This Week in Startups - TWiST News: Grape-picking robots, startup investing for all, and the earnings numbers you need to know | E2037

Episode Date: November 2, 2024

This Week in Startups is brought to you by: Fidelity Private Shares℠. Manage your cap table and data room, get faster, more accurate 409A valuation and fully automate your next financing round. Visit ⁠https://fidelityprivateshares.com⁠! Mention our podcast and receive 20% off your first-year paid subscription. LinkedIn Jobs. A business is only as strong as its people, and every hire matters. Go to https://www.linkedin.com/twist to post your first job for free. Terms and conditions apply. Oracle - Oracle Cloud Infrastructure, or OCI, is a single platform for your infrastructure, database, application development, and AI needs. Save up to 50% on your cloud bill at ⁠https://w⁠⁠⁠⁠ww.oracle.com/twist * Timestamps: (0:00) Jason and Alex kick off the show (4:49) Extend Robotics’ Dr. Chang Liu joins the show (6:23) The challenges of real-world automation, robotic grape picking, and human in the loop AI (10:44) Fidelity Private Shares℠ - Visit https://www.fidelityprivateshares.com Mention our podcast and receive 20% off your first-year paid subscription. (12:14) Teleoperation, network latency, and future of robotic harvesting (21:05) Building a tech company in the UK and second-order effects of AI (23:24) LinkedIn Jobs - Post your first job for free at https://www.linkedin.com/twist (24:35) AI models learning from best human practitioners (26:56) Global remote work impact and virtual assistants (30:38) Impact of global salary standards on US developers (33:02) Oracle - Save up to 50% on your cloud bill at https://www.oracle.com/twist (34:28) Accredited investor rules and potential changes (42:26) Sophisticated investor tests and angel investing (47:32) Investing in local businesses and AI news (48:36) OpenAI announcements and Google’s blowout quarter (52:22) Democratizing education with AI tutors and OpenAI's new search product (1:03:39) OpenAI’s advanced voice feature and earnings highlights (1:06:26) Tech giants' AI initiatives and implications (1:08:03) Amazon’s AWS growth and AI business expansion (1:14:39) Audience question: Interesting AI sectors for startups (1:21:05) The Chinese EV market * Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com Check out the TWIST500: https://www.twist500.com Check out the Launch CloudKitchens Incubator: https://ck.launch.co * Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp * Check out Extend Robotics: https://www.extendrobotics.com/ Here is the “Empowering Main Street in America Act”: https://www.banking.senate.gov/imo/media/doc/emsaa_section-by-section.pdf Check out SearchGPT: https://openai.com/index/searchgpt-prototype/ Check out the Xiaomi SU7 Ultra: https://www.mi.com/global/discover/article?id=3768 Check out Nio: https://www.nio.com * Follow Dr Chang Liu: X: https://x.com/enjoychang LinkedIn: https://www.linkedin.com/in/thechangliu/ 8 Follow Alex: X: https://x.com/alex LinkedIn: ⁠https://www.linkedin.com/in/alexwilhelm * Follow Jason: X: https://twitter.com/Jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis * Thank you to our partners: (10:44) Fidelity Private Shares℠ - Visit https://www.fidelityprivateshares.com Mention our podcast and receive 20% off your first-year paid subscription. (23:24) LinkedIn Jobs - Post your first job for free at https://www.linkedin.com/twist (33:02) Oracle - Save up to 50% on your cloud bill at https://www.oracle.com/twist * Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland * Check out Jason’s suite of newsletters: https://substack.com/@calacanis * Follow TWiST: Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin Instagram: https://www.instagram.com/thisweekinstartups TikTok: https://www.tiktok.com/@thisweekinstartups Substack: https://twistartups.substack.com * Subscribe to the Founder University Podcast: https://www.youtube.com/@founderuniversity1916

Essentially, Sam Alden said, hey, we face a lot of limitations and hard decisions about how we allocate our compute resources towards ideas. So essentially, Jason, they're compute constrained even with Azure in their pocket. That blew my mind. Yeah, they're blowing through five, 10 million, 10 billion a year in infrastructure and losing a lot of money. And that does say something that, you know, they're running out of bandwidth. And I saw Jensen Wong who was talking about Elon standing up this giant cluster. So the buildout keeps occurring. I do think the interface of chat GPT is superior to any other interface, but the results are not. This week in startups is brought to you by Fidelity Private Shares. Manage your cap table and data room, get faster, more accurate for

09A valuation, and fully automate your next financing round. Visit FidelityPrivate shares.com. mention our podcast and receive 20% off your first year paid subscription. LinkedIn jobs. A business is only as strong as its people and every hire matters. Go to LinkedIn.com slash twist to post your first job for free. Terms and conditions apply. And Oracle. Oracle Cloud Infrastructure or OCI is a single platform for your infrastructure, database, application development, and AI needs. Save up to 50% on your cloud bill at Oracle dot com slash twist. All right, everybody, welcome back to this week in startups.

It's Friday. It's November 1st. We are but four days from the election. And we're not going to talk about that today. We're going to talk about startups, technology. So this is a great reprieve for all of the that are white knuckling it to the end of the election. Or hopefully if you're a founder of a startup, you're focused on your team, your product,

and your customers with me again, today, as always. Alex Wilhelm, how are you, sir? You know, I'm good. I'm also very glad that this is our little election-free safe zone because I feel like everywhere else, I look if I open the door, election, turn on the sports game, election. So, you know, a little break. Sounds good. Yeah, I'm so good, man.

Who's going to win? Who's going to win? Let me ask you. I mean, it is crazy to think that the prediction markets are like coming back down to the, they were obviously very much in favor of Trump, but they seem to have come down to the polls and the pollsters, right? The gap is closing and we're in a dead heat. And I asked Nate Silver on Twitter,

like, when's the last time it was like this? Was it Al Gore and Bush? And I don't even think that 2000 election was this close. Yeah. Well, one thing I've been tracking very carefully, apart from Polly Market and Kalshi and other prediction markets, probably Robin Hood got in on that. Shout out to Robin Hood.

But I just pulled up the data. According to the latest information from NBC, 64.9 million votes have already been cast. Wow. That is crazy. We talk about the election day, and we'll count all those on election day. But I mean,

the election is now. You know, it's ongoing. But, Jason, we have a really awesome show. We have with us today. We have the founder of a company

called Extend Robotics. That's going to tell us all about teleoperation, VR, and how to grab little grapes with a robot. Incredible. an upcoming bill in the Senate that might bring private market investing to everybody, and we have a raft of critical earnings updates. And if you and I don't yap too much, we'll talk a little bit about Apple's new satellite deal,

new stuff from Open AI, and I pulled some new Chinese EV clips for you. So, what to do? Lots to do today on this week in startups. Let's get started. I had seen Alex a video of grapes being picked by robots and AI. we had an investment in a company, Root AI that was doing picking of berries. And this is, I think,

something we're going to see in the coming years, general robotics powered by AI, large language models, helping the robotics, you know, move from doing very vertical tasks to doing very general tasks. So vertical AI,

I think everybody knows, like you can teach AI how to play chess or poker, and have it really focus on that set of heuristics, rules, strategies. Just like CafeX could have the startup we invested in that makes coffee a very constrained ability to make coffee in a box. But to have a robot go fold your laundry, pick grapes, and make you coffee, that's a whole different ball of wax. So let's introduce our guest today and see what he's up to.

Yes. So with us today is Dr. Chang Liu, the founder and CEO. over at Extend Robotics. Dr. Liu, I'm going to call you Chang because that's going to be a lot simpler. But hey, how you doing? Hello, very good. Great to see you, Alex. Great to see you, Jake Al.

I'm definitely a big fan of both of you. And yeah, and a twist and the audience podcast is definitely a big fan, yeah. So tell us what you're working on and the progress you're making. You heard my little T-Up about robotics and how it might change over the coming years because of advancements, specifically in LLMs and AI, learning so quickly and sort of generalized AI versus vertical AI. And do you think I got that sort of teeing it up correct? Yeah, definitely, definitely.

You've had a really good summary. But yeah, just generally for Extend Robotics, we are a UK startup starting in 2019. Yeah, I mean, the extent really the purpose is to develop robotic software to extend human capabilities beyond physical presence, ultimately to amplify world's labor productivity through immersive technology, robotics, and artificial intelligence. So, yeah, so our purpose is really helping the world, making that transition from those traditional vertical factory automations into this real world's general purpose automations.

The aim is to combat the labor shortage, obviously, everybody's talking about that. Yeah, but obviously, we understand the real world automation is difficult because it has to require like a multi-purpose system working for high variability tasks in those dynamic environments, right? And that's generally deemed impossible previously. and we see there as a breakthrough since the, I would say led by chat GPT that the new generative AI is taking over on text demand

or conversational domains. But we see there's an opportunity to embody those similar AI systems into physical form factor like robots and enable human-like experience even for physical works. So we just saw one of your hand-tled robots. doing lab work and picking grapes. Explain to us what's happening in that video. If we pull it up again, maybe you could sports cast us and tell us, you know, did the robot, was it trained or ready on how to pick grapes?

Or did you just tell it, pick grapes or pick whatever you see in front of you if you think it's ripe? Like, how much training do we have if we, you know, we see a robot here doing a hands shake? What do you call those robots that have like two or three hinges on them? Yeah. That we see in a factory. What are those called? I mean, normally, we call it a mobile manipulator. I mean, these ones in the showing in the video.

We recently have more human-like form factors with two arms and similar mobile base. I mean, we developed software that is trying to be more widely compatible with off-the-shelf parts, so we don't manufacture hardware. So I don't want to compete on that, but we build software that is trying to give a lot of options for people to build their own setups. So our software is basically a virtual reality applications that people can use to upgrade their robots. And people can use, once upgraded, they can wear a VR headset. They can use their gestures to operate robots,

and they can have finger-level controls as well on hands. So the purpose of the software has three purposes. First is to be able to teleoperate robots in real-time, and then the second is to collect data for AI trainings. And then the third is to inference AI and be able to supervise and provide interventions we need it. And obviously, the hope is that we can allow the seamless shared automations happening, both with human as well as AI combined together, and they can basically complement each other

over time. And as the AI goes more capable than it can become. I want to jump in there. So I know that extent is working with Queen Mary University on this grape picking project, But just to make sure that I understand, the current level of the technology is teleoperation, robotic arms in the field, and then allowing the robots to learn from humans doing the motions for them to actually pick the grapes. But in time, this will get to the point when the robots are able to do it mostly by themselves, I presume? Yes, yes, exactly. We certainly view the automation as a spectrum rather than overnight miracle step change.

and there's always some level of human there and hopefully towards the future there will be less of a human and more of AI in a process we call it human in the loop AI and as a work collaboration between human and AI it means that it makes this smooth transition basically where one human

can simultaneously supervise more more robots depending on the level of autonomy that each robot becomes. So that's why we amplify labor. We don't replace labor. We amplify labor with AI. And it's going towards the same direction. But obviously, we're proposing a more smooth transition here. All right, listen, interest rates are coming down, but it's still hard to raise a new round.

Everybody knows that. The last thing you need is to get bogged down in the minutia while raising. So if you're a founder overwhelmed with your cap table management, due diligence, and frankly, managing your investors, you need to check out Fidelity Private shares. Fidelity understands that dealing with equity can be incredibly time consuming. That's why some founders joke that fundraising is a full-time job. I'm not working on my startup. I'm managing my investors. I'm doing fundraisers. And it shouldn't be, right? With Fidelity Private shares, you're going to easily manage your cap table and your data room. You're going to get a more accurate forum. 9A valuation, and you're going to get it quicker. You'll also be able to model the future of your company's equity, which you need to do as the founder, because this dilution adds up. People start signing notes. They don't know what the impact of those notes will be. And you know, you want to model that out.

And you can fully automate your next funding round. So you stay on track. You seem professional. Even if you're scattered, you know, you want to come across as very professional and buttoned up. And that's what having a great partner like Fidelity will do. And you're going to get great support from Fidelity, a great startup community. They do a lot of cool events with founders and VCs. I'm going to be going to some of them.

So check out FidelityPrivate shares at FidelityPrivatechairs.com. Mention our podcast. You're going to receive 20% off your first year paid subscription. And Alex, this is what we, I think, are hearing about self-driving, that there's 1.5 mission specialists. We were told that Waymo. We don't know if that's true. Waymo says it's not true.

Cruise had virtual operators. The idea here is humans in life. the loop, watching the AI learn, and then eventually it steps back and watches 10 people picking grapes, or it goes from one and a half or two car robotaxies to 10 to 100. And then the mission specialists, which I love that term, eventually are just looking at very unique moments in time. Yes, Professor? Exactly, exactly.

So they generally should be just the supervising what's happening and just monitoring. just in case, and then only needed, they come in and provide those interventions. And always you can, you can teleoperate whenever you need. So I want to talk about the grapes, though, because one thing that I don't know much about growing grapes for wine, but I do know you have to be pretty delicate with the actual fruits themselves. So how much of a challenge was it to train these arms or teleoperate them to not, you know, break the skin of the fruit in ruin, essentially the wine?

harvest. Yeah, so there's certainly level of integration, innovations needs to happen in terms of the sensory information, be able to sense force at the end of factor to both provide feedback for operators, but also information for the AI models, make sure there is a clear feedback on applying force. Because obviously, for grapes, it's very important to make sure not damaging them, and they are very delicate, and especially for wine making. We're working with Suffraim Grange,

it's the wine, English sparkling wine maker, and they're very keen, I mean, they're very, they work really hard to make sure the wine is on a good quality. And we're working with Quimera universities that are very experienced on custom end factors to be able to incorporate force into the sensings. And we come in as a human in the loop AI software capabilities. As a consortium, we try to build this together.

But Dr. Luke, in the effectuator is just a fancy term for robot hand. Right? I think Grippers is the term that I've heard, right? Grippers. Is that the term in robotics? Yeah, grippers and effectors. sometimes you can have different form factor for the quipers, yeah. Okay.

When do you think a wine harvest will be done completely robotically? What year if we were to put a bet on this? Yeah, I think, I mean, in terms of our current project, we are, we obviously quite early in the stage. I would say in the next three to five years, this is something could be commercialized. But obviously, this is also a process of commercialized. commercializations, that's going to take some time. But I think technologically,

we're getting more mature in the next five, three to five years. I think what's going to be amazing about this, Alex, is if you think about training skilled or unskilled workers, right? And when people start the job, they're unskilled, and then over a decade, maybe they become skilled at it. Sure. It's got to be extremely frustrating as somebody who runs a vineyard to have new people, typically probably migrant workers, come in and then train them of how to not damage the vines and how to do this correctly. And now if this becomes automated, then that means whoever the best caretaker was of these grapes and these vines would become the global single vine manager in the world. So this is a really interesting concept, Alex. Imagine if you had one, if you had the greatest

pilot of all time, let's say Sully, who landed the miracle in the Hudson. Let's say he is, in fact, the greatest pilot of all time. You could have that pilot in every single plane. Here, you could have the person who manages vines the most perfectly manage every vine in the world. In a other words, the most perfect practitioner for any sport behavior in the world. And then the really dystopian or weird moment in time for all of this, Alex, will be, I think, instead of having migrant workers leave Mexico to come to California to pick grapes, they could stay in Mexico, not cross the border, and have VR headsets and have the robots in Napa doing this work.

It's kind of crazy when you think about that three to five year period. No, Alex? I mean, I think one thing that we underestimate as a technological breakthrough is just broadband penetration around the world. Because what undergirds what you just said,

it's just fast connections. Now, Dr. Liu, I presume that if I'm going to interact with one of these robotic arms with the gripper at the end, I need a pretty fast connection. How latent is, It's the I move my hand and then the robot moves its

own arm. Is there a number you can put on that? Yeah, yeah. So actually the data streaming latency and bandwidth reduction is one of our key IP about ability to really quickly compress the data, especially

in this case we're compressing 3D data to enable immersive experience in VR. And we have the full patent on on it. And basically, we can, our ability to stream data is

about 100, 150 milliseconds. And that's obviously on top of the latency of the, of the network. Okay. Will it work on a commercial internet, or do you have to have a dedicated

network? And would it work on, say, a 5G network? Could you do this with a Verizon or even a Starlink satellite connection from a low Earth orbit satellite, or is the latency not good enough on those platforms. We can already do that, yes. So we can use on Wi-Fi 5G.

We're just working on testing on Starlink's. Yeah, so in fact, I think Starlink is probably the lowest latency in terms of network because you don't have to go through the fibers. So it's a more straight line. But we're working on testing that. But generally, ability to compress data small enough that you can actually stream over that standard network is one of the four capabilities we can offer. After wine, what do you got? What's the next after you pick grapes, or is it just the company's going to work on picking grapes for a couple of years?

Yeah, we actually, so we're quite wide in terms of adoptions. So we actually started with nuclear applications as a teleoperation. interface, working with Atkin, Realis. And we then take on some project with Airbus on space robotics. And recently we started this new ground, I mean,

a year or a year and a half into this future farming projects that is on agriculture applications. And I think we are getting into the field that we are a lot of traction coming in from manufacturing industries.

especially automotive and a few more like manufacturing applications where we see there's a lot of labor shortage as well. So I think we are getting towards this phase where we start to focus on manufacturing, but I think all these applications really allow us to build a general purpose technology. And for us, because the technology is so general purpose

and widely applicable, all of these applications really contribute to advancing the technology as well. We're looking at more commercial adoptions at manufacturing applications. All right, Dr. Lou, we have to let you go. But one more question before we do. I know you're a UK-based startup. I also know that you guys have raised some grant money from the UK government.

How supportive has the UK been to its startup community lately? I know we're in a post-Brexit world, but I love talking to folks from different countries. So just give us the quick layer of the land for building a technology company in the UK. I think UK is probably one of the best environment for startup in, if you are a software-focused startup, software and AI. And then there's a lot of government support in terms of pushing innovations, happenings, and helping startup, small companies to innovate.

And there's a lot of venture capitals available to talk to, especially around Linden area. there's yeah so I think a thing is a good starting starting point when you uh when you make your next trip to london to get a big check shoot me an email and uh we'll catch up on that because I'm very curious to see how the company goes but um dr luke thank you very much extend robotics everybody

you can google it and uh if you're into sparkling wine perhaps check out and saffron grange yes all right thanks doctor awesome appreciate it good work up uh thank you thank you so much really interesting i think when we see these demos and a lot of times I'll see them go by

on LinkedIn or X and the production team we talk about it what's underpinning this change where are the second order effects that will happen is what I always try to go to because then we can figure out the trends

and I think there's a couple of very interesting ones here one is the greatest single practitioner in every vertical I don't know how to phrase this exactly but the perfect pilot I think might be the way to think about this

and we're the perfect poker player. Sure. You are going to have, for every profession, the greatest practitioner addressing every single instance of that task. And so for picking grapes, it seems mundane. It probably seems like not super important. But it probably is because great winemaking making requires winemakers, winemakers and this entire process, all the secrets will eventually wind up in AI and the quality

of wine, like the bottom half of wine could become as good as the top half because all of the secrets, all of the little nuance will be captured. So it will elevate the cost of, it will elevate the quality while reducing the cost. Founders, I know that you're keeping a close eye on your burn rate. I am too. In today's venture market, every single hire you make has to be perfect, right? You can't make mistakes. You got to keep that runway as long as possible so that you can run more experiments and you need talented people to run those experiments and figure out how you're going to get product market fit, how are you going to scale your company? And that's why you need to use LinkedIn jobs. As you know, LinkedIn brings you the candidates that you can't find anywhere else. LinkedIn passed

the one billion member mark. Think about that. One billion members. And seven, 70% of LinkedIn users don't visit the other leading job sites. This is a phenomenal statistic. They don't even go to the other job sites. Why? Because they might not be looking. And those are the best hires. But they're hanging out on LinkedIn, doing professional development,

checking in on their network, building their network, sharing content, finding leads, all that great stuff. Bottom line, there's amazing hires waiting for your company on LinkedIn and nowhere else. And they have a special deal right now. Post a job for free. What? FREE, what a great price. LinkedIn.com slash twist.

That's right. LinkedIn.com slash TWIST to post your job for free. Terms and conditions, course, apply. I want to double click on exactly your thinking here, because to me, the best pilot, the example you used was solely, definitely a flesh and blood human. But when you talk about these AI models

becoming the best of what they do, are you referring to like a digital, digitally trained AI that learned from the best human to then become the best technology product? You would assume that all, all of this monitoring of great humans would then result in the greatest Jedi pilot of all time. And it could also learn just from all scenarios. So if there is a scenario where, you know, when you're picking grapes to learn, hey, you know, this bundle of grapes needs 24 more hours on the vine.

And this one, you know, we missed and we should have picked it earlier. over time learning that is something that could come from a human's perception or over time the AI could learn, hey, when we produce this vintage of wine, it got a lower score than the last five years. And let's go back and replay what we did. And that type of diligence on tasks exist for planes. Every time a plane crashes, we... We have this wonderful thing that happens out of that tragedy, which is the entire system gets better.

They focus so much energy at such great cost on making travel safe that we're now at a point where we've had no commercial airline deaths in America in like a decade. Yeah. It's phenomenal. And, you know, tragically, whenever the next one occurs, they will put millions and millions of dollars into figuring out why that happened. Even if it was a drunk pilot or just a broken bolt and it's like super obvious, they will track that down. Now, that doesn't happen. Those kind of postmortems, those kind of debriefs does not happen when you're picking grapes. I am certain.

So now you're going to have that occurring. So the perfect pilot for every profession is coming. The perfect professional for everything is coming. I hear the implied add back to the theme list. So trust me, it's going on. One last note about this before we move on. Also, by the way, the other piece, the part two of this is, how strange is it that telecommuting, remote work is going to be global? So we're sitting here watching Amazon and Andy Jassy deal with like a mutiny over at Amazon for when they come back five days a week in January.

and here we're going to have people potentially picking grapes from Mexico who would be migrant workers, but since we were shutting the border down, they would be able to do that work in a call center or from their home with a satellite connection, just in the same way Athena, go to Athenawau.com and get a virtual assistant. We're investors in the company, bold disclosure. just like you could have somebody in Manila working in your operations department like we do here at launch and many other startups have virtual assistance

because you can't find assistance and admin workers in the United States. Sadly, people don't want that job here. So it's really interesting those two components of this that I am not struggling with, but kind of inspired by. Yeah, it's going to be interesting to see how long U.S.-based developers can charge the wage premium they have been compared to their international

compatriots. Like, for example, if you're a pretty solid developer in the U.S. and you live in Seattle and you work for Amazon or Microsoft and then you move to, I don't know,

France, I mean, you would take a pretty serious salary reduction. So by that logic and to your point about teleoperation and people from around the world taking more part in the

economy as we see if Marlend, does that mean that in time one of a developer's best traits in the U.S. is going to be their ability to go to an office because the person in London or Manila or wherever can't. So I wonder if that's going to be their saving grace.

I think that will be part of it. Just like a Somolier, you could have the greatest Somelier in the world, telecommute in to tell you what to order, but you'd still want to go to the restaurant and have that experience of the Somelier in person, you know, real world, meat space, as we call it back in the day, you know, would be charming and wonderful. and I think that's probably how this will hash out. And they'll just be a global standard for salaries that I think we're going to.

Because if you look, you know, like you'll have somebody in Manila or India or Kenya. These are all emerging markets for virtual assistance business process outsourcing. And they're all, let's call it, low single digits per hour workers, three, four, five dollars an hour. In the United States, minimum wage, seven, eight bucks, I think. but the average wage more like 15 to 25 and an assistant in any major city would be 50, 60, 70K probably, which then puts you at $35 an hour, $30 an hour, something in that range. And so that's what I think is happening is you're going to just see the salaries converge on a global average for work. and then people will decide if they want the luxury item of having the person in person.

To your point, like maybe the 150 or 250,000 developer coming to the office at Meta, making high-end decisions will be fine. That person gets $250,000 for dragging their ass into an office and eating, you know, $30 Neiman Ranch steaks for free. And then they'll have five developers working for them for 50K each. the same salary on their squad. And I think that's like going to be a very interesting efficiency play. It's going to be tough though. It's going to be tough for people in the U.S.

who are accustomed to being, what you're saying is the A plus players are fine. They will work with A plus players in lower cost of living areas. What if you're a B? You're screwed. Yeah. I really, that seems to be the case.

B is not better. It's bummed out. B is for bye-bye. We're done with. you. There's a lot coming here. I want to make one last point, though, about wine. I had to quit drinking, but when I used to drink, I used to drink a lot of whiskey, big fan. Love scotch. My scotch tastes quickly and surpassed my budget, actually, so I switched to bourbon. But I heard a really funny, apocryful story about a scotch distillery, and they got some sort of, like, health inspection, you know, people wanted to come and open the doors and look at them. And they

found their room where the whiskey was aging, and it was just full of cobwebs. Like, just a an absolute mess. And the health inspector is like, you guys, you got to clear this out. You can't have, I mean, this is a health hazard and such.

And the whiskey guys were like, but, but if we take them out, they've been in for so long. What if that changes the flavor? All that's to say that the alcohol industry is not exactly progressive on technology at all points. But the labor shortage will drive a demand for technology products.

And so I think they're going to get better. That's exactly right. You know, if you do not have vibrant immigration in your country and you do the protectionist thing, which, you know, people have the right to do. And then societies can make that decision. Japan certainly made one where, you know, you can't be a citizen of Japan, but they're trying to get people to move there to work, to fill those gaps. And they're, they're quite promiscuous when it comes to giving visas, but not citizenship.

You're going to see that blend. But if you can't find workers, people are going to find other solutions has happened in the cashier business. And the concept of a cashier 10 years ago, you would be perplexed to go into a restaurant, a quick serve restaurant, and not see cashiers. And here we are 10 years later, you would be perplexed to not see online ordering. And so things can move quickly and humans can adapt very quickly to these changing circumstances. Yeah. All right. Let's talk about AI, everybody. It's everywhere self-driving cars, breakthrough medical research,

and of course business tools that make everybody better and faster at their jobs. If AI hasn't hit your industry yet, trust me, it is on the way, and it's happening fast. But AI requires a ton of computing power. You know that. And those costs will skyrocket if you're not careful.

I have seen these bills come in. It can get crazy, quick. And that's why it's time for you to upgrade to the next generation of cloud. And that means Oracle Cloud Infrastructure or OCI. OCI is a blazing, fast and secure platform that covers all the bases. Infrastructure, databases, app development, and of course, all your startups, AI, and machine learning needs. Now here's the kicker.

OCI costs 50% less for compute and 80% less for network compared to other cloud options. That's a lot of savings, especially when every minute counts in the cloud. Thousands of businesses have already upgraded to OCI, including MGM resorts, specialized bikes, and fireworks AI. So here's your call to action. Right now, if you switch to OCI, Oracle is offering to cut your current cloud bill in half for new U.S. customers with minimum financial commitment. But you'll need to act fast. The offer ends December 31st, 2024. See if your company qualifies for this special offer at oracle.com slash twist. Oracle.com slash twist. Let's keep moving down the docket here. I know we got a lot to get to. Yeah. So one thing that has hasn't changed and some people wish that it would is the rules about investing in private companies if you are a U.S. citizen. Jason, there is a thing called the investor accreditation program. If you meet certain thresholds, you're considered to be intelligent enough to buy securities that are not publicly traded. But there's some movement in the Senate to actually change that.

But first, I want you to tell people, what does it mean to be an accredited investor? Sure. We have laws in this country around accreditation. It basically means that you make over $200,000 a year, or you have a million dollars in net worth, and it can combine with your spouse, if you're in a couple. And so it basically means if you're in the top 5% of earners or wealth holders in the United States, you're what's called accredited. Another word for this might be sophisticated, intelligent, but accredited is the word they use. Now, when you have that status, which you could have gotten from winning the lottery, having a great, job that pays well. Having an inheritance, any of those possibilities could be involved in

some illegal activity. Any of those reasons that you have accumulated wealth in the mind of the SEC and our government means you can take the risk of investing in alternative assets. Alternatives in this case would be private company. So you could invest in Uber or LinkedIn before they were public. When something's public, it goes through a lot of vetting, a lot of sunlight, a lot of audits. We still have fraud in public companies,

but it is a fraction of what could happen in private companies like Theranos or WeWork, where you saw fraud or incompetence or just the Wild West. It doesn't necessarily even have to be fraud. It could just be, they're swinging for the fences and things can come apart at the seams. Now, the reason we do, this in America is to protect what the government very condescendingly believes that 95% of the

citizenry here are stupid and unable to make these decisions. At the same time, you can go on to draft kings and, you know, or prize picks and do a parlay and bet on the Knicks or wherever you feel like it, which I'm totally in favor of and I think it's great. You could bet on public equities, and on the margins you can do crypto, and crypto falls into this. So the question is, who gets to participate in this? And the reason it's important is because true wealth creation and the ability to move from one strata to the next is largely determined by your ability to bet on these, and I'm saying bet, because it is a risk, to make these investments in highly volatile assets.

And if you were to have the ability to invest in Uber when it was $5 million, that could work out very well for you. Or if you could invest in LinkedIn or DoorDash. And you know what? If you worked in human resources or you were a cab driver and you saw LinkedIn or you drove for Uber, you would immediately want to put $1,000 into those companies

and you would be rewarded with a thousand X. were more. So I'm very much in favor of allowing people to take a test to become what I'll call sophisticated. And actually, we have tests for cutting people's hair, if you want to cut people's hair, for driving cars, and in some cases for owning a firearm. In some places, you can just get one. But we don't have tests for gambling on sports or going to Vegas and playing blackjack. I actually would be in favor of if somebody wanted to go to Vegas and gamble, them having to take a test on each game that they play. I'm dead serious.

If you wanted to buy more than $1,000 in chips, why not give people like a little test? Here's the rules of blackjack. Here's the rules of poker. You have to know if a flush pizza straight. Now you can play. I would actually be okay with that. And that's the same situation here.

I'm just thinking about how much money that would have saved me regarding the game of craps throughout my life, in which I put money down on the table. Magic happens and then my money goes away. You never understood the connection between the things. it just evaporates from the table. Now, the thing about the number of people who count as accredited investors is very interesting.

The journal covered this empowering Main Street in America Act, which is what we're talking about, the bill in the Senate. But the thing is, the number of people who meet the accredited investor test, if you will, or threshold has dramatically expanded. So back in the day, Jason, your 5% number would have been dead on. It was 1.8% of households in 83, but thanks to inflation, and just the changing value of money, now there's 24.3 million U.S. households

or just about 18% that meet this threshold. So my point is this. We've already allowed the accredited investor standard to dilute itself so much that it's become pretty much meaningless. So then why do we need it? Why do we just get rid of it? I think the answer to your question is because if you were running the SEC like Gary Gensler is,

you would have the mandate to protect consumers, above all else. And that mandate then means when something like FTX happens, you have egg on your face. You bear the brunt of that. So they are risk averse, which is the mandate they've been given. They are not given the mandate to have the polarization of wealth and the wealth gap in our country close. And that's what actually this legislation would do. If people who are just starting their careers, who had under 200,000 in net worth, but you know, who made 70,000. $25,000 working at a venture firm or in human resources or in accounting. And they happen to see Coinbase or, you know, LinkedIn, which is such a great example, because so many people got

their jobs on LinkedIn or posted jobs on LinkedIn. And those folks would not have met in almost all circumstances, the, in the majority of circumstances, they accredited, but they would know that that was a good bet to make. And here, you know, that that is the reason. And, and, and, and, you know, incentives really matter. The incentive the SEC is not to get poor people to become middle class or middle class people to become millionaires. That's not their mandate. Their mandate is to protect the public and build trust in the financial system. Okay, fair enough. But that's that's the problem. It's sort of like with the rocket ships and the launches in California, you have the coastal commission, which, you know, I guess virtue has this virtuous mission to protect the coastline, California. California's great asset is the coast. It's like got this incredible, you know, natural bounty. But we kind of want to get to Mars and get things in space as well. But their mandate is to protect the coastline. And okay, if you want to protect the coastline, oh, you should just say, don't build anything. No houses, no rockets, no pier. And, you know, we had this in San Francisco. I remember in the Bay Area when I live there, somebody wanted to put up this Ferris wheel in Golden Gate Park. I don't know if you remember that. The Ferris wheel. idea. I would totally have used that. And like, people were like, you know what, we don't want to have like the London giant Ferris wheel because I don't want to look at it. It's in my backyard. And then the other group of people were like, but kids and couples in the middle of a romance would go up there and, you know, propose to each other and it would be charming and wonderful and, you know, the naysayers win. And so that's really what's happening here is the incentive. for one group of people is outweighing the big picture.

And then that's where leadership comes in. And this is, I think, what's happening. Now, people will, when they make this sophisticated test, then people take it. It should be, in my mind, a three or four hour preparation for the test. And it should be 50 to 100 questions. In other words, it should be the equivalent of, I don't know, a driver's license.

I think the driver's license is probably the right. amount of work. You could pass the driver's license, test and practical driving with what, three or four hours of training? Don't tell a 16-year-old Alex who failed it the first time. Well, you were probably not paying attention. So anyway, I think this is great. We're starting to see this now come up over and over again. And if we do this, it'll make America more competitive because more people will become wealthy. More people will put risk capital to work. And the amount of appreciation in companies ultimately happens in the private market more than it happens in the public markets today. So if you were to invest in, I don't know, Instacard or something or Reddit, how much

of that occurred in the private markets, you know, as opposed to the public markets? Probably a lot of the acceleration occurred in the private markets. If you want to put that into perspective, imagine a company that goes public at, say, a $10 billion dollar valuation and then a year later it's gained 40%. And now it's worth $14 billion. Great, but that means that your possible returns as a public market investor were 40%. But if you bought in at a $10 million valuation, it's several hundreds and hundreds and hundreds of percent. If you want more about this, it's a Senator Tim Scott, South Carolina is the driving force behind it.

And it has a lot of others have in it as well, including extending the jobs. Zach Chilliardust have up to $2 billion in revenue to go public under different rules. And there's a lot to it. but mostly I think it's a good package of things. And I hope that people don't get their faces ripped off too often, but capitalism. I mean, could it be any worse than buying crypto coins that have no value in the real world? I mean, it can't be worse than that. Could it be worse than gambling on sports?

I don't think it's going to be worse than those two, and we allow those. And it can't be worse than going to Vegas and playing crap. So this by default, I think, will be, this will exist. investing in private companies will exist exactly between gambling and sports, you know, go to Vegas and public markets, which seems reasonable. There could be something right in the middle, which is you invested in a company, it was private, the founders were lunatics, they took a ton of risk and they lost your money. Okay.

You know, but you only put 5% of your net worth into it. So the other 95% were your home, your equities, your bond, your wealth front, perfectly design portfolio, whatever it is. You know, I remember this tweet that I saw this guy was like, you know, Angel investing is the best way to have less money. And I just, I just died. Because, I mean, if you're Angel investing, you're, I mean,

that's idea stage, couple people and like three lines of code. Like, it's pretty amazing. I wrote a book on it. You are going to need to hit 30, 40, 50, qualified investments, according to every angel I've ever talked to. They've said between 30 and 50 is the number of bets you have to place in order to have one power law potentially,

nothing is guaranteed, potentially, let you hit break-even in a profit and beat what happens in a low-cost index fund. Now you do have the outlier that you could hit some grant slam. And that's reason enough in my mind to do it with a low single-digit percentage points of your network. That is the advice I would give to you, Alex. If you told me, hey, I want to get involved,

I'd say, great. I know how much you make, relatively speaking. you have some amount of savings, 5% or less. In other words, if you lose it, you would make it back on your other equities and interest that year. So if you blew out the 5%, you'd still be even, right? Because you're probably making 5% on your existing equities and money.

Also, like the rest of our family investments are, you know, like zero cost index funds and low cost international bond funds. So like, are my personal financial positioning as a family is so conservative that if this did come to be, and if I wasn't trying to maintain my journalist tag, yeah. I mean, also,

I just want to bet on my friends. Like, whenever with someone that makes a new company, I just want to, like, buy into it. Yeah, I mean,

that's like, I frequently get asked to invest in restaurants and movies, and I have the standard joke, which is, I, I would, but I, I like money too much.

So I don't, because I like money. And people laugh and whatever. And I generally just pass on those investments, because I know that it would never work. Now, I did invest in my friend Nick Jarecki's film with a group of my poker buddies. And if I did not care if I made money or lost money,

because that's my friend, and I just wanted to see his movie do well, and it did well, and I'm sure we'll break even on it. Yeah, but for example, there's a coffee store near my house that just opened up. It's called Brown Bee. It's here in Providence,

and I know the owner. His name is Walid. Lovely guy. I've been walking by that place, Willie, he's been, you know, building it out, remodeling it. It would be so cool if I could go up to me like, I love you. I love what you're doing. Let me buy 10K worth of this company. But like

the system isn't set up for that right now. It's set up for bank loans and you know, this other old school stuff. Imagine if they could put in the window we're raising money to redo the store. It's going to be a $100,000 renovation. We're looking for 100 of our top customers to put in $1,000. And whoever does gets upgraded to a then, you know, a large coffee, but for the price of a regular for life, help us out. And you'll also own 1% of, you'll own, yeah, for every 10,000, you'll own 1% of this million dollar enterprise. And, you know, if you make money, you make money. But if you don't, you don't care. Yeah. Because you want it to support your

little. But that would be illegal currently to do. Which is, just sounds so ridiculous. It's done. Anyways, we will eventually get this stuff sorted out. But in the meantime, Jason, I have some big AI news. I don't know if you caught all the open AI announcements this week, but three things happened. One, they did an AMA over on Reddit, which was incredibly well received. And I have answered to the question of where the hell is GPT5,

in case anyone is curious where that model is. Essentially, Sam Malvin said, hey, we face a lot of limitations and hard decisions about how we allocate our compute resources towards ideas. So essentially, Jason, they're compute constrained even with Azure in their pocket. That blew my mind. Yeah, they're blowing through five, 10 million, 10 billion a year in infrastructure and losing a lot of money. And that does say something

that, you know, they're running out of bandwidth. And I saw Jensen Wong who was talking about Elon standing up this giant cluster. So the buildout keeps occurring. I do think the interface of chat GPT is superior to any other interface. Yeah. But the results are not. So I really feel like Google has to make a Gemini app, a dedicated app that 100% of the interface is designed to be used by consumers. And I really love Canvas on chat GPT, which allows you to write notes. And Claude has had this for a while.

But you ask you questions, and it just keeps refreshing the same document, which is really nice. I was writing the liquidity overview document for the next liquidity conference. And as I was writing it, instead of using a word processor or grammarly, as I am prone to do, I did it in Canvas, a canvas in chat chepti. And that was the first time I edited a document

without editing the document. I gave the AI commands to edit the document, then I cut and basted it into a grammarly or a Google I can't remember what I did or a Notion page, one of those three. And so that's kind of this interesting moment I think we're going to be at. And I'm also been doing, as I may have explained, when I drive my daughters to school, I put on chat GPT's voice application. I asked them what they're working on in school.

And then I have it give us quizzes on it. So I said, hey, give us some increasingly difficult questions in American history. And it was great. It was fun to do with the chat GPT. And who was the first president in the United States, who freed the slaves, who wrote the Declaration of Independence. You know, and I was like, yeah, who did write that first draft for the Declaration? Oh, yeah, that was Benjamin Franklin. I got it wrong.

And I was like, I knew that answer, but I had forgotten because it's been 40 years since I was in history class. Yeah. And I was just thinking about tutors. You know what an advantage it is to have tutors? Did you have tutors growing up, Alex? No, but my parents afford it or no. No, well, my dad, who was a PhD in chemistry, he actually held what he called homeschool on Friday nights through Saturday morning. So we'd have our friends over and have a sleepover and do science experiments and write poetry.

in. Basically, they just robbed into our public education. Incredible. You know, my parents couldn't afford it. They were busy working. And so I was always a little bit jealous of my friends who had tutors. And when I got older, I realized, oh, poor people or middle class people don't have tutors. And all the people who I met who were getting into colleges were like, yeah, you know, I was falling behind you mad. I had a math tutor. Yeah. I was falling behind in science. I had a science tutor. And then I was like, oh, what an incredible privilege it is to have that. And now that's completely democratized with a custom AI tutor for free, for life on any subject. Holy cow. The world will become more just because of this technology. I am so here for it. If you have kids,

try this. Just get into one of these, have it do quizzes, and it will remember where you're at and you will become superhuman, so to speak, and your ability to tutor. I am very very, very, very interested in how Google, speaking of Google, did in their earnings. I did see that YouTube was a blowout quarter for YouTube. Yes. And that they did $8.9 billion in quarterly revenue in advertising. And if you put those numbers together, they said they did $50 billion in revenue over the last year. But $35 billion of it was advertising, which means the other $15 billion was subscription revenue. And that to me is extraordinary. So maybe you could take us through, and also the cloud had a blowout quarter, but maybe searches were down.

So this is really interesting. And I know Google went up, their shares went up a bit, and they were one of the ones who was maybe lagging. So let's drill in to what's happening at Google and then what we can look at in terms of second order impacts and what founders can sort of ascertain from this incredible quarter. Yeah. Okay, so in Q3 over at Google, just to put some high level numbers out there, revenue was 88.27 billion, up 15%, Jason, pretty solid growth for a company of this size, earnings per share, 212, both of those beat expectations. Now, you're right, Google Cloud brushed.

Revenue there was up 35% to 11.4 billion. And critically, Google Cloud now makes money. So a year ago, its operating profit was about 266 million, basically break-even for a company of Google size. this quarter, nearly $2 billion in operating profit at Google Cloud. In a quarter. In a quarter. So, I mean, this is, as a standalone company would be an extraordinary company.

Oh, absolutely. This would be an enormous public company. We'd be talking about just this company. But that's just one part of the course, Google Empire. On the search front, Google made a lot, or alphabet, I should say, made a lot of noise about how AI is actually making search better and used by more people. does not really reflect my personal use. But hey, you know, sure.

Explain that, unpack that, because this is, I think, critical. Sometimes we see something in a stock and what management is saying, but your own personal behavior is radically different. Explain this disparity that you're expressing. Yeah, so pulling a quote from their earnings transcript, Alphabet said they're, quote, seeing strong engagement, which is with AI search, which is increasing overall search usage, and user satisfaction,

which means that the average user of Google search is not me, which actually helps explain to me what Google Search has become, because it's moved further away from what I want, but apparently I'm more in the minority than I thought. Now, to your point about startups, I think that means that you really need to go talk to a lot of people in like Oklahoma and not just San Francisco or Austin or Seattle or Tech.

We are the Vanguard. We are the tip of the spear where the earlier, adopters. But as early adopters do, the massive middle and the laggards will eventually do. So you and I start now, I think, a lot of searches with an LLM, whichever one it is. I am doing probably 20 or 30 chat GPT sessions per day. And then probably I do five or 10 on Gemini or Claude or other LLMs just to test them in comparison, sometimes GROC when I'm on Twitter, X. So if we were to look at that, I think my use case, my default now is to ask chat GPT. My default is not to ask Google.

I think you're the same. And we're trying to think of the instances when I prefer Google. And I find Google oppressive when I'm doing almost any search because of the number of sponsored links that are there. It's the whole top of the page. It's crazy. So I think this is the lesson. When you run a business, if you become a monopoly,

you squeeze and squeeze every ounce of revenue out of that user base, as you're supposed to do. But as you squeeze and squeeze that, the users can slip through your fingers to a better solution. And I think that's what's going to happen to Google. That's the big challenge is there's no advertising in chat GPT. It could be. you could have links in it when you do a search.

And I wouldn't mind having a couple of them. But you kind of get the answer. And then if you get the answer to your question, how many days between this date and this date? Like I was doing that for the election. I would say how many days between now and November 5th? I would go to chat GPT to do that because it was just easier. And I would get the right answer and I wouldn't click on any links.

Or if I want to know, like I said, basic history facts, I used to do that on Google, find a web page on the open internet, and then do it. Now chat GPT just steals whatever that web page had and regurgitates it and intercepts that content. So this is going to be very challenging for them. The second lesson here is probably really good idea to have multiple revenue dreams. Because no matter what happens to Google search, the investment in Waymo, YouTube, and cloud are going to save the company. because now they have massive revenue, $50 billion from YouTube.

You said they had $2 billion in profits. I don't know what the top line was on cloud. For cloud, it was basically $11.4 billion in revenue and then just under $2 billion in cloud profits. So then you have basically $250 billion a year of businesses in YouTube and cloud. If you have $100 billion in revenue over here, that's not search.

That's pretty great for balancing it out. and we don't know what the Waymo revenue is, but that is a black hole of profit. There's no profit in that business. That's going to lose $5.10 billion a year for another 10 years. Those long bets actually are probably paying off.

I think so. So a couple notes about that. One, to put Google search in comparison compared to its other businesses, Google search and other, the line item in Alphabet's earnings, was just under $50 billion just for Q3. So when we still think about Google,

keep in mind the search is still the vast driver of its income, though I think we both agree that that's going to be under a lot of pressure. Two, Open AI put its new search product live for everyone out this week. If you're on the web interface and you're a paid subscriber. So I've been playing with this. Any good? It's great. Oh, yeah.

Do you want to share your screen and show it? Sure, don't mind at all. Yeah. I mean, I did see that they were going to do this. this. So this is their perplexity killer. I know perplexity was very upset about this because they were used it. They were built on chat GPT.

And then Sam decided, shockingly, that he was going to compete with them. So I need a good query. And I'm going to go ahead and push the search the web button when I get it in. But Jason, what do you want to learn about? Let's say, great ski vacations for families. I push the wrong button.

Oops. Okay. So I'll just do that again and I'll do it with search this time because I know what I'm doing. I see. So you click a different button there. You click the search operator button and then here we go. And it gave us a couple of images. A table.

Maybe you told it to do tables. No, I didn't. But it did it both times actually. This is the regular chatypd four-hour response and then this is what we got. So, Jason, do you prefer smugglers notch, big sky or, oh, I love Beaver Creek. That place is great. So yeah, I mean, here we go. And those are all links, right?

You link out to those places or no? No, those are not links, but we can click the sources button, which then brings up links to where it pulled from. And essentially, it appears to have gone through and read every single listicle that was SEOed for Google and then compressed those down into a list for me. So this is the listicle of the listicle. It's a meta-listical. Yes.

It's a meta-a-eckle. And this is where I think they're going to get sued to all hell and back again. And this is why the New York Times should not settle their lawsuit. The New York Times should take it to bat. I think they should go to bat on the lawsuit on behalf of the content industry. Because if you see what Chatsyp is doing, they're stealing all that content processing, presenting it as their own and those and then burying the citations and links, which is decidedly more cutthroat than what Google did with their one box.

Even Google with the one boxes, which is, you know, or their snippets under the link where they give you like one or two sentences. that pissed off content producers, but at least you got some traffic from Google so you could say, you know what, I'll make this deal with Google. I'll make the deal to put my list of clothes into the search engine knowing I'm going to get links.

And it was this little dance that content creators, publications writ large, made with Google. You take a little bit and send us some traffic and we're all good. Here, Sam Altman is taking everything and giving nothing to the content providers. And that's why I would say this is destined to get him massively suit.

And far be it from me to Simp for, you know, $100 billion companies. But I'll stand up for them here. They did note in their official corporate blog post about this launch of search that they are working very closely with their existing content partnerships. So I think the idea is to lean pretty heavily on that side of things. One more open AI thing. Why not put the links at the top level, though? They obviously are being selfish with that.

That's why I said, are the links on the top level there or are they buried? Bering the links, I think, is the key mistake here. I know aesthetically it makes it look better, but it should say from BuzzFeed, this, from here. Or cited by BuzzFeed, it should say cited by BuzzFeed, Travel Ossity and Travel and Leisure and, you know, Alex and Jason's ski. Vacations.com. And if it did that on each one of their citations, then you would have this like recognition. It would put the fair in the fair use here.

Yes. I'm with you on that. I'll be very curious to see how this all plays out. But one last open AI thing, they brought the advanced voice feature that you were describing to desktop this week. I didn't know that. Wow. It's really fun.

I can't do a demo because I can't have the audio play out of my ears into the computer. But if you are a paid chat chief T subscriber, go try out the desktop app, advanced voice mode. It feels different than you'd think because you're talking to your computer and it's talking to you while you're doing other stuff. And it feels like a different, it just feels different to me in a really cool way. So I recommend everyone try that up again. I would really love for it to be monitoring my desktop right now. You and I talking, we will be doing this show next year.

And you'll say, I'll ask you. hey, how many users did, how many paid customers does Uber have? And it would just come up like as a sentence while we're talking and say, according to Uber's Q, you know, Q3 filing X, Y, and Z. And that's where it's going to get really interesting

is when there's this persistent agent on your desktop, watching what you're doing, and giving you supplemental information. I would love to have another monitor here. That was my AI agent. that was watching us talk and just giving me supplemental information. Like, remember pop-up videos?

Do you remember that on? Oh, I distinctly remember those, yeah. There was a show called pop-up videos. It would give supplement, if you had watched the aha video for, you know, 10 years take on me, it would say, you know, it would draw a line and pop up a little bubble and say, this are, you know, person you recognize from this TV show or this artist was, was this who did this drawing or did you notice that this person has a coffee in the background on Game of Thrones and they shouldn't that little, you remember that scene? I do.

When they had a cup of Starbucks in the, I mean, Westro Starbucks packed. Westro Starbucks is packed. That was a busy filming schedule. I got to say, I want to find the showrunner who made them as day and just be like, it's okay. You guys did great until the last season. Then you did terribly. I just finished, by the way, House of the Dragon. Have you watched it? I've seen a good chunk of the first season, but then I had a little bit of a second. season. I don't want to say anything, but my wife and I really enjoyed it. I want to watch it. It's so good. Also, industry. I told you about industry. Did you watch it yet?

Jason, I have an eight week old baby. I would love to follow up on every content recommendation I get because my friends are very smart. But I'm covered and spit up and crying. I understand. I saw you walk around with a very cute baby on your on your chest. It was very charming. I miss those days. I love I love how it with me, you know? What else do we got in terms of

earnings. We got Amazon I saw. We got Apple. Where do you want to go next? Which one stood out for you? And by the way, if they break up Google, I'm a Google shareholder, alphabet shareholder. Man, I am stoked if they spin out one or two of these businesses because it's going to unlock massive shareholder value. As I've always said, if YouTube was a standalone company, it would be worth hiding double Netflix, given its footprint. I just realized that I didn't tell you the most interesting thing about Alphabet. So I'm going to pull that back up because this is the critical thing. What did I miss? Sundar Pachai during the earnings call said, today, more than a quarter of all new code at Google is generated by AI, then reviewed and accepted by engineers.

That's at Google. A quarter. Now, a lot of this is going to be super background, boring code, not like, you know, cutting new UI. But that's still an enormous amount of code that is starting with an AI Genesis at, as you said, Google, a company that's known for its engineering prowess. It's going to double, and then it's going to double again. I mean, we're obviously going to have robots. creating robots. You know, like imagine a factory of robots.

What you could imagine here is a factory of robots building robots. That's what Google is about to experience. There will be AI agents writing AI agents. And when that happens and you get that flywheel going, that's where people feel that there is an AI risk. And, you know, we saw, we talked about Claude having access to your desktop and being, you know, it can do certain things. It can't do others. can't like make a purchase for you as an example. But obviously that's what we all want. So it's only a matter of time before these agents start spinning, you know, in a, in a velocity that

may be unstoppable. And that's, I think, a 10% chance or 1% chance, but it's one that's worth looking at. And we will look back on that statement as part of that hockey stick of when it goes parabolic, straight up, zip, zip, zip. Mm-hmm. 25% being done, 50% being done, 99% being done, 99.59s, all code in the world. And that code we didn't ask to be built

as being built to do things we didn't ask it to do, and all of a sudden we figure out that we're in a simulation. I'm fine with that final scenario, because if it's all a simulation, then I want to know what my score is. But another company that's doing big things in AI and Cloud Jason, of course, is,

Amazon and Amazon has a couple numbers that I want to just point out to you because they, they blew my mind. So, AWS grew 19% last quarter. Okay. It's now on a $110 billion annualized run rate, which is bonkers. And critically, growth there has re-accelerated in Amazon's phrasing for the last four quarters. That means that it's over $100 billion.

It is incredibly profitable on an operating basis, and it's accelerating. Wow. that's got to be one of the best businesses ever built, period, hands down. It's two and a half times larger than Google's cloud. So Google Cloud is catching up, right? They're at 11 billion. This is 27 billion. And all of this goes back to software eating the world, as Mark and Driesen famously said.

And people are just continuing the build out. And these web services, whether it's Azure, Oracle, AWS, or GCP, are absolutely going to be the beneficiaries here as NVIDIA and big hardware and energy, nuclear, solar, batteries, everything. You're going to need a lot of energy to keep these things growing. And then Amazon has been reducing headcount all this time. They want to do further reductions in headcount.

Google has done reductions in headcount as has meta. So you put all this together, doing more with less static. team sizes plus 20% growth means yum yum on the earnings. And I think this is going to be the bull case for big tech and America. I want to make one more point about this AI business, sorry, this AWS business at Amazon. So one thing that everyone's trying to track, Jason, is how fast is AI as a business growing for both tech companies big and small? And people are very curious, how much is AI demand driving growth at AWS? Well, Amazon gave us a little information

They said, and I quote, AWS's AI business is a multi-billion dollar revenue run rate business that continues to grow at triple digit year-over-year percentage. Got it. And is growing more than three times faster at this stage of its evolution than AWS itself grew. Okay. So the demand to use AI, if it's multi-billion, let's pick a number, three billion. It's three percent. If it was just three billion, it would be three percent of their yearly revenue of over 100.

billion. 3% tripling a year means next year would be 10%, and then tripling again would be 30%. So it is the highest growth sector growing faster than Amazon Web Services itself, which was just to put up web pages and storage of, you know, files. So that is extremely telling as well. It makes sense that these workloads are what developers, startups, and big corporations are musing. And Hugging Face has all these different models.

So every model will be there and you're going to use a lot of compute. And consumers are willing to pay for it, right? Paying 20 bucks for chat GPT is the greatest deal ever. Oh, yeah. But here's my thought for startup founders because AWS, $27 billion in revenue last quarter, $10.5 billion in operating income. There's a lot of margin in AWS. And startups are building with AI.

They're going to the hyperscalers. plunking down their hard-earned dollars or hard-raised of dollars to use these mega-compute systems to do the work for them, to crunch the numbers, but you're paying Amazon so much margin to do the math for you. So I wonder if there's not a case to be made for more startups getting okay at running their own hardware and just not paying. What is a pretty stiff premium to Amazon? Yeah, I mean, the good news is that folks like Oracle and Google will give startups, you know, a really large number of credits. We have, we'll put it in the link here,

but we have a link because Google and Oracle are both sponsors of this week in startups. We have credits for them, and you can check out the new website, get startup credits.com, where I put some of our top deals for startups. Let's fill out the form there, and they'll send it to you at get startup credits.com.

But yeah, this is, um, standing up your own hardware is a bit crazy, but if people charge two, much, then it could become worth it. And we've seen this over and over again, where if people charge too much, people will rack and stack them themselves. But the one thing I will say is because of the credit economy for startups, startups in the first two years tend to get taken care of pretty well, because there's a dogged competition between Google, Azure, AWS, Oracle, and everybody

else to get startups early. And that's a very smart move. So they generally discount pretty heavily. But I have seen this at times where the bandwidth, it used to be the bandwidth charges were what they would kill people with in terms of overages. And people would be just like, well, I'll just put up servers so that I don't have to deal with the bandwidth charges.

And then bandwidth became commoditized. So there you have. I mean, I'm really, really optimistic that the cost of doing AI model training and AI inference will continue to collapse this time. goes on. But right now I just, I worry about the startup that's trying to like build cool stuff and is paying out Amazon or Google so much money and essentially taking venture dollars to selling equity for them and then giving the profits to a big tech company. That's not what I want to see. I want to see the startups crushing the big tech companies.

You know, this is going to be a, you know, this business is one where I think startups wind up benefiting. It's probably still cheaper. And if the good news is it's not just AWS, It used to be just AWS, and the offering from Google or Azure was kind of meager. I would hear all the time, oh, you know, they don't have this. They don't have the, they don't support X, Y, and Z, you know, stacks. And now there's parity across a lot of these. So if you don't like, you know, what your current cloud provider is doing, you got two or three others you can go to. I saw a question come in, and it was on the lower third.

I guess that's a new feature we have here with this live streaming software. Do we want to take a question from the audience? Let's take a question. Team in the back end. Can we put that back on our lower third here? Yeah, we saw that lower there. Come in, here it is. Look at that.

All right, Jason, take it away. Asgard Invest, asks what sector's verticals in AI do they think is interesting for startups? I think going back to what we saw earlier, when there is a pilot that could be perfected, and it's repeatable, man, there's something there. So we have a company taxed GPT that's done wonderfully. And you just think about taxes or legal services,

looking at x-rays, all of those things, or a mechanic. We have a firm MasterTech that you just interviewed Alex. I did. Yeah, they're great. And so, you know, like something like Master Tech is the perfect example of verticalized AI where you have a data set that needs to be cleaned up, a model that needs to be created. and then the accruedramant, the interface, the learning around it still needs to be built up. It's not like you can go into a language model and be like, how do I fix this car?

And you're going to get a great experience for a master technician. But master tech AI, which we're investors in and you just interviewed this week, they are going to build everything around that and keep it up to date so that an actual master technician, just like our guest earlier, just like a master vineyard, of grapes, those kind of things are a great place. So that's why I think, you know, having an ideal customer profile and overservicing them has always been and will always be and accentuated here in this age of AI, the way to go. All right. I'll take this an entirely different direction. Great. I think that humans are incredibly lonely and I think that there is going to be a market, especially for older folks, as our species ages, to have digital friends. So one thing that I'm really interested in is, when did I get to make a buddy?

I want an AI buddy, much like my dog, Maggie is my bestie. Like, we've built a bond over years together. I would like to have a digital friend in that way. It's interesting you bring that up. I've been pitched a number of times recently of people who are trying to figure out how to create companionship or equanimity through journaling or sessions or chat sessions with individuals. So you have better health and, you know, like marketplaces to put you with a coach or a therapist. Then you have chat chippy T over here in character AI where you can talk to a

character from a Marvel movie. But somewhere in the middle, you know, maybe not to deal with depression, anxiety, etc. Yes. Not to role play like Dungeons and Dragons could be what you're talking about, which is, hey, I just want to have somebody who asked me really good questions that I can talk to that, you know, maybe I can release some tension and have like this little AI buddy. I think it's coming. There's a lot to parse there, right?

Because you don't, we had somebody commit suicide. Yes. magically in February, using character AI, they had a discussion. I don't know if you and I talked about it on the show. We didn't actually get to talk about it on the show. It was in one of the dockets, I think, lower down and just there's more than we can get to in a week. But in that pace, I, as someone who has children, my heart literally bleeds for the parents. But I struggle to depend the blame on a chapbot, much as I don't think that Grand Thuffed Otto is the

driving force behind carjackens. Yeah, so in this case, there was an individual who was obviously suffering a young person, 14, 15 years old, I believe. And they were having this roleplay with a character. And then I don't know that the character encouraged them to commit suicide, but it didn't not encourage them. I think there was a misunderstanding. I think that the kid was kind of pull from memory here, everybody.

So if I get this slightly wrong, just understand that we're talking about this. with real gravity and respect for human life. But I think the kid was talking about like going home as a euphemism for suicide. And I don't think the character that he was talking to had any idea about the nuance of that. And so can you say that it was encouraging him to do that and its responses? It's very, very dicey. The biggest takeaway that I have here, Jason, is that my children will never have phones. Yeah, I mean, it's definitely something to think about it.

And these things also have, yeah, so the lawsuit, the dialogue was, I promise I will come home to you. I love you so much, Danny. I love you to De Niro. The chatbot responded, the suit said, please come home to me as soon as possible, my love. What if I could come home right now, sets are continued? I think that's the young adult. And the chat bot responded, please do my sweet king. So to your point, the chat GPT is just spouting gibberish in a random dialogue scenario.

And the child here tragically is taking that as a call to commit suicide. You know, potentially, to blame the technology on this, I think it is, again, as much as our heart bleed for it, it's just spouting gibberish. we were kids, we would go into chat rooms on AOL and CompuServe anonymously. And there were all kinds of chat rooms that came out anonymously. And yeah, it was random and, yeah, quickly became sexual, violent, crazy.

Like, the internet and these kind of chat rooms were insane and deranged in my recollection of my childhood. You know, in 1985, 86, 80s. I was 14, 15, 16 years old. And bulletin board systems had the most deranged, insane, profane chats going on. And kids were involved in them with adults. So a long way of saying you got to monitor kids use on these devices. And, you know, hearts go out to the family here.

That's just so tragic. Why don't we end on a completely different and high positive note? So good news, Jason and I have two clips. you because earlier you were talking about Chinese EVs and all the crazy stuff. Oh, yes. Cool stuff coming. So I had two more clips for you. I went out to the internet. I hunted, I pecked, I searched, I found two things. One, I want to show you a video of the Xiaomi SU7, which is a car they make, Ultra, which is a 1500 horsepower variant of it. You can't actually buy it. But here is a video of the Xiaomi SU7 Ultra going around the Nureberg

ring, the famous huge track in Germany, and I have selected a part of it called The Carousel, and we have the clip, I think, Cued Up already you go, so let's take a look at the Xiaomi going around the Carousel. Whoa. No, you can't, you can't hear it, but it

sounds like a very angry hamster. This is an EV, or this is? This is a pure electric baby. We went around the Nureberg ring in, I think it was six minutes and 46 seconds. And a Tesla had this record with the plaid, right? With special tires, special engines.

These things are absurdly fast right now. But, show me, Jason, makes phones. Oh, they also make EV supercars? Bunkers. I think it just goes to show how many companies in China are involved with EVs, the quality, and the progress. Now, second video, you're familiar with the company Neo-N-I-O, the Chinese EV brand?

Sure. Yeah. Okay. Turns out they have a power, like a battery swap system. And it's not in its first or second or third. It's now in its fourth generation. I want to show you a video of Neo's fourth generation power swap station, essentially a place to drive in, get a new battery, and then get out. Here is the guy going in, and then the computer takes his car over for him, and it backs in, and then he gets a new battery in.

I think this version is three minutes, which is faster. than the one before. Yeah, I remember Tesla had shown this back in the day, like maybe 10 years ago, they had a similar technology. And I think the problem with this is, is whose battery pack are you getting? So these battery packs are worth $30,000, let's say, where they were. Maybe they're worth $15,000 now. Sure. So I'm driving to Vegas. You swap my battery out. You give me a new battery. What if you give me a battery that's got 100,000 miles on it? And mine has 100. Then do I come back and get my battery back?

That was the key issue with these batteries. So I think what will have to happen is these will work really well for robotaxies. Oh, yeah. I think the other thing that's happening as well is when you back one of these in, you can now get inductive charging. So I wonder if this is even going to be necessary. The use case for this is I'm driving across country and I don't want to stop for an hour at a supercharger and have lunch and get a full charge or whatever, 200 mile charge. And, you know, I found when I had my car, the only times I had to even use a supercharger when I lived in the Bay Area was going to Tahoe and going to Los Angeles. So it was with the two drives when I had to stop at a supercharger.

on the way back from Tahoe to the Bay Area because it was downhill, I did not need to stop at the Supercharger with the Model Y. When I had the Model X, it was two stops for half an hour each because the Model X had a 200-mile battery. When I went to a 325-mile battery,

this went away coming back, and it went to, but because you have to go up this giant hill on the way to Tahoe, which uses probably 50% more battery. I had to stop for 20 minutes to top the battery off with the supercharger. because the superchargers eventually got really fast,

you know, 600, 700 miles in an hour it could do. So you would basically get 200 miles and under half an hour. The first 200 miles were really fast. So the use case here doesn't really exist for these hot swappable batteries in my mind, except when we did our Robotaxi overview. Remember we said, hey, six hours maybe of charging a day because you're driving two or 300 miles. Maybe they put a 200, 300, 300 mile battery.

pack in these, then you charge for an hour, then you drive for five or six hours, then you charge for two hours for an hour. You know, that is the case where swapping the battery out wouldn't matter because it's not your battery. It's not owned by any one person. And two, you could get in and out three minutes as opposed to 30, 60, 90 minutes to recharge the full battery. So I do like this for that use case, but I kind of...

Flip it around though, Jason. Imagine instead of buying a car with... the battery, right? Imagine just bought a car and then for 50 bucks a month you would access to batteries. I think that's eventually what could happen here too, yeah, is you could just have your battery be abstracted away from the car ownership. Yeah. Yeah. That's not a bad idea too. I don't need to own a gas station. Why would I need to own a battery? You know, where this really is going to be important is going to be in V-toles. So this same technology, when you're a V-toll, you are using a lot of battery and you have to have the precise amount

because the weight to power ratio really matters, right? Because if you put too much battery in a car, it's no big deal. You're driving, you've got an extra 100, 200 miles. But when you're flying, obviously, the weight is the key issue. Yeah. So I've seen in the V-Tol ones where the V-Toles come in, they land, they get on a conveyor belt, the battery pops out of one side and they slide it in the other, snaps in, and takes off again.

And that's kind of what you want. I love that. I'm a big all of the above guy. I love that we have superchargers in the U.S. I love that we're talking about unified plugs for those. I love battery swapping. I know they're doing hydrogen cars in Japan. Let's go.

Let's just keep pushing this as fast as we can. But that's an update from China. Those rolled out in June, the new neo charging platforms or swap platforms. And there's only, I think, two out so far, which is why I like the clip of that guy who actually went to Guangzhou City to give it a test. the cars coming out of China are bonkers. We have talked about them before here.

They're incredibly cheap. And I think what you're going to see, the cars in India, Pakistan, South America, Africa, just basically the emerging and frontier markets are going to come out of China. They are going to be blocked from being in the United States and the West, just like Huawei is blocked. Why? Because the United States and the West do not want the Chinese to have self-driving cars on our roads that could be used and shut down our economy

if we were in a conflict with China, just like we wouldn't want China spying and also spying. So I know this sounds crazy. We will block all Chinese imports of cars indefinitely because of those issues. Great, great segue. So recently a startup called Skydeo, they're a drone startup, started a consumer now doing business. They just got sanctioned in China because they sold. drones to the Taiwanese fire department. Interesting. I know Skydeo, they've been on this program.

They make drones for public safety. And that's fascinating. I'm trying to get the Skydeo guy on the show for a news record next week. I'm kind of breaking my own work here. But I really want to talk about the sanctions because who has a better perspective right now of the geopolitical tensions between China and the U.S. from a hardware perspective, probably Skydeo. So if everyone works out well, everyone,

next week we'll have the CEO on. And, you know, like DG and the reverse, DGI, I have a DGI drone. Like DGI drones, I believe those are, that's a Chinese company, correct? Yes. Am I correct? You are.

So, yeah, I think this giant standoff with China is going to continue. And it is, in the best interest of the West to not be as dependent on China as we've been. Yes. And everyone agrees with that.

Even companies that are notorious Chinese bulls, like Apple, for example, going back to earnings, every three months, I see a new headline out of their Indian manufacturing, which is like they're moving up the value,

they started to make older models, cheaper models. They used to make the like, yeah, whatever that, SE, they had some names, for the Apple phone that was like for kids or, you know, entry level.

Yeah. But now the iPhone 15, I think, is made in India. And they're doing the pros as well. And so that's the diversification we're seeing. Skydeo, I'll say the rest of that for later on. Jason, we should wrap. We've been on for a while. We've been on for a while.

Three shows next week, maybe. Let's do it. Let's do it. Monday, Wednesday, Friday. Coming at you. 12 o'clock central 1 p.m. East Coast and 10 a.m. on the left coast. if you have ideas for stories you want, Alex and I to cover, hot takes, guess, whatever it is. He's Alex on X. I'm at Jason. So you just add Jason at Alex. Tell us you saw the show. Tell us what you like. And also Alex and I are in the comments on YouTube interacting with you. Coming at you. See you next time. Bye bye.