and investing in a $100 million

A-R-R company that's growing, like, 20% at, like, eight times A-R. Both of those are called growth investing, but they're, like, as different as you could get in every single way. So, like, our crossover is doing venture again, I think, yes, but they're sticking more to their knitting, like, is like,

well, what about... What about... ...doing doing Series B's, like, no, but they're, you know, several crossovers that have invested in like majorly in the big labs for example. Was there any investments that Tiger did that you wish you did? There has to be some right. Because like I remember FTX was like spraying money like crazy too, but then they invested in Anthropic and like they kind of made it all back in one trade.

Right. So do you and Tiger hasn't I don't have a full list of of their investments or anything like that. But do you think it's possible that in the fullness of time, they just did okay or did they actually light money on fire you know it's funny i mean this is probably just a rumor um so this is you know assume this is total hearsay but i had heard that there was like maybe an internal like you know justice for john courteous trend at tiger where like you know like he did push for them to to get into data bricks and like that's gone really well and it's like a truly generational company and like you know it's probably going to be an amazing investment for them and like And so I don't know their full investment track record and returns, but I do think there is like a potential where like, I don't know if Databricks is a $500 billion company, you end up with like a pretty good portfolio, even though you had a bunch of strikeouts.

Like I think if you look at someone like Neil Mehta's track record, it's like is his hit rate like 90%. No, there's like a ton of flameouts in there. But he's just invested in situations where he's put, you know, 400 in and gotten $5 billion out. And it turns out like if you just do that even at the growth stage, like the power loss still exists and you still. and you still have like an amazing track record. So, I mean, they're huge investors in Waymo. This is Tiger. Would love to be in Waymo, I think. I mean, it's like an unbelievable technology company. Obviously, they're huge investors in Database.

I'm sure there's lots of investments like that for them. Yeah. That's fascinating. Well, this was a really great conversation. We'll have to have you back really, really soon because I'm sure we could talk for two hours. Yeah. Yeah. Yeah.

Anytime, guys. Thanks so much. Thanks for the alpha. Talk soon. Bye. Cheers. See it. See it.

The little jab back at Daliant It's perfect. It's great. Yeah. We're not going full like Jerry Springer yet, but we like a little bit of drama on the stream. Yeah. When the SaaS companies are spying in each other, we're covering it. When the Wall Street Journal's putting a robotics company in the Truth Zone, we're covering it.

That's right. And coming up next is just a great story about Zipline. We got Keller coming in the studio to break down the news that Zipline, the drone delivery company that has been around for a long, time, although not operating on America Shores, went and found a less regulated environment, delivering life-critical medical supplies and blood, I believe, scaled their business, did all the R&D risk in a low regulatory risk environment and is now ready to come back and has some really amazing partnerships. They are back.

We're excited to invite Keller onto the show and give us a breakdown on what Zipline is up to these days. Keller, welcome to the studio. Hey, thanks for having me. Thanks so much for joining. I have to say first, fantastic launch video. There are so many launch videos these days that I mean, I'm, you know, suspect number one in propagating this type of content that's, oh, let's pull top gun footage and, you know, like, let's put free bird over it and we're hard tech and it's us welding. And there's a lot of that and it's really cool, but it was getting a little played out. And you went a very different direction with this announcement. And so I just really enjoyed the launch video. But can you take

us through exactly what you're launching, what this means for the company and where the company stands today? Sure. Yeah, day before yesterday, we officially launched our next generation service in Dallas. So as you mentioned, Zipline has been operating these autonomous delivery services across eight countries over the last 11 years. But really, this is the first time we're seeing like major metros start to launch and scale in the U.S. Our first customer is Walmart. We're already announced a number of other partners like Chipotle and Sweet Green Mendocino Farms and a lot of the biggest health systems in the U.S. who are all relying on this technology to automate and accelerate their deliveries from businesses or hospitals or warehouses directly

to homes. Do you feel like this opportunity is underhyped today? It felt like there was a when drone delivery, you know, maybe it was, I'm sure it was when you started the company, you believed in the vision, but then it actually took years and years and years to like get to the point where we are today, which is like it's a reality. You're partnered with Walmart. I'm going to be able to drop a burrito on John's backyard if I want. I mean, you guys started in 2014, right? So complete overnight success. Yeah, yeah. We love an overnight success on this show. So congratulations. You're going to be a household name soon.

Everyone's going to be like, oh, yeah, you just did it so quickly. Maybe I should get into that market. Maybe I should have. Hopefully it makes sense, but I think there's this trend in technology where people get excited about the potential of an opportunity. And then the reality sets in that it's really, really hard to do. And then there's this period where, you know, it gets less attention. It actually is good for you because there's less people going into it. And then now I can just imagine a future where we're just seeing zip lines everywhere in the sky.

So yeah, I mean, talk about your maybe excitement today. I'm sure it's more than more than ever. Yeah, I mean, you know, I agree and there's so much to say on that front. But I think, you know, there is this obvious hype cycle curve of, you know, extraordinary, you know, excitement. This is coming tomorrow. I mean, when we started, you know, the CEO of one of the largest tech companies in the world was on 60 minutes, promising that they were going to be doing drone delivery to every home in the U.S. within two years.

Yeah. And so we always assumed we would be a. fast follower to them. We thought they were going to lead and we would be a fast follower. Had you told me that 10 years later, Zipline would be crossing 100 million commercial autonomous miles would be the largest autonomous system on earth of any kind. And, you know, that big technology company would be less than one one 10,000th of the scale. I mean, would have seemed impossible. I think that the main takeaway for me is that, you know, that 10 years, because then you have the hype and then you have the trough of disillusionment,

I believe it is what it's called. And I think for especially for hard, tech or hardware companies, that trough of disillusionment might be like five to eight years. And so for us, it was really important to walk rather than talk. It's really easy to talk. And maybe that's, you know, apropos of what you were talking about on the launch video side, John. But, you know, yeah, if you just focus on walking and focus on these small use cases, I mean, you know, we started by delivering just blood to 21 different hospitals. It was such a very narrow, clear thing that we needed to do. And, you know, we started. And, you know, we started, And we're also willing to do very unfancy things, like operate in countries that are hard to get to and get our hands dirty and figure out really hard, unfancy problems like how do you operate in all kinds of gnarly weather, you know, rain, thunderstorms, snow and icing conditions. These are hard problems. It's probably a bit similar to autonomous cars in that way, too, you know, that there's a lot of hype and then the reality sets in of what it's going to take to make these systems work at scale.

Can you talk a little bit about the regulatory environment? It seems like you probably could. have just gone straight to the American market because of the regulations, but you found a way anyway, and I love that. Is that actually a bull case for the regulatory regime in America working as intended? And we should have companies that can go test things elsewhere and then bring it back when it's mature and the technology is ready, or are there specific regulatory changes that you'd like to see over the next decade to either spur more innovation or just make what you do easier?

Yeah, it's a good question. When we were launching, we really had this sense. There was such an unclear regulatory environment with regard to what we were proposing. We thought someone would build an automated logistic system for Earth. That seemed like a really important idea. But it was so unclear from a regulatory perspective that I think we really just concluded, we have got to go do the most obvious life-saving thing imaginable, because that would maximize the chances that we could.

you know, I mean, it was both powerful because it was an amazing mission and because it would maximize the chances that we could innovate and get started and operate in the real world. And so we needed a public health care system to do that. We didn't want to work with a whole bunch of, you know, health or skeletor health systems, which you can have in the U.S. We've got to be good if we could have a public health care system. And so that immediately caused us to go to certain parts of the world. And then I think also there's just a lot of, you know, people think, oh, the regulatory environment must be so different in Africa than in the U.S. actually not true. Pretty much the same. regulatory regime when it comes to airspace because planes fly back and forth between them. So you need the same rules. I think the difference was finding a government that behaves more like a startup, a small, agile government that is willing to make decisions and exhibit more like, yeah, just executive decision-making. So those are really the things that made Rwanda such a powerful place for us to get started. And it was about six years later, you know, at that point we had about 50 million commercial autonomous miles and zero human safety incidents.

That's the moment when we thought, hey, this is a good time to bring this massive data set to the FAA and kind of help them see like this technology is ready for prime time. It's safe. Once we did that, it then took about five years from then until, you know, where we are today, where, you know, Zipline is the only company in U.S. history that's been awarded like full permission to fly beyond vigilantes site in all 50 states. So I think part of it was building that data set, showing that it can be safe. and then part of it was working directly with the regulator in the U.S. to help make sure that the U.S.

doesn't fall too far behind in this core area of new technology. Yeah. The immediate benefit, you know, as your network rolls out, I love the idea of just being able to press a button and get, you know, an item, you know, in minutes. You know, it's obviously been a dream, I think, for so many people for so long. Can you talk about some of the second order effects that you're anticipating at? we're here in LA, I can imagine like a lot of traffic on the roads is just like items being delivered and it just so happens that it's a human in a car right now. And so with Zipline,

you know, when Zipline rolls out here, I can imagine that there would just immediately be less congestion. But that's just me, you know, making a bold, you know, potential prediction. But I'm curious how you think about it. And the cool thing is we're already seeing this. I think people don't necessarily appreciate how massively instant delivery has scaled. in the last five years. And it was obviously accelerated by COVID, but we're gonna do five and a half billion instant deliveries in the US this year alone.

And we're using three to four thousand pound, three to four thousand pound gas combustion vehicle driven by a human to deliver something to your home that weighs five pounds. So you know, you don't have to be a physicist to realize like this is actually a bizarre solution. You know, we're essentially we're using technology it's 100 years old to solve a problem that's like five years old and to serve a market that's

growing super fast. So we think it's super obvious that if you want to deliver something fast that weighs five pounds, you probably want to do that with a vehicle that weighs 50 pounds and you want the vehicle to be electric and autonomous. As soon as you realize that, then I think, you know, the future, you know, you know a secret about the future that most people don't know. The other thing is like I personally, if I'm ordering delivery, it's like, do I really want this person to have to go to CVS and get Advil? Like technically that it's like, you know, a form of employment, they're opting into it, but like, it's like, oh, I'll just do it myself. Like, it's been of a hassle. But when it, when it's fully autonomous and it's literally like

an extension, you know, of the service that I'm using to get the item, it's, you know, you'll use it a lot more. And actually just, just on that front, you know, the customer behaviors we're seeing just in the last year in the U.S. are quite mind-blowing. I mean, first of all, you know, I may have thought, oh, maybe it'll be these like tech adopters and some of these, some of these, you know, I don't know, you know, nerdy or, no, no, it's like moms and grandmas who are like our power users. And, you know, more important than that, I mean, you talk to, you talk to them and we have, we have users who have ordered 300 times in the last 12 months. So this is like a part of their daily, it's way different. It's like, you know, I was, I was talking to a woman who said, you know, she's 78 years old.

She orders, you know, she goes to the grocery store once a week and then she orders about three to four to. She orders from Zipline three to four times a week. And, you know, as I was, you know, as I was talking to her, she's like, yeah, you don't get it. I mean, this saves me like five hours a week. It's totally priceless. There's no way I go back to the old way of doing things. When the weather is bad, you know, I don't want to risk my health or I, like, fall down and get hurt. And she has a lot of friends who might, you know, either single moms who it's not as easy to get out of the house or, you know, older people who it's not that easy to get to the store.

I mean, I think people underestimate, yeah, if you make it super fast and convenient, this is something people actually use. every day, not every week or every month. Yeah, even just thinking parents, right? I think every parents had the experience. It's bedtime and like you realize there's no diapers, whatever. And so then it's like, well, are we going to like get the kids and go to CVS or whatever to like get diapers and they don't have the brand? So it's like I just think of, I can just think of so many use cases where it's just like,

oh yeah, we're all, you know, we're going to delay bedtime by 15 minutes. On the average. And Jordi, as someone who has a, you know, one year old and three year old at home, like, these kinds of systems operate 24-7. Yeah. Which is also pretty game-changing relative to the way we currently think of logistics. It's a really big deal to have something that is always available whenever your kid happens to wake up or isn't feeling well. Yeah.

On the topic of the evolution of the actual technology that you're building, I always thought the name was interesting because obviously it's a metaphor for a zipline that just goes from one place to another. But then you were catching the drones with something that looked kind of like a zipline. and now you have the mothership drone dropping a smaller drone with something that looks like a zip line. How, I guess when in the evolution of the company did you come up with the mothership and baby ship? I don't know what you call it. The delivery zip is what we call it. What do you call it? We call it a delivery zip.

Delivery zip. Okay. Yeah. So how did you come to the pairing system and why is that important and what the evolution of the technology you've been? Yeah, I mean, I think it's a really good question. So, first of all, you know, the plane spent 10 years operating these kind of more long range systems. Platform 1, you can actually see some of them sitting right here, waiting to out and begin making deliveries. Fixed wing, right? Yeah, it's a plane. And so this aircraft can fly, you know, 300 kilometers

on a single battery charge. So it's all about range and serving like very rural hospitals and health facilities. For Platform 2, you know, it's becoming obvious to us that home delivery is, you know, It's by far like the Holy Grail. I mean, that's what automated logistics really has to solve globally. And the most important thing is you want to be able to deliver quietly, like silently. You want to deliver silently. It needs to be extremely safe. You need to be able to deliver gently and with dinner plate level accuracy.

It's kind of how we describe it. That's actually super hard to do. And, you know, you see a lot of other companies talking about. You don't have to, you don't have to, you know, stress that sounds very hard to do. If you, well, if you, I mean, if you see the way other companies are talking about solving this problem, a lot of times they're talking about like descending, you know, an octocopter of death. I mean, it looks like a lawnmower, you know, within 10 feet of your home. It's incredibly loud. It's incredibly disruptive to your neighbors. It's not really a part of like, we think, you know, new technology, it needs to be part of a beautiful, serene world that we would like be proud to hand to our kids. And so, you know, the big advantage of designing the system in this way is that the, you know, the, you know, the, you know, the, you know, the, um, the, the aircraft is staying 100 meters in the air. The aircraft is, first of all, designed to be extremely quiet. And then on top of that, it is far safer

because it is staying far away from you, your family, your pets. Only thing that's coming close to your home is the delivery zip, which you can see behind me. This thing is, you know, inspired by Eve from Wally. I was going to ask, was there specific sci-fi that you and the team just love that you kind of reference? I think we're really into Solar Punk. Have you guys heard of Solar Punk?

Oh, yeah. Yeah, of course. Yeah, I think that really to me is nobody talks about that. And, you know, so much of, I mean, a huge sci-fi nerd and obviously so much of it is kind of apocalyptic and the robots are out to kill you. And we're like, well, what if in the future robots are trying to save your life? You know, that's, that's a different take. And I think I completely came across in the video. The video was very saturated, very orange, very, like sunny day. And it was just so pastoral and lovely. I really like that as like a different vision of the future. It's cool. And to, and to Jordy's point, you know, know, I think that these kinds of systems, people are often like, oh, it's going to be so loud, and it's, you know, the sky is going to be darkened with drones. It's like, it's kind of funny. It's the same way people actually originally felt about cars, if you go and like read the, you know, if you read the newspaper articles from like 1910, people thought, you know, cars were going to be the scourge of cities. But I think the reality is actually, this is going to take a lot of delivery vehicles off of the roads. This is going to reduce traffic in our neighborhoods. This will reduce pollution in our neighborhoods,

improve air quality, they'll reduce noise because these systems are ultimately much quieter than cars. I think there are a lot of pretty exciting advantages when you just, yeah, I think, I think the world 10 or 15 years from now can actually be far more beautiful. Like we can hand space back to humans. Like your kids could be playing hockey in the street again, which I think today you just don't do anymore. You don't want, you know, between all the cars zip it. Yeah. Could you talk about any thoughts, you know, ideas? around larger payloads. Like, you know, we've seen like airship,

uh, airship startups emerge that are, you know, basically building like big blimps that could deliver, you know, something closer to kind of a competitor to the cargo ship. Yeah, the cargo ship or some of these bigger, you know. But even like the flying car startups that are trying to ferry individuals around,

I'm sure you're like loosely aware of the technology there. Yeah. And in many ways like, you know, doing these like ultra small drones and with super precise delivery, it feels a lot harder than just, just like lifting up, you know, basically an airship bus and going from point A to point B. But is that at all? I'm sure you guys have thought about it. Any immediate thoughts?

You know, the market that Zipline is focused on instant delivery is one of the biggest markets on earth. And it's just such a huge problem. And keep in mind, really today, it's only available to rich people in the U.S. You know, it's way too expensive. Like even just in the U.S., it's not universally available, not to mention the 7.5 billion people who don't live in the U.S. globally. So we actually think that as you automate, as you decrease the price, as you expand access and make it universal, I mean, we're going to continue to see 10 or even 50 X growth in instant delivery globally. So whereas 5.5 billion deliveries might seem like a lot in the U.S., we actually think there's probably demand for 50 billion deliveries, instant deliveries, in the U.S. So I guess, long story short, we've got our work cut out for us, just focused on logistics.

I think, you know, we deliver, this system delivers an eight-pound payload. It's designed to deliver an eight-pound payload. Eight pounds is actually a lot. You know, that's like a big grocery bag. It's dinner for, you know, four to eight people, depending on what you're ordering. And I think ultimately these kinds of systems, if you can get a system that can do, say, a billion, you know, instant deliveries of these kinds of packages, it is actually likely that you would be able to then scale that autonomy up, autonomy stack up relatively easy to carry humans, for example.

I mean, I think carrying humans is a far harder problem. And it'll be interesting to see, you know, where that plays out over the next 10 or 15 years. Well, last question, then we'll let you go if you have time. I would love to know about, obviously, everyone's thinking about tariffs and whatnot, but how are you thinking about scaling up manufacturing? It seems like you, you know, you have product market fit. You've solved the regulatory risk, the technical. risk. What's next in terms of scaling up manufacturing? Are you building like a gigafactory for

these things at some point? And is that just like an entirely new challenge that you foresee on the future? You know, Zipline has always manufactured everything in the U.S. And so one of, you know, one of the advantages, you know, obviously there's this huge competition playing out between, you know, it's kind of great power competition between the U.S. and China. And I think that a lot of that competition is going to revolve around key areas of technology and everybody's talking about drones. I think the current perception is that China dominates drone manufacturing. And that, you know, that is true for like DGI quadcopters that, you know, are plastic and take pictures. But the good news is there are, you know, there are a few companies that are completely focused on manufacturing

in the U.S. There are U.S. companies driving different classes of vehicles. You know, Anderl would be another good example, NEROS. And ultimately, Zipline isn't even, it's not even because it's more efficient from a global supply chain perspective to manufacture in the U.S., which in our case it is, but actually the most important consideration is if you're a truly innovative company, you want manufacturing and engineering to be right next to each other. Like Zipline, so, you know, I'm actually downstairs at our headquarters. This is now kind of like an engineering prototyping space. About a third of a mile away, we have a very large factory, and we're scaling that to build about 55,000 aircraft a year in their first full year of production. You guys should come visit if you want. But I think,

you know, that, and by the way, that's in South San Francisco. So I think people are often surprised that you can, like, achieve that level of scale in California. You're doing the reindustrialized meme. Yes, exactly. But it's just all about, it's all about engineering and manufacturing together, like the ability for engineers to go get hands on their own parts to start to understand, but how does that part actually get built? What does quality look like? And to be able to rapidly change parts of the assembly process, if necessary, to achieve reliability, safety, and cost, these things are so much easier to do if you do them all in the same place. So I have so many more questions. We'll let you go, but I have one more. Are there any,

like, huge developments in either, like, open source technologies, you know, see all this development with AI that's getting better and better, or even just like partners. For a while, there were companies that were thinking about drones as we'll build the operating system for drones and will vend into a company like Zipline. Have there been any other key partners or technologies that have really, you've kind of built on the shoulders of giants, so to speak, or has it really just been like you've got to build everything from scratch? Yeah, it's interesting. I hoped that what you just said would happen. And in fact, when we see, started, we were using a lot of off-the-shelf components.

You know, we used an off-the-shelf IMU, off-the-shelf GPS system. We were using something called RTK differential GPS. We were using, even for the first few months, we were using off-the-shelf autopilot. Like, all of these systems kind of failed. And I would say it's actually very similar. You know, when Tesla got started, they were like, oh, we're going to use an off-the-shelf Lotus Elise chassis chassis and an off-the-shelf battery pack that we're buying. We just combine them together.

we're going to have a good product. Turns out, like, they were wrong. To actually build a great electric vehicle, they had to design the battery pack from scratch. They had designed the entire car around the battery. Zipline found itself in a very similar position where no one is building electric aircraft at commercial scale. Zipline is the only company that is operating a full fleet of commercial aircraft at this scale.

And so it means we had to design the battery completely from scratch. We had to design a flight computer completely from scratch. We had to design the motor completely from scratch because, you know, ultimately what all of our customers care about is just teleportation. They just want something to go from point A to point B fast enough to save a human life. And that means we have to make it safe, reliable, cost effective. And the way to ultimately do that has been to design every component around the use case, really. And so that's, well, it's amazing too that you started in this like ultra high stakes use case, which is blood delivery where if you mess up, there's like, you know,

massive, you know, the biggest consequences. And then now, everywhere. else every other American can get their burrito or their cheeseburger, you know, reliably with the same reliability as, you know. Yeah, no, it's, it's fantastic. Well, I, we'd love to have you back on soon. Yeah, this is fantastic. It's so fun to watch the strategy play out. I'm just genuinely so excited.

I'm so excited to be a DAU, a Zipline DAU. I think you're going to be more than DAU three times a day. Yeah, he's going to be doing a thousand deliveries a day. As soon as you get hourly average user, H-A-U. Yeah. hourly. Yeah, as soon as you get heroin on boarded, it's game over. You're going to be profitable. No, I just don't think, I don't think this is priced in yet, to be honest. I think it's going to be like imagine people, you know, shopping like, you know, the whole like drunk shopping meme. It's not just like instant delivery on, you know, you want food or groceries. It's like, imagine being able to like see an item on an Instagram, you know, shopping or whatever, get an ad and get it like 30 minutes later. That's going to change. It's going to. It's going to. It's going to. It's going to. change everything this is so awesome i'm really excited for you so congratulations on the overnight

success thank you for your service yeah thanks for 11 years of hard work you make it look easy but yeah we really appreciate you come out on the show and sharing all that with us this is very fascinating thank you and seriously congrats thank you guys really appreciate it we'll talk to you soon talk soon bye i would amazing almost invest in that company at any price yeah who is in this company because they've just been grinding for so long. And it must be super capital intensive. So there's probably pretty deep bench. Not a lot of very well-known funds.

Sequoia. Okay. Google Ventures. Yeah. That happens sometimes. Another $1. Another $10 trillion to the Holy Trinity. What do you now?

Of course, of course. Anyway, that's great. Yeah, Sequoia got in the Series A. Let's see it. You love to see it. Anyway, let's move on. Let's do some timeline and then get out of here. We started late, so we're ending late.

You already heard from Quaid at Bezell, but I just wanted to let you know that at getbezzle. Your Bezell concierge is available to source any watch on the planet. This is an interesting fact that people might not know about Bezal. So they have auctions, they have prices, like Buy It Now prices. But also you can chat with a real person who works for Bezel and say, I want this exact watch. Maybe I saw it in a movie. Maybe I'm hearing rumblings about it at watches and wonders. You can tell them this is the one you want, and they will actually go out and source it

for you and bring it to you, which is great. Anyway, let's move on to the OpenAI announcement. Starting today, memory in chat, GPT can now reference all of your past chats to provide more personalized responses, drawing on your preferences and interest to make it even more helpful for writing, getting advice, learning, and beyond. Seems like a nightmare because I've been lying to chat GPT for about three years now about my expertise. Well, every time, because of the prompt engineering, I'll always say, like, you know, I'm interested in trains, but treat me like a train expert. I work in trains.

I actively own trains. So I, because I want to, I want to prompt engineer it to give me like the best data. And so now when I talk to it, it's going to be like, as a train conductor, you probably want to go with this train. I'm like, I was lying. I wasn't actually a train conductor. Explain this like I'm five. Yeah, it's like, oh, you're five. Oh, it turns out I didn't know you were five. John Coogan, the. John Coogan, the. John Coogan, the. five-year-old? The five-year-old. Oh, yes.

As a five-year-old. So now I have to go back into Chad GPT and tell it. Forget that I'm five. Forget that I'm a five-year-old industrialist who owns trains. And let me tell you about my real history and get you up to speed on the real memories that I want you to say. But obviously, this is a very cool product release. It makes a lot of sense.

They go on in this thread. So in addition to the save memories that were there before, it can now reference your past chats to deliver responses that feel noticeably more relevant and useful. This happens a lot because you have so many different chats going. You want to reference another one and you have to go in the search bar, copy, paste. Obviously, that's a product feature. And this is a testament to Open AI expanding from nonprofit research foundation lab into product consumer tech company. And this is the thing.

People have been saying, oh, the models have no moats, et cetera. This is one of the modes that will come up. And I'm sure other foundation models will build this and copy this. But the name of the game is staying just a little bit ahead forever. and that's what Google did and no one ever pivoted to Bing because Google search is always just a little bit ahead, right? Well, if you want to stay ahead, you should get an eight sleep. Yes, you should. And let's do a quick, let's do a quick sport check here.

I think I did pretty well. I was going to cover eight sleep during the F1 breakdown, but we'll have to do that tomorrow because Charleclair is an eight sleep ambassador and we love F1. I got a 94, little low on the time slept 96. You have an uncanny ability to beat me by like two points. How much did you sleep? I slept six hours and 49 minutes. Brutal. Brutal?

I mean, I slept eight hours and 11 minutes. You put up big numbers, man. But I get in bed at like seven. Yeah, you do, you do. I got to get to bed earlier. I got to get to bed. It's so the key to sleeping well is like you have to go to bed when you don't feel like going to bed. And then you actually get your eight hours.

Yep. And you can go to eightsleep. dot com slash tbpn yeah uh anyway uh there was an interesting interaction between uh john carmac uh quote tweeting someone named devon uh about SpaceX uh devon says i'm going to call it right now a lot of stuff is going to break on this mission talking about a new SpaceX mission it's by design it's as part of the plan don't get upset i'm not saying SpaceX plans to fail i'm pointing out that SpaceX has taken an ultra important principle from software engineering and realized it applies to all engineering feedback beats planning.

That's a good lesson. And that you see is why SpaceX doesn't do things the NASA way. The NASA way was to gold plate everything, plan and test and plan and test and generate mountains of paper detailing every contingency with every scenario plan. SpaceX just shrugs, says it's unmanned and sends it. Half the time it blows up. That's the whole point. They don't actually want it to blow up, of course, but they're anticipating that it might.

That possibility is part of the plan because one rocket blowing up or crashing is an actual end-to-end test, this beats many, many man years of planning and plotting. The key realization here is that knowledge only comes from empirical observation. Everything else is just speculative. The sooner you get into that feedback loop, the faster you run it. The more iterations you can do in less time. This means while others are planning and speculating, you actually learn something. Relevant data is the most precious thing in the universe and it's worth blowing up any number of rockets to get it because rockets are just stuff. They're just made of stuff and you can always get more stuff. You can never get more time. It's a great insight. It just means we're doing it cowboy style. And I, the post is good.

So many good lines here. I got to follow it. But John Carmack endorses it and says, I have never seen it expressed exactly like that, but I wholeheartedly endorse it. Feedback beats planning. My plea at meta was no grand plans. Follow the gradient of user value. And I was chatting with John Carmack years ago on, on X about this, talking about how just getting the VR cost curve down just every single headset just slightly lighter, slightly cheaper, slightly better. Like, that's what I wanted. I was pitching.

Like, I pulled up this video of the N64 and you turned on the N64 with put the golden eye cartridge in. Turn on the N64. This is like way before your time. But you turn it on and it would just blink on and you'd be ready to play. And it was, it was crazy. There was no login screen, no off, no intro, no credits. It was just turn it on and you're just playing.

It was amazing. And I was like, they need to get there for VR. He, you know, agreed and gave some other feedback that was really interesting. But interesting speed of execution stuff. Anyway, speaking of VR, if you're trying to sell VR headsets all over the world, maybe your big screen VR, you've got to get on Numeril sales tax on autopilot. Spend less than five minutes per month on sales tax compliance. It's that easy.

Yeah. Thousands of companies rely on Numeril. Yep. You can go to Numeril HQ to get onboard. they'll do a white glove onboarding. Oh, yeah. They are just absolutely fantastic over there. 25 states are now taxing software sales, John.

I didn't know. You know that? No. Yeah, actually, you did. I did. I said it yesterday. Well, Vitorio says, Thank you to Numerol for supporting the show.

Vittoria says, it's so over. They automated Italians. I saw you put this in there. I like this. You put it in the show notes. I like it. Did you see the hand movement, which is crazy?

It's the Italian movement. Italian. He's Italian. He's having fun. But very interesting art installation, good viral video. Lots of fun. I like this one from Framer. Thanks to AI, every meme can be turned into a cartoon easily.

It's so funny because you remember we joked about this early on. It was, there needs to be like, you know, South Park meets Silicon Valley where something happens on the timeline. It just immediately gets turned into a cartoon. Yep. And we kind of looked at that. And we were like, eh, it's like funny to do. do.

It seems really expensive. So we wouldn't do it. It wouldn't make sense. It's just like at the end of the day, it might just be a viral video. It's not really. And we have a friend who's already using voice AI to make these like incredible deep fakes. Incredible deep fakes. And they're very funny because they're not funny because of the AI.

They're funny because he puts so much thought into the punchline and how it leads into the punchline and he's so creative with it. And it's really his life's work. And it completely copies the like the accuracy of the voice. It's crazy. It's really, really good. And so this Framer thread is interesting because not only do they share this video of the Paris Olympics, that crazy breakdancing video, but Framer actually breaks down exactly how to do it. So you can kind of follow along with all the different prompts. You go into chat GPT, turn the images into cartoons, and then obviously change those into videos. But it was remarkable. It's a very watchable video. And I think this would be part of like the meme stack going. forward when an iconic event happens, it will be instantly turned into a cartoon. You'll be able to enjoy it as anime if you want. But the real creativity will come from that human insight of what would be particularly funny to do in a cartoon setting.

It can be very funny. Anyway, let's tell you about public investing for those who take it seriously, multi-asset investing, industry leading yields, trusted by millions. Go to public.com to get started. And thank you to public. They are the ones that power our ticker down at the bottom. stock market goes up, that ticker would be down immediately. It's all driven by what the host of the All-In podcast are tweeting, apparently.

Every time they tweet, it seems to go poorly. Somebody takes Sacks his phone away from him. But it's going up, it's going down today is not as bad as other days. We had a pretty big dip in the middle of the day, but we're doing okay now. Yeah, I mean, we almost hit the circuit breakers today, but we didn't. I guess that's a win. Any day where there's not circuit breakers is, is a Dom Parayande in my opinion.

It's great. Should we talk about the Open AI lawsuit? This is kind of interesting. This is developing. So Elon has sued OpenAI, I believe. And there's been kind of back and forth about Elon being a co-founder of Open AI, putting a lot of money into the nonprofit via donations, and then not getting equity in the for-profit when that conversion happened.

And the investments started rolling in. Now Open AI is suing. Elon or counter suing. And Open AI Newsroom writes, Elon's nonstop actions against us are just bad faith tactics to slow down Open AI and seize control of the leading AI innovations for his personal benefit. Today, we counter sued to stop him.

He's been spreading false information about us. We're actually getting ready to build the best equipped nonprofit the world has ever seen. We're not converting it away. And so that's an interesting take, basically saying, like the nonprofit is not going away. That actually isn't new. Sam did address that in that.

It's new positioning. But it's new positioning and it's getting out there. Yeah. And so if you think about it. They've never been really leading with, we're actually making the best nonprofit ever since founding. Yeah, because the VCs are like, well, look, we don't care about the nonprofit that it's going to continue.

We only care about the for-profit. But obviously, if you're a nonprofit and you have some insane equity position in this like Bangor Consumer Tech Company, that's going to be very good for funding your nonprofit. And so what will the nonprofit? do probably continue to work on AI safety and AI research and all this all this stuff which you know I think can be very valuable and so they want to they want to make it loud and clear that the philanthropic efforts at the open AI nonprofit are not going away anytime soon and and you can expect

for them to keep fighting so you know hopefully that all just resolves my take has continually been mom and dad are fighting let's try and get them to sort it out and build a glorious future together with amazing chat chbt and grok functionality for all of us to enjoy yeah i feel like if grok and chat chpt actually sat down just the two chatbots they could resolve it with enough back and forth well chatbot arena yeah yeah yeah yeah two bots enter only one can leave no no no they should be able to work it out they should be able to like you know do enough iterations sure on the situation to like you know, run a million simulations. And, you know, that's actually kind of the scenario laid out in AI 2027.

Like, there's this whole nationalization push. And the three scenarios that are proposed are essentially, one, the government just says, we're nationalizing you. We're taking the complete control because we're afraid of, you know, runaway AI. The other one is like the open brain, the leading lab kind of like turns inward and kind of goes offshore and really fights it and doesn't, doesn't, you know, get nationalized. But they actually advocate for the third, which is kind of a truce. And a deal gets broken, gets brokered. And the government winds up working very closely with the bleeding lab, but the lab is not entirely

nationalized. And that's kind of what you're describing where, yeah, we will, will better AI models that can go and do crazy simulations and get to more reliable truces. That actually might be a great outcome. Instead of this dystopia where everyone's fighting, everyone's suing each other constantly, it's like, no, actually everything's just balanced out because we have a million PhD lawyers talking at all times to make sure we have the perfect deal struck at any moment. That's right. Who knows?

Anyway, Scale AI founder, Alex Wang proposes a national AI data reserve to bolster US data documents. I don't know what this means, but it sounds awesome, John. Let's push it forward. Amy to secure a competitive edge over China in the AI. race. Very interesting. We used to have gold reserves. Now we've got data reserves. Big data reserves.

Yeah, I wonder what that would look like. I mean, certainly like sequestering some of the data and protecting it seems extremely valuable since deep seek seemed like it was completely reverse engineered from chatypete. Esoteric PDFs with forbidden knowledge. Yes. I think Nat Friedman should be the one on the case. A national AI data reserve should be the scrolls cannot be ingested into the next

LLM training run. They must live in Nat Frieden's house forever. Anyway, there's some other news, but first, let's talk about Wander. Find your happy place. Find your happy place. Book of Wander with inspiring views. Hotel great amenities, dreamy beds, top tier cleaning, and 24-7 concierge service. It's a vacation home, but better folks.

It is just delightful. Yeah. Brad Gersner, who recently led the round that we discussed on the show today, says Google is holding the line on CAPEX, as I said tonight on CNBC, tech CEOs are ready to run. That's why today's tariff clarity was critical to maintaining our leadership and global AI, smart targeted tariffs, plus a trade deal with China, plus a tax deal will extend U.S. national advantages. And so this is on the back of Google CEO, Sundar Pichai, reiterating their $75 billion cap-x guide for 2025, despite the tariff uncertainty. We've been hearing that Microsoft, Microsoft, Microsoft,

cancel a certain, it was kind of a big headline because it was like, I think it was framed by that one poster. It's so over. Microsoft canceled a one billion dollar data center. This is so terrible. And it's like, well, they're still spending 74 billion then. You know, it's like on CapEx. Like the capacity is going to increase. Maybe they're slowing down a little bit, but it still seems like. And then also like there's just the dynamic between all the hyperscalers that none of them want to get caught, you know, flat footed here. And so, and there's also the, the, amazing story about Zuck being like, yeah, we did reels. We got caught back, we got caught, we got caught flat footed on reels. We didn't have enough, uh, AI data centers to train,

uh, Reels recommendation algorithms at scale for a billion Instagram users. Uh, so we built out a re, uh, a reels training AI data center. And then we were like, let's just do two of those. And, and it worked out perfectly because then they were able to train Lama. And, uh, it had LLN functionality over all over the place. And so I think every, a mag seven CEO has to be taking that seriously and thinking about well yeah maybe we don't want to get over our skis on capax but we got to be in the game and that means double digit high double digit billions of capax every single year we're scaling our capax here at the show as well we are scaling up well let's close with is there anything else you want to do or you good let's save this this next piece

for tomorrow i think so it's important and i think we should give it some real attention anyway fantastic show really great interviews. I had a great time today. And just thank you for listening and thanks for dealing with the cyber attack that happened on the show earlier at 11. We had to start late as we fought off. Just an absolute nightmare for Ben and the whole team. They were stressed, but they got through it and they're feeling good. And I'm sure tomorrow will be a banger episode. That's right.

Right at 11. On the dot, we're going live. You heard it first. We've been pretty good about going at 11. Yeah. We used to be all over the place. So the general trend line is more accuracy, more professionalism on this show. And so thank you. If you've enjoyed the show, leave us five stars on Apple Podcasts, Spotify. Fun fact about Spotify, you can't just go leave us five stars there.

You got to click, listen to the episode on Spotify. Let it play for a while. You can put it in the background low. And then it will let you leave five stars. Spotify knows. Don't spam us. But we're going to get through that. We're going to get through that.

Thank you to ramp, polymarket, public, bezel, numeral, ad quick, eight sleep, wander for sponsoring this show. We are entirely corporation supported. We will never forget that. Thank you. Thank you. Have a great day. All right, folks. Bye. Cheers.