Tech Earnings Recap, Beautiful Data Centers | Mike Isaac, Kyle Harrison, David Marcus, Turner Caldwell, Everette Taylor, Puneet Mehta & Justin Wexler, Alexander Embiricos

You're watching TVPN! Today's Friday, May 1st, 2026. We are just barely live from the TVPN UltraDome, the temple of technology, the fortress of finance, the capital of capital. We had to rush back from Stripe Sessions. Fantastic interview with the Collison brothers, many others yesterday. Go check it out if you haven't already. Fun stuff. We didn't go nearly deep enough on the whole Kocian singularity. There's a debate on the timeline about it. We can get into it later. We need to have some more economists on the show to break it down for us. But this is their big pitch. Firms get smaller and the future, the Coase theory of the firm. Anyway, if you studied economics, you might be familiar, but it's jargon and you're going to be hearing a lot more of it.

A whole lot of mumbo jumbo.

I'm not even going to try. I'm not going to try with Jordi. Well, here's some, here's some less mumbo jumbo. Tech earnings, quad kill recap. 4 in 1 day. Last time this happened, I think it was 2020, 4 big tech companies all same day. It is random. No one really knows why. It just sort of lines up. It's like a solar eclipse.

Can you use astrology to predict the next one?

Potentially, potentially.

You would have to imagine it's something around the alignment of the planets that causes the retailers.

More like the holidays, more like the holidays because books close, then certain things line up. Also like Nvidia is never in the competition. Yeah, yeah. They're always like 2 or 3 weeks later. But anyway. Well, let's go through what actually happened. Google was the big winner. They absolutely crushed. The stock's up 10% in the last couple days. Really successful. And there's a few reasons. So the core business is still growing. Google Search, this was the question of, is AI, are chatbots going to be eating into the core Google Search revenue, the core business that drives all the other investments? The answer is unequivocally no. Google Search and other revenue was 16. $4 billion, yeah, up 19% year over year. I mean, 2 years ago, there was the whole like Google search is dead thesis. That has not been borne out, at least this year. Very, very strong results there. Then Google Cloud is like the major story, so much so that Tae Kim, when he did his earnings recap, he didn't even talk about top lines at Google, Amazon, and Microsoft. He just says Google Cloud revenue 63% year over year, Amazon Web Services 28% year over year, Azure, 40% year over year, and then Meta overall revenue, which we'll get into, 33% year over year. So he's not even looking at the rest of the business. All that matters for the three major public clouds are those cloud businesses, not even anything specific within the overall business. So long story short, higher CapEx at Google, totally justified in a world of cloud. Acceleration. Backlog growth is huge. Uh, cloud backlog, uh, nearly doubled to more than $460 billion. And importantly, this isn't like a 5-year, 10-year contract type thing. Half of that, more than half of that backlog is expected to be recognized in the next 24 months. So really good news there. Uh, Google Cloud hit $20 billion, up 63%. Now it is smaller than Amazon, and that's why AWS is, uh, is, you know, people are hunting for high 20 growth rates, maybe, so it begins with a 3. In Google, it's smaller, so it can grow 63%, but that's still fantastic. And cloud operating income was $6.6 billion, which is very strong margins. So everything is looking good. Search durability, demand for AI infrastructure, it's the full-stack AI play right now. Microsoft showed solid execution, but they didn't dramatically change any particular narrative. I think a lot of this goes to like the just naturally slow deployment of enterprise AI. But this is the focus for Microsoft. They are the enterprise AI play.

Yeah, one thing was notable is they weren't heavily teasing anything related to the next Gemini release.

Yeah, they wouldn't tease that at earnings.

No, no, they have historically. So anyways, maybe they wanna let the model speak for itself.

Yeah, probably announced Gemini 4, I would imagine. Announced it or a coding-specific model. Yeah, yeah, something there. I mean, there's certainly demand for it. They have the capacity. There's lots of opportunity. We'll probably be following that in 3 weeks when Google I/O happens. So Microsoft investors wanted to see Azure growth, Copilot adoption, and a solid justification for all the CapEx. The stock's down 2%. Nothing crazy, but Microsoft beat on headline numbers. Revenue was $82.9 billion, up 18% year over year. And from an analyst perspective, it was a clean beat, but there's more nuance here. So in terms of AI adoption, Microsoft added about—

Look at this beautiful LED wall.

5 million.

Yeah, let's give it up for that.

Still underrated. We ordered this bad boy maybe like 6 months ago.

6 months ago.

Took a while.

Took quite a while.

But it is fun. We need to figure out—

Worth the wait.

More things to do with it. So on Copilot adoption, this is the big question. So for Microsoft 365, like total seats, you know, you think about Microsoft Teams, Outlook, like the standard Microsoft enterprise seat, I think it's about $30 per user per month, competes with Google Apps Enterprise, right? But Microsoft's been doing this for much longer. They have 450 million paid seats. So, In terms of Copilot, they added 5 million, which is great. They're at 20 million now. So that's solid adoption, but it's still small compared to the 450 paid seats. Ideally, like every seat would have a Copilot alongside of it. That, I mean, that's the idea, right? And so the question is how quickly can they get that 20 million Copilot seats up to 200 million, 400 million? You know, ideally everyone has one of these add-on subscriptions and that's lifting the overall business. Market reaction was a bit choppy as the new OpenAI relationship gets digested. And so there's two sides to the Microsoft OpenAI deal right now. Of course, OpenAI is no longer limited to selling just through Azure. Azure no longer has exclusive access to OpenAI models. OpenAI is now available on AWS as well. And it's— and it cuts both ways. So on one hand, Microsoft they used to be the exclusive provider. If you're a sales guy at Microsoft on Azure, you're probably really excited to call somebody up and say, hey, you wanna use GPT-5.5? We're the only place in town, like you should come over here, you should migrate more of your infrastructure, you should be using Azure. But it was probably frustrating because a lot of people say, no, we're locked into AWS, we are stuck in that cloud and we'd love to, but we'll just use the API directly or we'll figure out some other workaround, but we're not moving for this.

Yeah, or people could be like, yeah, we're open to trying it, but it'll, take, you know, a couple quarters realistically to get ramped up.

Yeah. And so from an Azure perspective, losing exclusivity is a negative, but from a business perspective, the more growth for OpenAI is good for Microsoft's equity stake in the company. And so these two tensions, like the capital balance sheet side of the business versus the Azure sales and acceleration side of the business, like these two—

Not to mention Copilot versus Codex.

Sure. Yeah. And so these two sides of the business are sort of butting up against each other. And Satya Nadella ultimately has to go and, you know, renegotiate the contract. And they did. And it seems like they're in a good place because a lot of people were sort of saying like, oh, well, is this going to get messy? Is there going to be some lawsuit here? It seems like there was a very clean negotiation, renegotiation, and OpenAI got rid of the AGI clause that allows them to stop sharing models. But they are sharing revenue and there's like, like the overall deal.

Yeah, basically the, I think 2032 end date.

Yeah.

Is just the IP share instead of this like arbitrary AGI.

So over at Amazon, Amazon still of course the CapEx king. They top the CapEx forecasts. But as scary as that would be, AWS is reaccelerating. And so that's very good news. So the stock moved slightly upwards. Q1 2026 sales for Amazon overall were $181.5 billion for the quarter. For the quarter there, they're like on track to make a trillion in revenue soon, up 17% year over year, beating expectations. AWS is expected to see 25% growth. That was the expectation, 25%. That's sort of where they've been. They came in at 28%. So they beat on AWS growth and that's really, really important, especially for their scale. This was very good. Big news. The ads business is still cooking, churned out $17.2 billion in revenue, and the chips business just crossed $20 billion run rate, which is very, very good. And they have big deals. AWS is in this interesting position, less of this like Google pure play full stack and more focused on working with both OpenAI and Anthropic at this point. And so that all justifies more investments in Trainium and AI infrastructure broadly, and the market seems to like it. Huge CapEx numbers, but reasonable response from the market and the stock. So, Meta. Meta had a rough go. Huge drop, nearly 10% down following earnings. It's sort of jumping up and down now. But the core business was extremely strong. So Q1 revenue was $56.3 billion, up 33% year over year. So, high growth for the overall business. Ad impressions rose 19%. Average price per ad rose 12%. They raised the CapEx outlook from this. They have this range. It's within $10 billion of $125 billion. They added $10 billion to both. So instead of saying, hey, we'll be between $115 and $135, now they're saying we're going to be between $125 and $145. And now, will they actually get more compute for that, or is that just a reflection of rising prices for all the inputs that go into compute spending? That's a big debate because there's one bull case where, oh, they're optimistic, they're going to buy more compute. But it's like, no, there's a possibility that they're just spending more to get the same amount of compute as they thought. But the big problem for shareholders, the big answer is it's not as clean of a story.

Can you imagine, just take one second and imagine that Meta's a private company and their pitch is that we make AI agents for selling products and they're doing $200 billion, they're at a $200 billion run rate growing 33% a year.

Yeah.

How's it price? $20 trillion?

Probably. Probably.

With those operating margins?

Yeah.

I'm joking. It truly is one of the most incredible businesses of all time.

But they don't have the cloud.

But people are— yeah, yeah, they don't have the cloud thing. People are— he's kind of acting like the most AGI-pilled CEO, like hyperscaler CEO, right?

Yeah.

He was talking about like basically talking about like RSI on the earnings call. He's like, it's not just about like coding, you know, coding. Like he really seemingly believes and he doesn't want to be in a position where someone else has actual superintelligence and he does not.

Yeah.

And so, and that's just like a little bit too fuzzy because when you look in the rest of the market right now, it's like, productivity tools, right? And so the market just like can't really read into it. And they're just kind of like not even giving him full credit for the business that he has.

Yeah, and I forget if he's ever mentioned the idea of selling tokens, selling AI infrastructure, getting in the cloud market, but it's way easier for an AWS.

He kind of loosely mentioned that last year when he was like starting to ramp stuff up, like basically saying like, maybe we could do this. At some point.

Yeah, this is way easier to justify the, the CapEx at Amazon when you say, oh well, like, you know, if, you know, regardless of how this market plays out, maybe Anthropic takes some of the compute, maybe OpenAI takes the compute, maybe there's a whole bunch of new AI labs that take compute, maybe we'll use it for our services, our own models. If models commoditize, it's still valuable. If they don't, it's still valuable. There's a a whole bunch of different ways to shift that CapEx around and justify it. Meta, it's gotta show up in the AI figures, it's gotta show up in the ad business, at least right now. And so there are questions about how the big CapEx spend feeds back into cash flow when the company doesn't have a platform with big enterprise AI contracts, massive RPO figures. There's not an easy place to just like stuff compute necessarily and at least get, a market standard return on investment. Of course, they could spin up reselling, they could become a NeoCloud or something like that, but at their scale, it's a whole new host of challenges if they go that direction, and I think that that's not the plan. I think you're dialing into it, that they are in fact recursive self-improvement pilled at this point. And so the other sort of tremor in the system was the fact that daily active people, which is something that they were, it's not users anymore, DAP, daily active people, uh, declined sequentially since the first time that Meta began reporting the metrics. So less people were using Meta platforms, uh, this quarter than previously. Um, but they had some good explanations for it. They said that there were internet disruptions in Iran and WhatsApp restrictions in Russia. And so this was not representative of a real change in internet user behavior. And I think we all recognize that that is reality in the sense that there's not some up-and-coming social network that's eating Meta Platforms family of apps lunch, right? Like the TikTok threat came and sort of went. The Snapchat threat came and sort of went.

Sora.

Sora came and went, right? And so there was a different post I saw that social media use is declining, especially under young people, but it was already so high, not really showing up on the data.

It will truly be insane if we get this new alien technology, we get LLMs, which in digital form can convince you that they're a person. Yeah. And we end up getting no new big social platforms. Like—

Might just strengthen the current ones.

Well, yeah, no, I mean, I think that's the base case right now, but it's still wild. I think you would expect, I think again, if you rewind, to like 2020 and you said, hey, we're gonna have this technology where you can effectively have a digital twin of yourself, you would think that somebody would figure out some new use case mechanism.

Yeah, maybe, I don't know. I continue to think it just strengthens the incumbents, Roblox, Instagram, WhatsApp, Facebook. Like these platforms will just get more content on them that will probably be LLM-generated, LLM-enhanced, even if it's not LLM-generated, them written or developed. It's, you know, a creator taking less time doing other boring things and spending more time on their core output. So we'll keep tracking it. So overall, what did we learn this earnings season? One possible takeaway is that there's a little bit of a fracturing in the overall AI narrative. So the AI narrative over the last 12 months has basically been like, say the biggest number, biggest CapEx number, biggest deals, just grow, grow, grow. Now every hyperscaler reports a big CapEx number, but each company does have a different story. So Google's the full-stack platform, Microsoft's focused on enterprise adoption and distribution, Amazon's most aggressive on infrastructure and partnerships with OpenAI and Anthropic. They have the cleanest, less, it's a less competitive, relationship with Anthropic and OpenAI because they don't have the DeepMind equivalent necessarily. Meta is all about ad optimization, and Meta is also this interesting, like, high-risk option on frontier AI. Like, it's possible that if TBD Labs, MSL, if that pays off, you wind up getting an OpenAI-sized or Anthropic-sized business out of it in the oligopoly world where they have a breakthrough, they get to something that's 5-5 4-7 level, and then they're competing at an equal level and they're playing in that oligopoly. And so, like, the Meta traders are sort of, or the investors are sort of weighing both of those sides. You have a very solid ads business, and then you have this call option on potential Frontier AI, which is still early, but showing some really solid signs. So, the market likes short-term revenue evidence right now, And there's still many strategies that will play out over the rest of the year, but this quarter was focused on, can you justify your CapEx this quarter? And so we'll keep following it. Let's go over the timeline and see what people were saying. The headline numbers, of course, were huge cloud growth. AWS 28%, GCP 63%, Azure at 40%. Absolutely amazing. Massive numbers. Remember, people were hoping for GCP to be at 40% and it went up 63%. Absolutely massive. AWS, people are hoping for 30, they got 28. Not too bad. TakeHim, we mentioned you can go subscribe to his Substack at takehim.substack.com. He has a write-up on the state of AI after Google, Meta, Amazon, and Microsoft earnings.

Tae Kim's been right a lot lately.

He's been doing well. And there's some people reflecting on, you know, crazy run-ups, crazy numbers, crazy hype. Is this the dot-com bubble? Always fun to go back and comp to previous bubbles. The dot-com one is still fresh for a lot of people, because some people in tech were alive during it. Many people in tech were alive during it. This poster, Ren, is comparing the price-to-earnings multiples today versus the dot-com boom. So today's Mag 4 PE ratios, the companies that reported earnings, Meta's at 16x price-to-earnings, Google's at 17x, Amazon's at 24x, and Microsoft's at 25x price-to-earnings ratio. Compare that to during the dot-com. Microsoft was at 73x, Cisco was at 200x plus. Yahoo was trading at 800x earnings. The NASDAQ as a whole traded at 200 times earnings during the peak of the dot-com bubble. Today's bubble is trading at 16 to 25x earnings on companies generating hundreds of billions of dollars in real cash flow. Now, they're drawing down on that cash flow, but even in some bizarre world where, you know, AI is a bubble and it's fake and we all go back to pencil and paper, we'll probably go back to do digital advertising and spreadsheets in the cloud and SaaS, and all of these companies will still benefit and be in fantastic positions.

Yeah, the only thing that's— or one thing that's imperfect with these comparisons is when you look at the peak, you can say, like, OK, Meta, Google, Amazon, and Microsoft have relatively reasonable PE ratios. But if you look at the rest You look at all the companies that have been started in the last 5 years, especially in the private markets that have been marked up from 20 to 100 to 500 to a billion and beyond, many of them have absolutely no earnings. And so you have to look at maybe something more like a Yahoo as a decent comp, right? Where Yahoo was trading at peak at 800x.

And there is some pushback in the replies here. Reasonous Four says, AMD's at 130, Tesla's at 350, Palantir's at 220, and Intel is at 900 times earnings.

I know. So it's very, very convenient to just take the 4 of the greatest companies ever in the history of the world and say, there's no bubble. They're reasonably priced.

I think it is useful to go back to the fact that there's a lot of hype in tech, but what's driving the vast majority of the market cap right now is still fairly low PE companies. Intel is so much smaller on a market cap basis than Microsoft, Amazon, Google, Meta, that even if you're overpaying on earnings for that, you the overall market is anchored to a pretty decent earnings engine still. And that's like the bull case here. What else is going on?

I'm excited. [Speaker:JAKE MERL] George Hotz is back with a white pill.

[Speaker:JASON_WILLIAMS] What'd he say?

[Speaker:JAKE MERL] May 1st, he says AI will create jobs. It's nice to see Jensen talk about this. And it's super obvious when you think about it. AI and immigration are fundamentally the same. There's new people showing up and hopefully everyone understands how and why immigration creates jobs. Wants are effectively unlimited. It's classic Jevons paradox that if we make something more efficient, we end up using more of it. Or a cool aphorism I learned at Facebook, if you make the site 10% faster, people spend 5% more total time on it.

Isn't this funny? He worked at Facebook for—

he was like an intern.

He was part of the machine. Now, just like you get the wah-wah crying people about how immigration lowers wages for native-born Americans, and we gotta keep the hardworking immigrants out because you have Some right to be lazy or something. You'll get this about AI. AI will outcompete some humans at some jobs, but protectionism is for losers. The important thing is that the overall pie grows and inequality stays somewhat in check, not by redistribution, but by design. There will be more to do than ever before.

More to do than ever before.

We pull up this video of Jensen.

Yeah, let's do it.

In the production chat for you.

Yo, you have it?

Yeah, we can play it.

Great. I'm excited for Eric Suford's deep dive here. We'll get into that later.

The Sufornator?

And reminder, we have the Isaacinator coming on in just 7 minutes. Mike Isaac has been covering the Elon Musk versus Twitter trial.

Is it like a 4-day work week with trials?

It is.

Are they living in the future?

It is. Wait, wait, wait. So can we pull up this screenshot, because did you read Mike Isaac's full thread on his experience covering the trial? Did you read the whole thing? 'Cause he's been live tweeting it, and it's very funny because he will post, so we gotta zoom in here. This is a—

Did you turn this into an infographic?

I turned it into an infographic. So this is a minute-by-minute chronological log of off-topic/personal, no-slash-minor-inconvenience moments because Mike Isaac will talk about, oh, Elon Musk just said this on the stage. Oh, the lawyer fired back with this. The jury said this, the judge said that, right? But then in between that, he has been logging very, very minor inconveniences that he's experienced. So he talks about his lunch, the roasted tomato and corn pizza. He says he forgot a pillow for his back. —for him to sit on.

Wait, corn, beans? I don't know.

He said he bought— That's a very Oakland thing. At 16:07:45, this is transcribed from X, so I think the times are UTC or something. He brought a water bottle, but forgot to fill it in the rush for a seat. Now he's increasingly thirsty during an intense moment of testimony. Someone's laptop suddenly blasted a loud YouTube video in the gallery. This is a garden hose nozzle, it said. Courtroom mishap. So he's going through it all. It's very fun to track his— and we'll dig into his experience both in the court and actually with what's going on with the case. But let's play this Jensen clip. This sounds fun. Jensen is insisting that jobs will be created in the future. Let's see it.

What does he have to say? There's the things that— the things that are said are very counterproductive and in fact hurtful. On the one hand, maybe— Maybe who he's talking about. Maybe a scientist thinks that by warning people that AI is going to completely permeate and proliferate across radiology and therefore radiologists are gonna get wiped out. On the one hand, that might be considered warning and therefore helpful, but in fact, the counter would have been hurtful. If we convinced everybody not to be radiologists and we now need radiologists, that actually is hurtful to society. It is hurtful if we convince all the young college graduates to not be software engineers and it turns out the United States need more software engineers than ever. That's hurtful. And so we have to be mindful of how we communicate the importance of this technology and what it's able to do. To advocate for policy and advocate for guardrails on the one hand. On the other hand, scaring people with things like saying nonsensical things which are not going to happen, that this is an existential threat to humanity. There's 20% chance that is existential. That's ridiculous. That it's going to wipe out 50% of new college grad jobs. That it's gonna completely destroy democracy. I mean, these kind of comments are not helpful. They're not based, they're made by—

[Speaker:JORDI HAYS] He's subtweeting in real life. He's not vague posting, he's subtweeting. I do think the 50% numbers are so funny because it's like 50/50 is like the most neutral you can be about some prediction, it feels like. I mean, I get that it's not, in that case it's like 50% of of jobs, but whenever you issue like a 50/50 proclamation, it sort of tells you nothing. Like, we were looking at like the prediction markets on the trial and we were like, "Oh, it's 50/50. Okay, so that actually doesn't give me any information about what's going to happen in the future." And it's very easy to make a prediction that's like, "Oh, I predict this will happen, 50/50." And then if it doesn't happen, you can be like, "Well, yeah, 50% chance it wasn't gonna happen. Like, I had it really high that it wasn't gonna happen." And then you can also say, you know, I predicted Yeah, it's so, it is truly universal when we talk to these companies that are building AI products or selling AI products.

It seems like the ambition level in every single industry is just going up. Even Gabe from Rogo was saying, yeah, there's an opportunity to cut costs or there's an opportunity to get aggressive and like take market share. And he said, like a lot of firms are saying, like, let's just do more work. Let's grow our revenue. Let's take on more clients, right? And so, it's harder and harder to doom about the job market.

Yeah. You need to doom about the task market. It's over for tasks. Tasks are getting automated. No, it's serious. Like, there are tasks that, like, The reading of the X-ray or whatever it was, the radiologist scan, what do they actually read? Is that an X-ray that they're looking for tumors in? You know what I'm talking about, right? The famous Hinton— Yeah, I don't know if it's exactly an X-ray. It's not, right? It's a CAT scan or something? It's some scan where you can look at— Yeah, they do a scan. And of course, image generation and image models are very good at detecting cancers. But that is just one of several tasks, many tasks that a radiologist does throughout the day. And so radiologists are currently making more money than ever and they're in more demand than ever. And I was telling you before the show about truck drivers. They don't just drive the truck. There was a survey that said 70% of them carry weapons because they are effectively, they are security guards for the truck, as well as probably a bunch of other things that you don't think about, light repairs on the truck, you know, stopping for different things. Doing different things, rearranging things in the back, making sure that the load is balanced. It's not purely just—

If you're driving with kids in the car and you drive by a big truck on a road trip, you can go like this.

Yes. And then— That's another thing that the self-driving truck needs to do.

A computer could never look out the window, see some kids in the car, and honk.

But, well, yeah. I mean, we'll see. Anyway, we have Mike Isaac in the waiting room. Let's bring him into the TVP and Ultradome. He's at the New York Times. Mike, how you doing? Yo, what's up?

Good to see you. Great to see you. So wait, so what's up?

Yeah, so 4-day work week, are they living in the future? Yeah, is this universal basic vacation days? It's—

court is incredible in that I got to sleep in till like 9 AM today, which is— which was great. But I made up for it because like literally every day this week I've gotten up at 5 o'clock in the morning to get my cold ass down to the Oakland courthouse and stand outside for 2 hours. So I don't know, but it's fun.

How does it work? Who gets priority? Is it first come, first serve? Like if some random person shows up before you.

They can just get it and then you're just out.

Or do you get to like show some sort of like press? Like how does it work?

Yeah, I wish I was cool enough to like cut. Well, here's the thing.

You should get an artist pass.

VIP pass? Yeah, right. No, not VIP.

Artist pass is a level above. Okay, artist pass.

So there are 20 reserved seats in the front row for press, but the issue is only one person per outlet gets it, and we are doing like live blogging for like the big moments, like opening statements and for Elon. And so myself and my colleague Cade came on the first, the whole first week. And so we've had to trade off. He's gotten the like press skip the line thing and I've been with the other folks in the 30 unreserved seats in the back. That are watching YouTube videos. Yeah, yeah. Oh my God.

Or literally watching—

Or Stripe Sessions. Literally one dude fell asleep. I was actually impressed.

I'm asleep. Yeah. It seemed pretty entertaining. Didn't seem like you were falling asleep. I was keeping up on the, through the live blog. Be reading my nightmarish Twitter too.

It's been fun, honestly. Have you guys ever done like court case things or have you ever been sued or been in court? Never been in court.

I went to mock court in high school where— everyone picks a role. I think I was—

Yeah, we LARPed in court. Literally LARPing in high school.

But other than that, never. I— the whole—

this whole time, it seems— it seems insane. Like with the recent NASA, the moon mission, I was like, this should be a livestream plus a pay-per-view for the key moments. Yeah. For this trial, it's like we have huge budget deficits and we have these incredible media products. Why are we not doing pay-per-views, right? But that's the funny thing.

I was just gonna say, that's the funny thing about federal courthouse stuff. There's a lot of different rules around filming and electronics. I covered some cases in DC, and I can't even bring a laptop or phone in those courthouses. So this is actually a very permissive judge just because she believes in press access and stuff. And we've had There's way more access than you would normally get in federal court cases.

Okay, so you've been to these court cases before, but if I follow your Twitter, it feels like you're making a bunch of rookie mistakes. You only had egg bites at 5:30 AM, your bite energy's wearing off, you forgot your butt pillow, you forgot to fill your water bottle. Like, is this amateur hour or are you a professional?

What's going on? I think that you guys might know me well enough that that is kind of how I operate most of my life, like kind of chaotic. But it is— Focus on the important stuff.

That's right, yeah, yeah, yeah.

I mean, look, I can file. I may be like bleeding and hungry, like limping across the finish line, but we're getting, I'm tweeting for you.

What is your, what are some like high-profile cases that stand out that you've covered in the past?

So I got to do, let's see, my first one was insanely boring but important, the Apple versus Samsung thing back in the day and like them, suing for Samsung copying like literally everything they do. Um, I did, uh, Dallas actually. Zuckerberg's like, I think maybe one of his first testimonies on the stand when ZeniMax was suing Meta for the Oculus acquisition. You remember that? Yeah, that was in Dallas. That was super fun. Uh, that I almost got kicked out of the courtroom for tweeting. Whoa. Um, and then I did, uh, the DC FTC Meta one, and I almost got kicked out of the courtroom again. Actually, I I did get kicked out of the court for wearing the Meta Ray-Bans. These are not it. But I wore the Meta Ray-Bans. And then they started putting, I was fucking super stupid for doing it, but then they started putting signs up saying, "Do not wear these glasses in court." But you weren't, you were just using them as glasses or you were being sneaky and you were recording?

No, I was like, look, I can't record, I won't record.

I was trying to play by the rules and like they were my prescription glasses.

Yeah. Oh yeah. But the bailiff, The bailiff was like, it's too much, it's too high risk for them because you could just, you could have like turned off the, the light or whatever, something like that.

Uh, talk about the, the fans. Are there really Elon Musk fans in the courtroom? Like, what, what motivates someone to go and watch that live? Is this their UFC front row ticket? Like, why are they there?

Uh, so you guys would have fun. Like, it actually is A lot of court cases are boring to people who don't care about this stuff, right? Like, you and I may be super into— like, the FTC trial was super fun for me because it's like, oh my God, Mark Zuckerberg emailing Sheryl Sandberg and talking about Path. Like, this is incredible. Yeah. And like, the average person has no idea what we're talking about, but this is like a circus. There are people who genuinely love Elon or are genuinely worried about the end of the world happening. And I think it's a really good thing that there's public access to these courts. Like, I think like the average person can come in and show up. And that's what I think after the buzz of Twitter and like people seeing that this is an event, we got much longer lines and like folks who are local, like just like I know a PM in tech from Meta that came. I know like a guy from Box made it in. Just if you get a seat, If you get there early enough, then you can get a seat and you can just hang out. It's like, and I think it's really great. I think it's great that people are, yeah, are there for it. You know, it shouldn't just be me.

Yeah. What about the, what about the jury? Does the jury seem excited and honored to have the opportunity to be a part of a case like this? Or are they nodding off? It'd be so funny to be like so out of just like, off the internet.

You can't be super biased.

You're probably not a PI. Sam, this guy Sam and Greg, and this guy who makes cars, and I don't even know what they're talking about. A whole bunch of mumbo jumbo. Like, there's got to be one person on the jury that was just so not tapped in that they're just confused. Yeah, I think jury selection was super interesting for that.

I can't say— it's actually interesting, I can't say too much about the jury right now because like there's all these rules about— oh sure, like some random person could go up to them if you identify them or try to alter it, something like that. Totally. But I will say, like, during jury selection on Monday, it is an incredible slice of life, and you get like how familiar or unfamiliar people are with the tech industry despite being here, you know. Like, some folks are like, I have no idea what you're talking about, I don't know what I don't know what AGI stands for. So it really, it really played, and voir dire and jury selection is so important for cases like this. It just, it makes the dynamic of the facts don't necessarily always matter, but the vibe can really matter, which I think is a benefit for Elon, honestly.

Yeah, were they trying to weed out, they're trying to weed out people that are a little bit too excited about the case, right? You want the people that are like in order to have a— but, but then talk— I didn't understand why is the jury just like giving like an advisory decision? What is the history of like, why do you, why do you have a jury when the judge is ultimately going to make the final call? It feels like just kind of putting on like a show because like theoretically the judge could just sit through a bunch of depositions and make a call? Yeah, sure.

No, I, I think, um, so I do think they want, more often than not, want a jury of these CEOs and companies' peers to be the deciding factor in what they feel like is good for a civil claim. Like, that I think is fairly standard. But to your point, um, the judge can throw out their verdict, which is like, is— and judges I don't think tend to want to do that because like they want to have reliance on this is the public, the public should have a say in what goes or whatever. But the judge can do that. I will say also she is responsible for if they are— if Elon or sorry, if OpenAI is found liable, judge decides on remedies, damages and things like that. So she still has an active role in that regard and in steering the case. But I really do think that Courts often prefer or often appreciate a peer, a jury of your peers making some of these decisions. So I don't think it's gonna be like completely disregarded is what I would say.

[Speaker] Yeah. [Speaker] How has Judge Rogers done so far in your view? Just reading the live blog, she seemingly has like zingers, pretty like good, like one-liners at times. I'm just like kind of imagining what it's like in there, because obviously I'm just reading text, but she's had like seemingly like some pretty good comedic timing. Yeah.

Oh my God, she's so funny. She's like real— like, as you might imagine, there's like a number of different types of judges and how they handle their court or whatever, and she just takes no BS from anyone, including the lawyers. And like when they try to like tap dance or break the rules or whatever, she's like, no, shut up, or like, get back on track. Or no, no, no, actually the best part or the most insane part. So one woman in the overflow room who is just, not just, it was a civilian going to attend and watch it, started recording, which is again against the rules, if not the law in a federal courthouse. So the judge brings her in and in front of a room of like 100 people, just like dresses her down, yells at her saying, did you not see any of these signs? What are you doing? I will kick you out. I will, it was like, I would've like peed my pants and started crying if she had done that to me.

It was deeply, deeply intense. Yeah, it's teacher, you know, berating a student. Yeah, class is in session. Class is in session here. It was brutal. How have you been processing Elon's position positioning. It feels like the two stories that I've heard him sort of telling are one about his commitment to technology, humanity, saving the world through Tesla and the electrification of the internal combustion engine and SpaceX making humanity multi-planetary. And then, and that's like a very high-level high concept pitch. And then he also sort of brings it down and starts beating this drum on like, you can't steal a charity, you can't steal a charity. Is that the correct framing that he's trying to go like high and low there? How much have he been beating each of these drums and how useful is that?

So I totally agree. That's the framing. And I think this really goes to the point of a lot of these trials are Pageantry is the wrong word, but let's say theater in that you are— this is a jury trial and it's a different thing than just convincing a judge who I would say is much more attuned to the facts and merit of the case. And like hammering in on the evidence and like something that may be boring to you or me or whatever, or the jury is gonna be more important to a judge. Elon, I think from day one has leaned into the persona of Elon and just from him being on the stand and saying, I care about humanity. I mean, he says he does whatever he does or doesn't, but like just leaning into this, like, I'm a world-changing entrepreneur and this stuff is existential and I'm the one who has cared about it. And like, that may work on a jury, you know, like there are people who still love him. And, you know, OpenAI is really hammering the facts of what they feel are their side of the case and saying, Elon has known about, uh, Elon has never been in the dark. He quit in a huff. He, uh, he's made it very clear he hasn't been there. He's trying to sue now, or he's trying to file this claim now because he's catching up, because he's behind as an xAI competitor. But like, again, this is all stuff that maybe it doesn't play— like, this is why jury trials are so risky for companies a lot of the time, you know. It's really— it's fun for me. It's probably not fun for, for everyone in there, but it's fun to to see it play out, if that makes sense.

How did the distillation comments come up?

That was like, that was the news of the day yesterday. You guys obviously were attuned to it.

Which you broke it, right?

I think I did, yeah. I think it was one of those things where I was like, holy shit, this is news. And I think like folks like started figuring it out, but I was like, gotta put it on Twitter. And so, The point— so Bill Sabat is lead counsel for OpenAI. He was sort of talking about— the point was made in the context of they want to hammer home Elon is creating a competitive product. And also he keeps stressing that this is doom, world-ending technology. But at the same time, he's ripping off the TOS and using the technology to improve his own technology in an explicitly for-profit company. And so the point is like hammering him as a hypocrite. And I think, look, like, I think Elon has history in his businesses. It would not be controversial to say, to like say one thing and do the other. And I think he sees it as like rules of engagement. That's the game in the, you know, no-holds-capitalism land. But like, that is how it came up. And that they didn't go super deep into it, but they wanted to make the point he's using OpenAI's tech. He's breaking the toss and he's partly distilling it. And I want to say also that Elon went out of his way to say everyone kind of does this. It's like an open secret in the industry, which I think is like also kind of true, but that doesn't make it— there's like, it's fraught, I guess, is what I would say.

Yeah, yeah. It's a— there's obviously a continuum there, which is why he tried to hammer that it was like partly. I imagine most car companies Have taken a rival car for a spin? Have they taken the car apart? You know, there's a line there and there are laws, but that's a separate issue, of course.

Has anything come up with a jury around prediction markets? Because I was thinking we've now seen insider trading across every possible prediction market, right? From, you know, I'm sure from like— Man, that's very interesting. You know, even the Maduro thing was interesting because a lot of people were like, oh, he's just, betting on himself. But then I saw that, I was like, hey, that's sending a signal to the entire world that an attack could be happening, which puts your entire team at risk, right?

So like, very, very clearly, like, U.S.

military personnel should not be able to trade against our own military's actions ever, right? And that they need to come down really hard on that. Are you— but a jury—

a jury like on this, are you just trading on this crazy—

are you, are you going to hierarchy. The reason, the reason that the jury thing is, is, is bad is it creates potentially an incentive for the jury to basically work together and say like, hey, like we all have to be here for like a month, like we could at least make some money on it. And then you only need to get like 4, you only need to get like 4 or 5 of these people.

The chat is saying monetize jury duty.

Oh my God. No, but like, it seems like very important that this does not happen, right? Because there's very, like, they tried to select a jury that just doesn't care about the AI race. They don't care about this or that. And, but if they are self-interested in some capacity, they could be like, well, my decision is not even really legally binding. It's just advisory. Like, I may as well. you know, I don't know, right? I think that's a great point.

And like something that I imagine like court systems aren't even prepared for fully yet because like, well, there hasn't been a big, there hasn't been a big trial.

Like there's been, I'm sure some like Epic, Apple stuff that was slightly big, but nothing where nothing, nothing, anything at this scale with this much, even just this much volume already from people that are, generally interested in, in, in the story.

Yeah, the outcome. And like, I think the other thing you should know is like, they're not sequestered. Like, they show up kind of like before, like anyone else, right before they— yeah, like they literally— like, we see them walking in and that the marshals— there's like a ton of like US Marshals there. They're like, is anyone a juror in this line? The juror gets to go in. So let— and at the end of every day the judge is like, goes back over the rules saying, do not discuss this case with anyone, do not watch it on TV, do not look at it on your phone, which is like a verbal command, right? Don't go on your phone. Just don't. Exactly.

Hey, don't use the most addictive thing that has ever been created in human history. Don't use the thing that you use for 6 hours a day.

That you probably turn your lights on and adjust your thermostat. Totally.

I mean, it's something I had not thought of, Jordi, but it's a really good, like, when does that come up at some point? You know, like, that's very worried, not just seems like, but just ever.

Yeah, it seems like something that the court should be paying attention to heading into a decision from the jury. How much is the history between, OpenAI's counsel, Bill, who represented— he represented Twitter when Elon was trying to get out of the Twitter buyout. And ultimately it feels like the X/Twitter acquisition has worked out pretty well for Elon. He's made a number, a number of plays. So hard to, hard to imagine him like, you know, deeply regretting buying it, even though at the time he was happy to get out until Bill said, no, we're doing this.

I, you know, it's funny, um, he is— I, I can't underscore enough how different the tenor changed. I mean, look, it's like, it's opposing counsel, so it's always going to be different. But Elon went from like, I am an entrepreneur, sort of like Rosalie explaining to you what I think the future should look like, and like very concerned the future of the human race to openly antagonistic to Savitz questioning. And the thing that I'm very curious how this plays with the jury is Elon was very, he's like, I'm a literal guy. The questions you're asking me are too complicated or they're not yes or no questions, which is fair as a grievance. But also the thing he either doesn't understand or doesn't care about understanding is that that's just lawyers. Like, that's the whole point of a cross-examination, is to ask these questions as, as reductively as possible to, to get a type of answer they're looking for. And his job, or his, uh, you know, pre-trial sort of discussions with his own lawyers, is to know how to navigate those essentially while also telling the truth. So I don't know, they just don't like each other, or at least Elon doesn't like Savitt. And you can It's very clear that he's just like mad at that. And whenever Elon gets sassy, I would hear like clearly people who are fans of Musk like laughing behind me or being like, yeah, you got him. Like, it's just, it's very funny. It's like kind of a, it's different than the usual vibe is what I'd say. That's very interesting.

Outlook, what's going on Monday?

What's the outlook for next week? It feels like this week was the Elon side for the most part. Are we gonna flip? To Sam and Greg and some other OpenAI folks, or is there an intermediate step? Like, do you have a clear view of what the next couple weeks look like?

Yeah, so it's a little rough because we have witness lists in full that they presented and are in evidence, but you don't really learn who's coming until like very soon that week. It's actually a giant pain in the ass for me and for reporters who are trying to schedule, but like, So we, Jared Birchall, who's like Musk's family office guy, just finished testifying. We're gonna get this guy Stuart Russell.

That was today? No, no, no, sorry, that was yesterday.

And you said Birchall was very dry. Like, was that intentional? That feels like that does not work in favor of like swinging a jury. Was that more for the judge? Like, what was the goal of that testimony?

I think it was to really just sort of show how, Musk was trying to set up this, like, structure things as a nonprofit and, like, hammer home, like, he's always wanted it to be a nonprofit and he's not sort of like— so it was, I believe it was Musk's witness who OpenAI then cross-examined and OpenAI used the occasion to show an email that had, like, a proposed equity structure for Musk. And so, like, They both kind of used him differently, but I think like Berchall is like essentially an accountant, books guy, like behind the scenes. So I think it was like, here's a peek behind the curtain, here's how they were dealing with the finances, here's how Musk was like, this is only a charity. And then OpenAI's lawyers were like, actually, check this shit out. So, but I think like, I honestly don't know, they also, struck some of his testimony because this is a little complicated. But if you remember last year, OpenAI or Musk made a bid with Ari Emanuel and some other companies to buy the whole company. And that opened the door, or that sort of complicates things for Musk's side because there might be some of those discussions around that bid admitted into evidence. And that may not be good for Musk because if there's like weird compromising emails in there. So like it's gotten complicated, but next week we're getting Stuart Russell, safety researcher. We got Greg Brockman, maybe Sam Altman, maybe not. It's 4 days a week in court only, so we may not have time, but I gotta be ready. I'll be there most of the time, but I'll be there for Sam for sure.

And then I don't know. Sorry, rewinding a second. Did they bid? Did they bid on the PVC, like the, the for-profit arm?

Yeah, it was the— yeah, and they, they wanted to just sort of— I, I don't actually know what their plans were afterwards, but like it was like trying to take over the asset, those assets basically, and then, uh, you know, morph it into what they— I mean, they knew that it wasn't gonna get accepted as a bid basically, but it was like And then Burchall— sorry, Burchall in court was saying this was us trying to sort of establish a pricing mechanism to like value the actual entity itself. And that was going to help them somehow. I actually am not quite sure how that would help them somehow, but someone in the chat is probably smarter than me on that. But like, it was them trying to sort of take that over. And I think OpenAI said publicly at the time, this is like a stalling tactic. They're trying to like slow us down while also using this in court in this concurrent lawsuit later, basically. So it's all like really messy, and I think even Musk's side took a risk there if those emails get into discovery. But it's unclear to me if that's gonna happen. If we— if they are, then we may see them next week.

Why would— why would that, that kind of discovery process around that bid be coming up now when everyone involved was well aware that it had happened quite a long time ago?

So I think because of a line of questioning with Burchall yesterday about the bid, that like, if that happens on the stand— and again, like, a lawyer gut-check me here because I'm a stupid tech reporter— but if that happens on the stand, then it gives an entry point for OpenAI's lawyers to be like, okay, well, we need to, you know, now admit— basically the judge said to Musk's side, you open this line of questioning, now it's fair game to go into this, and then you can start calling new evidence in around that, like, to open up discovery. It's like really like strategy, and like you have to be careful in your strategy when you're asking certain questions in court. I've been learning a lot about it, so maybe it's a strategic misstep by Tobaroff, Elon's lawyer, but it's not quite clear yet.

Do you have an idea of the purpose of the AI safety researchers testifying, what the goal is there? Like, does that align with one particular side? Like, if there's more nonprofits, like, you still probably wind up with Anthropic, DeepMind, XAI. Like, it— like, having one, like, you know, more focused nonprofit going on for a long time doesn't doesn't necessarily lead me to like, oh, then we wouldn't be in an AI race. Like, it's not a clear solve for me, but I imagine that one side is trying to position this as important. But do you have any predictions for what that goal is? Totally.

So this is Musk's witness. And your point is well made, which is like, okay, even if you kneecap one, like good luck on literally everyone else. And like, there's been a lot of time spent, Elon talking, smack about Larry Page and how he doesn't trust him anymore. That's actually been really fun to hear. But the point, I believe, is this is actually a point of contention. So the judge does not, the judge prohibited going too far into doomerism into the world stuff. And she's like, look, that's kind of a sideshow distraction, extinction humanity stuff is not the point of this case. But Musk's side called this guy because they want someone, and this is my understanding, is that Stewart is like very aligned with the idea that AI is super dangerous and gonna harm us all. And so if you get—

how does that—

how does that make that point?

How does that not like— if this guy is going to come on and say that AI, like the most insane like doomer point of view, which I think is everyone's gonna have their own opinion around this debate. It's worth, completely worthy of having and talking about. But how does Elon square that with like, I'm trying to build the biggest cluster possible and distill on the rest of the industry's model and create tens of gigawatts of space compute and like, How does that, like, can't this kind of witness potentially backfire in some capacity?

No, I mean, you're exactly right in that squaring that circle is pretty hard. And like exactly what OpenAI is hammering on.

It's like, I should be trusted with the all-powerful.

We're gonna be in the Terminator situation no matter what. Do you want me in charge of the Terminator? Not some other guy, not some nonprofit, not some shareholders.

He is basically making, he's pivoting Tesla production to Terminators, basically. Optimus, yeah. Great.

No, I, and can I just make one last point is that I just think he misunderstood the Terminator because like we still survive and we fight back in that world. Do you remember?

I've been saying this, people haven't seen the movie. Movie. Jordi actually hasn't seen The Terminator, but in all these—

kidding me, dude?

Yeah, in all of these movies, like, it completely leaves out that there is a problem and then humanity overcomes it. And so it's like, if there is a Terminator scenario, like, I just want to know that you're on my side as the resistance fighting back. And like, whether the Terminator—

yeah, John Connor, hopefully.

I mean, I would— yeah, it's I don't know. I don't want to get into the full Yudkowsky thing, but like, like it is reasonable to be, to sort of steelman that. Like, if it is bad right now, it's not. It's sad.

Says I have to watch Terminator. I've got to watch Terminator tonight.

You got to watch Terminator. Okay.

So I just saw this on Twitter. Terminator 2 is coming back into IMAX pretty soon. Like you can go see it in the theater. So you have to go see it, Jordi. Like it's, it's Required viewing.

Let's go together. Yeah, yeah, yeah. I'm down. Do you have any— I would say that so far the trial is less— I mean, it's hard for the trial to go very viral because there's no audio video and it's just like live blogging. And you can imagine some of these scenes, like just the stuff that you're typing. I'm like, if that was on video, that gets like 50 million views in like, you know, in a few hours on on X, you know, across a bunch of these different aggregator accounts. But part is, is maybe part of the, like, my feeling is like, it's just so, it's just so like, I feel like the entire tech industry is like, it's just kind of depressing to see, to see these groups like fighting in this way when we have so much bigger problem, like we have so many bigger problems as an industry, right? The big problem being like public perception of AI. Is like already so bad. Like people, you know, don't like, you know, people. And then you just have these two peoples that were like, no matter who wins, we all lose or something like that.

Like there's some very negative stuff.

I mean, that, so I totally agree. And I do think actually everyone, not everyone, but like many in the AI industry have started realizing, oh, we have, Like, not everyone loves us. Like, this is a perception problem in a lot of ways, which, like, to me is funny because, like, I— not that I'm smart, but I feel like I've known that for a little while. And— but it's the, the issue is the, the industry is coming around to that, or at least certain folks are, and know that they want to change that perception. But this case has been in the system for, like, years now and sort of, like, actually happening at a time where they wouldn't want it to necessarily necessarily happen if they're trying to change it. So the timing just sucks, honestly, for folks who don't like that, if that makes sense.

Yeah, there's so many different like little tidbits, and I don't know if they're gonna be exact quotes, but you can just imagine plenty of things that are gonna happen on the stand winding up on a Bernie Sanders post and being like, "Stop it, pause AI," right? Oh my God. Because like, you know, it's always like give the tech people enough rope to hang themselves, and like, then there's a lot of rope going out the next couple weeks. Totally.

Oh my gosh, it's totally, you gotta, one of y'all, both of y'all should just show up to the courtroom at some point and like experience it because I mean, you have an actual day job, but it's like literally, it's really interesting and like an experience.

So you can't record video, you can't record audio, but is there a world where you have like 5 of your own stenographers there taking a perfect transcript? Would that be possible with enough resources?

And then we could do a table reading on the show.

That's what I'm thinking is that you could use voice cloning. I don't know if you noticed this, Tyler, what was it? Grok launched voice cloning today or something? Yeah. Which is, why is that controversial again? Well, for like, I mean, we've had this technology for like 5 years and then no one released it because like you don't want people to make deepfakes of other people's voices. Okay, yeah, because it's like textbook, like I could, it's like, oh, I look great on the west coast.

I call you John, I need like, I need $500. Can you gimme your credit card for a second?

So obviously there's a bunch of risky uses of that. The good use is the comedic table read in the voice of Sam Altman, in the voice of Greg Brockman, in the voice of Elon Musk that we can all enjoy.

I was thinking more of like doing like a, like, basically end of the day. There's like, we do a play here where we have all of a bunch of different actors.

Oh my God. Full makeup, full costume.

So, you know, there are transcription services and even like—

You could do, you could be Mike Isaac by like—

Oh yeah, we need someone playing you for sure.

No, no, no, it's like, it's a cameo.

He's playing himself. Oh, he's playing himself, okay, nah. I think you gotta go full Nathan Fielder.

And the whole time just crashing out like, "I forgot my lunch, I forgot to fill up my water." I had a banana.

I'm good now. I had a banana. Oh my God.

Can I just say, there is an attorney there that looks exactly like Nathan Fielder, that I'm like, almost like, is this Nathan Fielder? Are you sure it's— oh, it looks like Nathan Fielder, okay. Every time I see him, I'm like, is this like a bit? Like, am I gonna be on TV in some way? It's very intense. I'm gonna ask him next year.

I mean, Elon Musk says the funniest outcome's the most likely, and that might be it. Seriously. What is the funniest outcome? The funniest outcome is like Elon wins And the penalty is like $50 or something like that. It's like, Elon, like, you win. Here's $150, enjoy. Oh my God. Just the most like inconsequential fine for the AI industry, which is flush with cash at all times.

Well, I find that like fine, like there's so many, anytime a company in tech gets fined, or not every time, but often it's like, okay. Sometimes it's a fine for ants. Yeah, like they did something that was bad. They generated $100 million of revenue and the fine is like $1.5 million. Yeah, it's like, did that—

I think the current Meta YouTube lawsuit was like maybe a $10 million settlement. Of course, there was like a knock-on of potential for class action.

Yeah, that was for one. That was for one individual.

That was one individual. But still, it was like you have this like landmark case, this big build, and then you're like, Oh, how damaging is it gonna be? It's gonna be like 5 minutes of revenue.

But the stock actually in Meta earnings as far, I mean, there was a lot of shit that took the stock down, but like, I think one of them was them saying like, this might have a material impact on, like these cases may have a material impact in the future if it opens the door. So like, yeah, I agree. Like it's always, remedies are always like, it can be existential from a financial point of view, but also just from a guess what, this is about to become your whole fucking life for the next 10 years or whatever, if that makes sense.

Yeah, yeah, we talked to a law professor who said like, the question is, does social media exist in the future? Like this is an existential moment. I'm not sure that we're quite there, but it is, there's definitely a risk and you need to consider that if you're an investor. Totally. Anyway, what's the game plan for next week? What's going in? Are you doing trail mix? Are you doing protein bars? I want you fueled up, ready to go.

CamelBak?

You should be able to do like a sponsored CamelBak, you know.

Did you ever watch Jury Duty, the show Jury Duty?

Oh yeah, yeah, of course.

I want to be the guy with like the chair pants and like the water thing attached to my back and like, I need, I'm gotta, this is blogging, this is back in my blogging all things digital blog days, man. This is, this is it. So I will be more prepared.

This weekend, this weekend, you should go sit in a chair for 8 hours straight.

Are you gonna do all 4 days? Are you doing all 4 days?

I think we're gonna switch off a little more. I apologize, I will not be my like disastrous falling apart self every day of the week, just maybe 2 days of the week. So, uh, Kate, you should follow Cade Metz, my other colleague who's there. Um, although he, he is much more put together and does not tweet like a person off their meds like I do. So, uh, it'll be a different vibe each day. But I will be there next week.

Last question, is this book material? Does it rise to that quality, this drama?

I feel like someone's gotta write a book.

I would write a book on your coverage of this trial.

Yes, that's what I wanna read. The Mike guy just exposed Trump's boss.

And then I would do a podcast about it. I would do a podcast about it, and then someone would make a book about the podcast about the book covering Mike guy's covering.

We'll create a content org. Ouroboros here.

Keep it going someday. I don't know, we'll see.

Anyway, thank you so much for taking time.

Great to catch up. Good to see you, Mike. Thank you. Thanks for the update.

We'll talk to you soon.

Have a good one, guys. I'll see you soon.

Goodbye. Uh, we're running late, but we got Kyle Harrison in the waiting room. Let's bring him into the TVP and Ultra Dome. He wrote a book about the Andoral Thesis, 300 pages of defense tech lore, history, knowledge. You know what's more valuable than any of these cars up here in the Hollywood Hills? Knowledge. And you wrote a book. Tell us about it. What was the process? What was the thesis of the book, The Anderol Thesis? It was an incredible experience.

I would say, I think the thing that I came to appreciate the most was everybody talks about Anderol like it's just any other company. All these high-growth companies that are defined by an awesome team chasing a big idea. Market turns out to be bigger than you possibly expect. You know, Ben Horowitz has to tell Databricks that they're underselling themselves at $10 billion. It's $100 billion. It's $200 billion. Yep. Like, those stories are very common. Yep. The Anduril story is so underappreciated because it is counterpositioned to— there's like over almost 100 years of military, geopolitical, industrial history that really define the company that people just— they just— they don't know. Like, they think about, oh, this is an interesting strategic choice. And it's like, no, this is actually because of things, decisions that people made in the '50s and '60s that we literally have to do things this way. That was the biggest takeaway.

Yeah, I feel like, uh, Shamsankar at Palantir with the First Breakfast has done a good job shining a light on the Last Supper. So many people in tech are familiar with that sort of key moment of industry consolidation that led us to this moment. Uh, were there any anecdotes or historical stories that you discovered that were more under the radar, more like less discussed as like, you know, the founding moments of this whole journey?

The biggest appreciation that I got was for this sort of like dual pairing of people pushing forward a specific project. So in the kill chain, which is Chris Brose, who now is the kind of co-president and chief strategy officer at Anduril, but at the time was coming off the armed services Committee. He wrote this book, The Kill Chain. He talks about this idea of like military mavericks. And I kind of combined this with this kind of founder, founder mode energy that people bring. And so you have this like founder plus military maverick pairing, where you have folks like the team. So Bernard Schrieber, I think his name, the guy who basically led the team inventing ICBMs. He absolutely needed Eisenhower as his kind of military maverick. Sure. To like pair that effort. There's countless stories of this, the modernization of the Navy, there's all these different things that were like that one-two punch was kind of required. Once you get away from like, you know, at one point the DOD was 36% of global R&D. Yeah, they had a pretty good handle on the cutting edge. Once you've— the further away you get from that, the more you need founder mode people digging in to figure out the solution. Sure. And it's heartbreaking because you see these like, countless examples of like prophets of urgency throughout time where like once we lost that, that sort of like desire for founders to come in and solve problems. And you have all these people who are like throughout the '80s and '90s and early 2000s, they're releasing reports and writing white papers and talking about how, hey, this is going to be a problem. Warfare is changing. China is going to school on how we do conflict, and that's problematic. And none of them had sort of founders to be able to come solve problems because no one was willing to pay attention to the actual problems. And so we lost that pairing.

There's reports of Vanderhall raising a new round, $60 billion potentially. Is this book a financial thesis? Is this financial advice? Is this something that investors, potential investors, maybe future investors—

Yeah, there's 4 chapters dedicated to price targets, price targets, multiple For entertainment purposes only, right across the— right across the— no, this is— this does not constitute financial advice. Okay, entertainment purposes only. Uh, you know what's crazy is, I think if I remember correctly, in secondary markets, uh, if, if there's any secondary volume that, that Andreessen Horowitz founders haven't already gone to war on, I heard somebody say, uh, that it was like $100 billion indication. Like, even just tracking like what people are willing to pay because of Matt Grimm's war, there may be no, no supply for the demand, but the demand is priced at $100 billion. So it's not, you know, where the new round ends up, whatever. The appetite continues.

Sequel, sequel, the Admiral Financial Thesis. Let's go. That's right.

This is investment advice. Admiral's been very, very acquisitive, and it feels like some of the most, like, meaningful innovation and like founder mode behavior in the last 20 years has come out of like Ukraine and what they've been doing to basically figure out how to like innovate and create these like R&D loops around drone warfare. Like, do you, any, I don't know if you can make predictions, but like does Anduril end up buying any of these companies Is that, is that, have you, have you uncovered anything on that front? Is that even, is that even a possibility? It just feels like they've done some very impressive work to protect their, their country. And a lot of it seems relatively aligned to Anduril's own efforts. Yeah.

So there's a couple of pieces of the puzzle. So first, like in terms of how acquisitive Anduril has already been, Our good friend Paki McCormick has described it well, I think, calling it an API into the DOD, where Anduril effectively enables this like plug-in mechanism of somebody can go very deep into a very specific capability in a way that like companies like the company that made Dive, like it'd be very difficult for them to sort of become a standalone large company, but by being acquired by Anduril, immediately they're able to plug in and get a $100 million contract with with the Australian Navy and stuff like that. So I think like the acquisition muscle plugs in a lot of this stuff pretty capably. The second piece of that puzzle is I think that a lot of times we do a disservice, like the venture capital industrial complex does a disservice by focusing everybody on this sort of like neo-Prime North Star of everybody needs to be the 10, 20, 50, $100 billion outcome. I think that there are a lot of opportunities for companies to be very, very good at a very specific thing capability, and then to plug into some of these systems, whether it's through partnership or through acquisition. And what's funny, probably the most frustrating part about the book is we wrote it over the course of 2 years, in part because we, you know, it's difficult to sit down, we all have day jobs and stuff. But in part too, because the world is constantly changing, right? When we started, when we originally sat down thinking about what would this would look like. You know, none of the things that happened in Israel, like Ukraine was sort of just continuing to evolve. Venezuela happens, all these different conflicts happen over time. But then on top of that, Anduril as a company continues to ship. And so even 6 months ago, I would've said, you know, what is the kind of like signal? And what's interesting is you look at some of their product launches, some of these through acquisition, some through partnership, but I don't know if you guys have seen their like anime style. Product launch videos, right? Right. So I don't know if everybody has noticed this, but, um, the last two videos had teasers of their future products. Hmm. So I think it was the video that, where they talked about Fury and then it was maybe Fury or the Barracuda. Um, and then at the end of the video, it kind of went under the water and you started to see these assets moving, but then it cuts away and you don't know what it is. A couple months later, they, they announced Copperhead, which is their like torpedo product. And that paired with dive and stuff. And then they have a video about that. And at the end it pans above the water and you start to see this thing kind of come out of the corner of the shot and it cuts away. And then just whatever it was last week or so, Anduril announced a partnership with Hyundai about building autonomous surface vessels, right? So it's like kind of going into all these different surface areas. I think that that is indicative of like where the signal is, is that whether it's through partnership, whether it's through acquisition, acquisition, a lot of the stuff that's happening in Ukraine, what you're kind of exposing is that there is significant opportunity for like networked assets. And Anduril absolutely has this mindset on becoming the sort of networked everything, the military internet of things is what Chris Brose calls it. Where can people find the book? Amazon.com. Our own Jeff Bezos is hosting it for us. We're not in the book printing business. So— You're going to do an audiobook for us?

Version. Oh, they print— they have a print-on-demand. It's great.

It's— you put in the thing, it takes a lot of— you think of what I'm thinking, John?

What you're thinking. It's pretty good. It's time to— it's time to ship books.

You got to write a book about PPPN, the novelization.

We want to make—

we want to make ad-supported business books. They're free. Yeah, but they come with ramp ads. Yep, that's great. You could— you get yourself a good designer.

It's as easy as it could be.

Last question from my side. Do you, have we reached peak early stage defense tech investing boom? Has it already passed? Is it, are we still on the up? Like, what is your feeling at this point?

I think that it is going to take, to take the air out of the balloon is gonna require some type of black swan, hopefully micro swan, black swan, micro black swan event. We don't want it to be too bad until—

Wait, like, wait, you said to take the air out of the defense tech investing sales, you would need a black swan event. So you're saying like, we declare global peace.

World peace. So I think it's going to take, I think it's more about the capability. I think it's going to take like, either there is a company that has been highly valued, highly sought after, kind of held up as one of the the golden children of the, the neo-defense industrial wave, that just something— it's terrible.

It's really bad.

Oh, like something very bad happened. And so then investors start to say, oh crap, wait, you're telling me not every company I invest in that does defense is going to be the next Anduril? I can't underwrite to every company? Maybe they're not $60 billion, but at least they'll be $20 billion. Like, if that underwriting muscle goes away, you have a whole swath of carpetbagger VCs that suddenly get very panicked. Like, something like that happens.

Social there were a whole bunch of like, like fifth wave social companies that didn't get the Instagram acquisition. They didn't get the WhatsApp acquisition. They raised a couple hundred million bucks and they just sort of burned it down and just went away and nothing really happened. And the VCs sort of woke up and they were like, okay, we gotta be much more careful if we, if we're investing in that category. So yeah.

Yeah, 'cause I think it's like, I think the problem is like there are sort of like three varieties of VCs are in and around like defense tech more broadly. And it's like, The people who invest and get it and they're actively built, they understand the like moral implications of the types of technology they're investing in. Then there are still people who have said, we will not invest in weapons. We don't wanna invest in defense, whatever. It's this middle ground here that is problematic. It's these people who, they have no real philosophical weight behind why they do what they do. They just see big number and pursue shiny object regardless of consequences. The line go up.

Well, we hope that the line goes up on your book sales. Head over to Amazon.com and buy The Anduril Thesis by Kyle Harrison of Contrary. Thank you so much for coming on the show. Can't wait to see you soon.

Great to see you. Congrats, dude.

Goodbye. Thank you. Up next, we have David Marcus from LightSpark. He's the co-founder and CEO, and he's launching Grid Global Accounts, a Bitcoin-powered system enabling global USD accounts, cards, and cross-border payouts. We talked a lot about crypto payments yesterday. We'll let you introduce the business yourself again. Welcome back to the show. Give us the update on what's going on in your world. Great to see you.

Thanks for having me back, guys. Well, look, so the big announcement we had this week— so Lights Park has been around for about 4 years. And in the last 4 years we've been building connectivity into payment systems all around the world, like 65 countries now. And the announcements that we added this week is Great Global Accounts, which is basically a global dollar account that any platform can issue and give to their stakeholders. And that dollar account is what I think is the most powerful money account ever created because you can move money in real time to 65 countries, payment systems, like really at any time of the day or night. You can move money on any chain with any stablecoin you want. You will have a Visa debit card that can allow you to spend that balance at 175 million merchants around the world that Visa supports. And Lightspark is becoming a principal member of the Visa network as part of that process. And it's got like this nice feature that enables those platforms to actually be ready for the, the complete change of interfaces as we move from websites and apps to conversational interfaces driven by agents. And so this idea that you can delegate safely money to agents so they can buy things or send money around the world for you is really part of that whole Grid Global Account platform that, that we've announced and launched this week.

Very, very cool. So, uh, how— like, when, when did you guys start thinking about this opportunity? Obviously, you know, everyone's thinking about stablecoins. We have more regulatory clarity than ever. Was this always on the roadmap or did you guys start to get customers saying like, hey, we want support for stables and not just BTC-related products?

Yeah, I mean, look, I think, you know, so we've been working on stablecoins for a long time, but the concept of a global dollar account that can actually live everywhere is really novel. And there was no chance of us actually doing that even 12 months ago, because the prerequisites, the stars were really not aligned. And right now, with the Genius Act here in the US, with MICA in Europe, and similar legislation all over the world, plus really, really great embedded wallets, The self-custodial wallets of like 2 years ago, where you had a seed phrase that you could lose, and you would never give that to a family member because they would lose money almost invariably, all of that is gone, right? So now you have really great wallets that you can log in with Google, with Apple, with a Passkey, with any login that's familiar, and never lose your keys or your money. And the last part of it is actually really stablecoin-backed debit cards where the networks have really leaned in hard on. And like now gives us an opportunity to really basically connect those dollar accounts to 175 million merchants around the world, making that balance really useful for anyone receiving money on them. Yeah, very cool.

Is the pivot from crypto miner to AI neo cloud that's happening across the industry having any sort of effect on crypto markets? Or is that a— because we talk to companies all the time that are compute constrained. There's this CPU crunch, there's GPU crunch, there's backlogs for power and data centers are delayed. And that's what's going on in the AI world. But Bitcoin as a network needs compute as well. Many other chains need compute. And is compute scarcity bleeding over in any way to the upside or downside. I just, I don't know the effect at all. Does it matter? Or is there enough compute for crypto to do everything that they want to do as a community?

Yeah, I mean, look, I mean, proof of work is definitely, and so, you know, more specifically for the Bitcoin network, proof of work is definitely using the level of energy that at times will compete with like AI data centers. Sure. The hardware is like, started GPU-based, but it's very ASIC-driven now. But on the energy side, definitely competes for the same source of energy. But look, I am personally an acceleration maximalist, and I actually think that we're actually going to unlock unbounded energy in the next 10, 15 years. So I actually don't think that we're going to have these types of issues of actually competing resources with massively forward-deployed amounts of capital on Bitcoin miners being replaced with GPUs. And also, those data centers for Bitcoin mining are so specific. A lot of those miners are liquid-cooled in a certain way and operate on certain grids that they're able to actually turn on and turn off the mining capabilities really easily, whereas it's harder for agents.

You don't necessarily want to do that if you're actively serving inference or even just training. You might not want to bring the train down.

Interesting. Where are you expecting to see the first real hockey stick growth when it comes to agents leveraging stablecoins? Right now, it feels like there's it's a lot of potential early experiments going on, but at least to me, it's unclear where we'll see that really breakout use case.

Well, I mean, I can share a little bit more about my thoughts and my experience. The first thing is I think that everyone competing to build an agent-to-agent protocol standard is kind of really future forward, and there's not much happening there, right? It's like there's a lot of big news and big announcements and very little volume. And that's normal because like, you know, agents are not moving money at scale by any stretch of imagination. We took a little bit of a different stance, right? Which is that we want to build a mass market money account that suits the needs of people and businesses all over the world. And then we've built a safe scope delegation framework that enables those agents to actually use the debit card, use the 65 countries reach of moving money around, use the multi-chain support for stablecoins. And then you can delegate that. And so what I've done with mine— and I've lived with it for the last 5 weeks or so— so I have my Grid Global account. And then I have an OpenClaw running on a Mac mini at the office. And I connected both. And so it just takes a scoped debit card, goes buy things online, which is actually more difficult than I thought, you know, starting this, but like it now works. It can message people on WhatsApp that I need to send money to and offer them like, you know, 7 or 8 different ways to pay them. And I don't have to worry about it. And it really works real time really, really well. And it's been so much fun to actually build on Open Claw and have like a money account that Lobster could use. 'cause it's been a lot of fun.

Yeah, yeah, that's awesome. Lobster counts. Well, thank you so much for taking the time to come chat with us. Congrats on the progress. Thanks for having me.

It's great to get the update. Always great to see you. I'm sure we'll see you back in the next, hopefully in the next 1 or 2 quarters.

Yeah, fantastic. Sounds good. We'll talk to you soon. Great to see you. Goodbye. Have a great weekend. Up next, we have Turner Caldwell from Mariana Materials. Mariana? Rare materials like the Mariana Trench. We'll get into it with Turner Caldwell, the co-founder and CEO. Help me understand how to pronounce this company's name. I don't want to get it wrong.

Uh, Mariana Minerals. Mariana.

Okay, where's the name come from? Uh, comes from the Mariana Trench. Mariana Trench. But that's because we were looking at—

Let me guess, they've got some crazy minerals down there, or do we even know?

There are some minerals at the bottom of the ocean. That's right, that we are So you're saying you yearn for the trenches.

That's right, there you go.

Okay, the business is an autonomous copper mine. Can you take me a layer deeper, explain the history of the business, the current shape of the business, where you're taking it all?

Yeah, sure, thanks. So we just restarted a copper mine that we acquired at the end of last year with a major focus on autonomy across the entire stack. So that includes the kind of orchestration layers around mining operations, as well as autonomous execution, which we'll do with partners. And then also autonomy around the refining processes. So the site is a mine and a refinery. [Speaker] Yeah.

But in practice, what does that look like? Like you go into the mine and there's some like ERP system or maybe just some like manual paper-based workflow. And you're saying like, let's go Vibe Code, a software solution for it? Like, yeah, or is it like—

yeah, basically the way to think about mining is that, uh, you have a whole bunch of different disciplines that have to collaborate. It's a big coordination problem. You have geologists, you have a geologic block model, you'll have a mine plan, uh, you'll usually have multiple mine plans, like a long-range, medium-range, and short-range mine plan. Uh, that has to be transferred to the mine operations teams that are running on like large-scale— or not autonomous— large-scale manual mining equipment. Then the, like, what is actually in the ore usually shows up on a piece of paper to the refinery saying, like, get ready to process this. Okay. And then the refiner, the folks who are running the refinery, have to figure out, like, kind of the optimal process conditions to maximize recovery of copper, lithium, nickel, cobalt, manganese, iron, graphite, kind of the whole thing, all of the metals through the processing circuit. And so what we're doing is we're kind of integrating everything into one integrated data frame first. And so there's a lot of work that goes into bringing all of the data feeds back into one aggregated architecture. And then we'll build world models of the mines, of the refineries that we then use to do a reinforcement learning training that we feed up into a reinforcement learning training pipeline. And then you want to close the loop loop between what reinforcement learning agents are recommending on the policies that are developed to the actual thing that's happening in the mine. And so we'll bring autonomous haul trucks in, we'll bring autonomous drill rigs in, we'll bring in autonomous loaders and dozers and graders and water trucks, kind of like all the assets that are being used to run a mine that today are, you know, 99% of mines are manned. And we want to go full closed loop between kind of the world models and the reinforcement learning agents and determining what the autonomous assets actually do. And then actually that's the refinery.

What do you think the potential efficiency gains are from going from legacy copper mine to the kind of end state, you know, fully autonomous system?

Yeah, I think it'll happen in two phases. I think the first phase is going to double productivity through just like better orchestration and better logistics management and better utilization of all the equipment. That's productivity and unit costs. You know, diesel is a good example for heavy haul machinery like half of the diesel consumption, up to half is from idle time in the pit on the equipment. And so there's a lot of short-term potential gains. But I also think that autonomy is going to fundamentally change. It's going to unlock a lot of operating modes that can't really be done with humans. And so the way that the mining industry has driven down cost over an extended period of time has been through scale. So the haul trucks are getting bigger, the shovels are getting bigger, and it's all focused on reducing your labour intensity, so maximising tonnes of ore, tonnes of ore moved, or tonnes of material moved per unit of labour hour. But as you move into autonomous systems and eventually design the mines themselves to be autonomous, I think we're going to see things get smaller. And swarm mining is going to get unlocked, and that's fundamentally going to enable you to actually be more selective in how you mine. Your roads get smaller, which means you can maximise recovery of the ore from the pit. So I think it'll all happen in two phases. It's like take the existing architecture, make it autonomous and maximize utilization and availability of equipment. And then there's going to be this like once autonomy is kind of like fully demonstrated across the stack, there's going to be this opportunity to actually attack like fundamentally how is power mines mined.

You started the company in 2024?

Yep. End of 2024. Raised a seed round in August. We're at 220 people now across the company. We have, we have two projects. One is a lithium refinery in One's a lithium refinery in East Texas that's extracting lithium from oil and gas wastewater, and the other is kind of first large-scale commercial mine.

Another one. He's got two.

Yeah, exactly. I have a question. Pitch for you. Robotic canaries.

Do they—

Well, we have— We have— We have— I'm sure. We have a Boston Dynamics spot that is basically like— mobile sensing system instead of the canary in the coal mine. And we use them mostly for like undercarriage inspection. So like you want to keep humans away from like heavy haul machinery. The closer they get to it, obviously the more hazards that are present. But also like the chemical refineries. Ideally, humans aren't walking through chemical plants all the time, kind of doing thermal inspections. There's a lot of stuff that's like auditory. So you're like listening for sounds of the motors. Sounding weird. Um, and the— exactly. And then that all, that all needs to feed back into, uh, into like, again, the integrated kind of like, uh, sensing, uh, integrated like data frame that tells you like how—

what's the healthier systems. Okay. Uh, did you start with copper 2024? We started with lithium in 2020. Um, and because, because it feels like, it feels like, you know, there's this whole autonomy story, more efficiency, I say all of that makes sense, but the price of copper's doubled since you started the company. And I feel like that just makes bringing a copper mine back online just wildly different from an economic perspective. And it's very interesting that you're sitting at the tailwind of like two AI trends, which are just demand for copper for data center development, wires, circuitry, all this other stuff. And then you can actually be more efficient, but you can be as efficient as they were a couple years ago reap twice the price. It feels like the oil dynamic, right?

Yeah, that, that's right. But I, I don't think— so the reason you want to be diversified is because like all these metals are cyclical and you can't really like live and die by a single commodity price. And so the plan was always to diversify, and the plan is to continue to diversify into nickel and cobalt and uranium and graphite and rare earths. And the, you know, as we, as we kind of work to build 10 projects in the next 10 years, um, but the You know, we started working on this project kind of like well before copper had its like little explosion and run that it's on because you don't really want to jump into projects in the middle of the hype. That's when you pay the most for an asset. That's when you're going to like the labor pool is most constrained, is like most in demand. The equipment suppliers that make the equipment that is specific for a specific metal are going to have longer lead times and have higher pricing. And so you really do want like We jumped into lithium when lithium was at the bottom of its recent trough and jumped into copper before it started to go on this run. So generally, we want to work on metals that we think have a long-term supply-demand imbalance. And for copper, it's very much felt. Grades are dropping, ore deposits are getting deeper, mineralogies are getting more complex, and it just means it's going to be a lot harder to make copper in the future. And that's what we want to work on.

What were you doing before this?

I was at Tesla for about a decade. Most recently was running our minerals and metals team. So I was building our lithium refinery in Corpus Christi in Texas and running a lot of our battery recycling work.

Very cool. Kind of like the absolutely most perfect background for a company like this. But any, any VC couldn't— a VC couldn't dream of a more perfect resume for this business.

Well, what's all the text? Text in the Gulf behind you for that map? Yeah, what does that say? I understand that it's a map of America, but what's in the— what's all the text for?

So these are all, um, uh, it's basically calling out like things that don't have space for the words on the map. Okay. Um, so it's just, it's a, it's a legend you got.

Uh, crazy detailed legend. Yep.

Well, our project is right on the border with Louisiana, so that's the, uh, that's why we got it in the background. Okay, I'm sad.

Well, you're a legend, and thank you for coming on the show. Legendary appearance.

Yeah, great, great to meet you.

Uh, have a fantastic weekend. Thank you so much for coming on the show. We'll talk to you soon. Goodbye. Another one. We got another guest, Everett Taylor. He's the CEO of Kickstarter, coming into the UltraDome. He's in the waiting room right now. Everett, how are you doing? Welcome to the show. I'm good, I'm good.

Sunny Florida right now, so I'm escaping the cold in New York right now.

So I'm feeling good. Amazing. Amazing. I'm very excited to talk about the Next Wave Fund, but first I'd love to know a little bit about your journey to Kickstarter, where the business is today, what your day-to-day is like, how big is this company? Sort of set the ground for us.

Yeah, so I mean, Kickstarter, you know, started in 2007, 2009.

I've been here about 4 years now. Yeah.

And, you know, we're the largest creative crowdfunding company in the world. We still are today. When I came into the company, though, the company was in a bit of decline. We were down 20% year over year. Really? Since then, you know, we've grown dramatically year over year. Last year, we had another record-breaking year. This year, we're already up 52% year over year over last year. So the company has been scaling and growing as we are transitioning just from a crowdfunding company to more of a creator economy company.

How has the AI boom, the Gen AI boom benefited or changed Kickstarter projects? When I think about the, you know, the catnip of Kickstarter projects, it's like a board game that probably took a lot of time to build and I mean, if the idea's great, I might not care if somebody used an ImageGen model to speed up the graphics for a tree on the back of some playing card. I could see this being a big boom.

Yeah, not at all. I mean, I'm sure a lot of Kickstarter creators historically were spending a ton of money on renders.

Totally, totally. But at the same time, there's, does this exist at all or is this just AI slop? And there's a community dynamic and there's an artist displacement dynamic. And so how have you grappled with that? People with the pros and cons of the last 2 years' crazy boom in AI?

Yeah, it's a great question. On the AI side, funny enough, gaming is where it's probably the most controversial, right? Yeah. Because gaming tends to be very artist-forward, right? And so we've gotten probably the most pushback and some of the most negative pushback on the AI side from gaming because they really want to see artists be able to have jobs and do their thing. And so, you know, we were one of the first major tech companies to have our own AI policy. Every Kickstarter creator, if they're using AI, they actually have to say upfront how they're using AI, crediting, crediting the artist if they're using another artist, et cetera, et cetera. But on the design and technology side, these new AI-powered, you know, tech products, especially in the hardware side, it's booming, right? So both categories are growing, but you, you see the differences in how people are accepting of AI.

Yeah, that makes sense. Talk about the partnership with Google. What's the news?

Yeah, so this week we announced the Next Wave Fund with Google to power the next generation of tech innovation. I'm a founder myself. I saw the struggles being an early-stage tech founder struggling to get funding. And so what we really wanted to do was give early-stage tech startups and small business owners the funding they need, the tools to be successful, um, the guidance that they need, and also be able to leverage Kickstarter's audience of millions of people to really get their first customers and scale their product. So this fund is focused on, uh, innovative products and hardware, software, gaming, and connected technology. And people will get $10,000 upfront, uh, non-dilutive capital, um, from Kickstarter and Google, um, to launch their products and their companies on Kickstarter. And we wanted to focus on these areas because Kickstarter has just a natural product market fit for technology and gaming. They are our two largest categories. And we also have a large audience for both.

So you're going to have a huge portfolio.

Well, Kickstarter doesn't take— Kickstarter doesn't take any equity. I mean, I have an Oura Ring right now. I wish we did take some equity on that. But yeah, we don't take any equity. We don't take any equity. And so this is really as a way to support, support people, give them the tools that they need to be successful and also build a relationship with Google, right? These people will have an opportunity to apply to Google Accelerators for their business as it grows and scales. They'll get access to our team that, you know, has decade-plus of experience in scaling new products. So it's a great opportunity for a lot of people.

Where can people get started? Take us through the— is there an application process? Where does the journey start for someone who wants to participate?

Yes, so you can go to our website and check out the Next Wave Fund. You can apply. This is for entrepreneurs or small businesses with fewer than 20 full-time employees. Like I said, your project has to be focused on technology or digital gaming. Sure. And unfortunately, We've been hearing this and we gotta shout out to our international creators, but this is for the US only. So at least one member of your team has to be a citizen of the US and be 18 years old, so.

Well, lots of opportunities. And thank you for taking us through it. Thank you for taking the time to chat with us. Very cool. Great, great to meet you. And great to meet you. We'll talk to you soon. Peace and cheers, Everett. Goodbye. There are many, many, many, Many more conversations going on the timeline today. One was kicked off, I think, uh, by Josh Kushner. I think he sort of started this. There were some other people talking about data center beautification, uh, and he summed it up, uh, perfectly. He says, make data centers aesthetically beautiful. And so people have been quote tweeting, posting, riffing on this, uh, sharing different ideas. More people would be pro-data center if every data center had a beautiful open to the public heated pool. That's an interesting twist. Full water slide. I think half pipe is a big idea. Monster truck rally constantly going on. These are ways to win people over, sort of a nitro circus going on in every data center.

Monster truck rally on the roof of a data center? Yeah.

Sort of a universal basic nitro circus. So you get, are you familiar with Nitro Circus? Are you not? Is this a 3 fingers moment for you? Are you like, ah, I'm familiar?

No, I'm just imagining they're building the ramp.

Just say you've never actually been to Nitro Circus.

I'm imagining they're building the dirt bike ramps over the dataset. Exactly. Backflipping over. Exactly.

Julie Young shares a trick, this one neat trick that Los Angeles uses to hide oil deckers. We have a few oil derricks in, LA that are disguised as synagogues. They are literally just sitting in the middle of the city. Lol, surprisingly few people are aware. There's a number of data centers that are disguised in Los Angeles as well. Usually you can tell because there's no windows in the building, but these exist in downtown, usually for edge computing, for like the content delivery network. You store in your Netflix videos there, that type of stuff. Not doing training runs, not gigawatt clusters. I did look up up how energy intensive Bitcoin is relative to what we think of AI. I think the numbers are OpenAI and Anthropic are around 2 gigawatts in terms of capacity, something like that. You know, we've heard about the Colossus data center. The average Meta campus is around half a gigawatt. Zuck's building something in the 1 gigawatt range. There's gigawatt a week plans, but If you were to put the Bitcoin network current sort of estimates using between 120 and 170 terawatt-hours per year, what is that in average power in gigawatts? It's between 14 gigawatts and 19 gigawatts. And so that's where you should sort of comp Bitcoin to if you're putting it in AI terms and in the numbers that we throw around, which are usually like a 1 to 2.

Before our next guest. Palmer Luckey has been getting into it with someone on the timeline. He says— Clifford responds to someone named Clifford by saying, dumb tweet. Toto has been fabricating advanced ceramics for semiconductor manufacturing for many years. They make more money from that than toilets. Strong technical moat in a rapidly growing industry. Clifford says, perhaps, fair point, but call me dumb under your real name or STFU, coward. Just like not realizing that he's talking to, talking to both of you.

But isn't Cliff Asness like a big deal? I think he's an author. He's, his papers, I've heard of Cliff Asness before.

That's funny because—

And so it is funny that they just like didn't over— Oh, he's the co-founder of AQR Capital Management, hedge fund manager. Is this the right person? Forbes estimated net worth of $3 billion. I believe. I think this is accurate. I think he, Yeah, global alpha. He started career in 1990, worked at Goldman Sachs, AQR Capital Management.

He missed the like Anderlecht, Bearfyte.

He just has to see the SPVs or something. Stable Asset Management and other institutions in the aggregate made commitments of several hundred million dollars to a new multi-manager hedge fund. He's published in academics. He's done a bunch of stuff. I've heard of him. Of him before, so this is very funny.

He runs— he has AQR. Oh, AQR, yeah. Interesting.

Anyway, very funny. I believe we have our next guests in the waiting room. Let's bring them into the TVPN UltraDome. Natomi Wonderko, welcome to the show. Would you mind both kicking us off with some introductions on each of you?

Yeah, of course. Hey Jordi, John, great to see you both. So Justin Wexler, one of the general partners at Wunder Co., working closely with Jeffrey Katzenberg. Yeah. And you've met all of us. Yeah. Sujei, Chen Li, Jeff, Anthony. So it's been an amazing run. I've been with the firm since early 2017. Cool. And also on the board of Netomi.

Okay. So yes, please introduce Netomi.

Yeah. Awesome. Great to see both of you.

My name is Puneet. I'm the founder CEO of Netomi. We are an agentic application tier company focused on the customer experience use case.

Okay. What specifically in customer experience? Is that different than customer service or are we doing surveys to understand how companies and products are being received, all of the above? What's the shape of the business today?

Yeah, we look at customer experience a little bit differently than how most companies in AI have have looked at this. So if you, if you think about it, historically businesses were designed as a sales, as marketing, as customer service. Yeah, because that's how the customer journeys were created and they had to keep those silos to keep cost efficient. But, you know, with the GenTech now, there's a resource that has become abundant. You could apply it through the journey. That's what we are doing. And also one of the key factors is if you think about traditional customer service, it looks at customer journeys and it says we'll wait for something to break and then fix it after the fact, and then hopefully we regain that trust. We are saying, hey, you got— you can address this upstream, you can address this in digital experiences. You don't even have to wait for the problem to occur. So that's what we are doing and that's why we are excited to partner with Accenture and Adobe in this financing that we just announced. And we're going to do this for all the Fortune 500.

Talk about the decision to go with enterprise versus small business. Was it ever on the table to go broader self-serve, or is this— is there something unique about the technology where you're— it's actually more suitable for enterprise use cases right now, or is that just your personal DNA where you flourish? Actually, all of the above.

So my personal DNA, I grew up in automated trading, building event engines for different firms on Wall Street.

You're born in the enterprise. I like it. Born in enterprise, exactly.

Born in the enterprise, born in low latency systems, born in event correlation, all of those things that just came together for this. But also we looked at where is money getting spent for this problem. So if you look at just broadly human capital spend is about $500 billion a year. About 75% of that comes from the world's largest enterprises, literally under 1,000 companies, right? So that's how fragmented the market is. And now with Natomi's technology, you know, you could run the autonomous front office. So when you do that, not only is that a cost efficiency which more than pays for itself, it really transforms the entire company because you can connect all the systems processes right in those agentic flows and in that sanctioned architecture presented safely to your customers. So yes, love that medium to high complexity, love the enterprise complexity comes in the DNA. That's what we are the world's best at. And just to add to that.

Yeah. So at Wonderco, as you all know, we invest in consumer, we invest in different areas of technology. But a big focus for us is enterprise and how enterprise is going to evolve with technology, particularly AI technology. I'm fortunate that I work with Jeffrey, who has a lot of great relationships across enterprise and thinking through how do apply Agentic capabilities in these organizations is very top of mind. We invested in Puneet actually before the— this AI paradigm of, you know, ChatGPT and everything that we know today. And so we've been a little early to this, but knew Puneet's DNA as these advances became more prominent was really the right leader to build for large enterprises. And so with Puneet, we sat down with leaders at Delta, United, MetLife, who have all had huge success deploying Natomi. A big proud moment for us is when the— yeah, no, it's pretty incredible. Anyone could go on the United mobile app and see powered by Natomi and interact with Natomi. And then earlier this year, OpenAI recognized Natomi on one of the case studies, calling it blueprint for deploying large-scale GenTech AI. So it just had a lot of great moments over the last couple of years. And that's what inspired Puneet and me to really sit down and think about, okay, this has been proven at scale, but what's the coalition around this technology that would really allow it to be in the hands of many, many more millions of consumers? And so that's why this round brings together Accenture. No one knows enterprise complexity better than them. And their seal of approval on this company is, is a huge sign to enterprises that this is what they should be deploying. Yeah, absolutely. And then with Adobe, the idea with Adobe is, you know, many times chat interfaces sit on top of websites. And while it's a step above prior technology, the next generation is going to bring those capabilities into the digital layer itself. So not having the website and then the chat disconnected on top, but fusing these two layers. And the work that Natomi and Adobe are doing together, and we can talk a bit more about it, is going to bring about a gentec digital experience.

Sure. Tell us about the round. I want to hit the gong. How much came together for this deal? What's the total? Go ahead.

We raised $110 million. There we go. Awesome.

Yeah. Well, that shook the whole —room.

It's a good one. Look at this.

Yeah, yeah, we are here to shake the room. Fantastic. So, so yeah, all focused towards enterprise deployment, all focused towards the last mile of AI. You know, we think AI really needs a big enterprise win that's ROI positive where large companies can go on their earnings call and say, this is what AI did for me. That's what we want to deliver for the entire community.

I love it. Well, thank you so much for taking the time to come break it down for us. Congrats on the round coming together and the progress. It's at this rate.

I'm sure you'll be back on this year.

Yeah, we're excited. We'll talk to you all soon. Have a great weekend. Have a great rest of your day. Talk soon. Talk to you soon. Goodbye. We have an update from the chat. Cliff Asness has been on Patrick O'Shaughnessy's Invest Like the Best. You know who else has been on Invest Like the Best? Palmer Luckey. They've both been on the same podcast, back on, didn't know about each other. They got to do a crossover episode, back to back. If you're looking to get up to speed on both Palmer and Cliff Asness, head over to Invest Like the Best with Patrick O'Shaughnessy. You can listen to both episodes. Although Cliff was on in 2018, almost a decade ago. He's got to go back on, set the record straight on who he knows, who he doesn't know. Anyway, there's more debate over the data center thing. —some very cool renders, some very cool ideas. I think the problem is that the plans are delayed. Plans are set years in advance. Any move to actually make data centers beautiful is going to take a long time. Hopefully—

Yeah, this feels like one of those things where data center operators would say, I would love for my competitors to focus on making their data centers beautiful and really make that a part of their strategy. Well, the Zucks of the world, you know, throw up like REI tents, throwing GPUs.

Exactly. People will do the minimum viable beautification potentially. But Dan Lincoln Harris has some comments here. Love that we are having this conversation. The questions that flow from Josh's comment are why and what we should aim towards. These should be places of wonder that we can marvel at, modern cathedrals "for the Silicon Age, not to worship them, but to stoke our aspirations. Square gray boxes do not inspire." Speak for yourself. I like a square gray box every once in a while. "Given the epicenter of this technological revolution was California, the light and space movement seems a fitting direction. James Turrell is the most well-known for this movement." And he shares some James Turrell images, which are absolutely beautiful. Good inspiration if you're a data center builder. You can also just throw some trees, do some AI imagery here. Perhaps people wouldn't be so against data centers if they were built in the style of the elf city Rivendell. That'd be kind of cool. Greco-futurism. They do look cooler this way, but it looks a lot more expensive than just some walls.

So who knows if we will see any— This video— Progress. What? Just in, skin exams are getting automated. Oh yes, this is a very cool product. Okay, break it down for us.

Let's pull it out.

We have video here. Yes, uh, I think anybody that's been to a dermatologist, it's like looking for potential, um, you know, skin cancers, things like that, abnormalities, has had the experience of like the, you know, the when you're, you know, that video or the meme where the The guy's like walking through security and the security guys just like, you know, does this like really quick pat down. It's like, that's basically what like a lot of dermatologists are doing where they just like take a quick glance and they're like, oh yeah, looks good.

Yeah, they might be busy, they might be sleeping.

Come back in like a couple years. But this robot. So a robot that just does it like really, really precisely, consistently, ideally, hopefully, hopefully a lot more accessibly. I can see this making a lot of sense. And then you actually— it doesn't even replace the job of the doctor. Yeah, it just helps them speed up the process.

Uh, where is this company? What stage are they? Do they have to go through FDA medical, uh, device approval process? That could be a couple years maybe. But this seems like once you get it, uh, through, it's going to be massively successful. Um, I wonder, I wonder where they are in the FDA process. Hopefully they and move along quickly. Arc Prize put out an update on Arc AGI V3, which Tyler, are you still on the human leaderboard for that or? I don't think I'm on the leaderboard anymore. That was still early on. Like I think there's actually more games now. Roasted. Well, Arc AGI V3 has been very tough for even the most frontier AI models, But the good folks over at Arc Prize have benchmarked the two latest and greatest AI models from OpenAI and Anthropic. GPT-5.5 scored 0.43%. Opus 4.7 scored 0.18%. That's state of the art. It is, it is state of the art. This is the weird thing about these Arc AGI tests is that they start with such a low baseline, 0%. 1%. And then they start, they start shooting up as the models get more capable. But it is very interesting and reassuring to see these powerful models that you sometimes hear these narratives, it can do everything. And then, eh, there's something that, you know, to continue working for, to look forward to in the future. There's more, there's more applications here. And so they found 3 total failure modes, true local effect, false world model, wrong level of ability. Abstraction from training data, solved the level, but didn't reinforce the reward. And so the Arc Prize account goes through a little bit of what's going on when an AI model tries to play Arc AGI V3, which is basically a 2D video game, simple enough for any human to progress through, but increasingly difficult for AI models. Okay. Yes.

We gotta talk about one of the most exciting new products created. What's that? In the last 100 years. Tell me. According to Katie at Business Insider, Amazon will now create an AI podcast about their products where two AI hosts discuss the products and take your questions as if it's a call-in show. Yes. Let's play. We don't need to play this one.

This one is so rude. But you can imagine the type of products that people are generating AI podcasts for. Only the silliest things will be generated. I don't know if people want to listen to a full podcast about every product they buy on Amazon, but you have had some very strong opinions about paper towels. You wanted paper towels from like the 1950s or something. Remember this whole thing?

You didn't want— Well, I just wanted a filter to be able to buy, only shop for things on Amazon.

Well, now instead of a filter, you can listen to an hour-long podcast about every possible SKU, and then you can make the most determined decision. Determined decision available. Well, we have Alex from OpenAI. He's a member of product staff here to talk about Codex and GPT-5.5. I believe he's in the waiting room, so let's bring him to the TVP and UltraDome. Alex, how are you doing? Hey, doing great. How are you guys? We're doing fantastically.

Close to the camera. You make me want to, you make me want to scoot up. Some people come on and it's standoffish.

Yeah, it's just like a little bit of— there you go.

Now you're locked in. You're locked in.

Okay, so, uh, Give us some updates on Codex. I mean, we've seen some crazy numbers, things that seem like they're going well, but what I'm most interested in is applications, usages, just as I'm talking to friends and folks that are, you know, outside of the most intimate tech community, what are the magical experiences? What are the prompts, the usages that are seeing adoption? And how should people even be thinking about Codex and ChatGPT these days? Totally.

I mean, it is such an exciting time. I feel like last year was the explosion of agents for coding. Yeah. You know, now we bring it to everyone else. Codex now is becoming this like amazing tool for just like general work or honestly anything you can do on your computer. Yeah. So the model launch, was it, was that only last week?

I think it was last week. I think it was.

It really feels like a month ago.

Yeah. Yeah, honestly, I lose track of time, but like, you know, API revenue for that model is growing 2x faster than any prior release. Like Codex revenue actually doubled in the last week. So, so it's just like the growth is insane. Yeah. And what's cool is that obviously the growth in usage and the way people are using Codex for coding is, is just like getting way better. But I think this is kind of what you're asking for. There have been, we're making Codex great for everyone. That's all the way from like what the agent can do and also just like how simple it is to use. Yeah. And so now we have, you know, Codex being used by like salespeople, marketing people, finance people, data science people, whatever you have, whatever you're thinking. Yeah, it's like 85% of the company uses Codex and we're seeing this happen outside as well. So yeah, I'm happy to jump into some fun use cases if you want.

I wanted to ask about like how you feel the, the perception around computer use has been. Like it felt like there were so many things that got released at one time that that kind of was quite magical and you see some posts popping up here and there, but do you feel like that's getting the attention that it deserves now?

Yeah, I mean, like, so broadly what we have is we have this amazing agent that can write code and if we wanted to do more work than just coding work, it needs to be able to do anything you can do on your computer. And so computer use is this big step because it's like, hey, well now it basically can do anything you can do on your computer. But the magical part of computer use and that the team really cooked here is that it can use applications in the background. So you could be doing something in one app and the agent can be doing something in another app. And that's important because it means you can actually delegate, which means you can give the agent hard tasks that take it some time to do. So yeah, I mean, the reception around that was actually awesome. I thought we got— people noticed sort of just how hard and magical that is, more so than I even expected.

So yeah, we've been really happy with that. Yeah, I've noticed that Codex puppeteering a mouse cursor has gotten good. It feels like it crossed some sort of like Turing test where when I see a video of Codex moving a mouse around, it doesn't read to me like, oh, that's a jittery, like the AI's using it. It just looks like, oh, somebody just recorded themselves moving the mouse. Is there a benchmark? Is that something that just comes from the new model or work that's being done on Codex? Like what is going on with the progress there?

The Sky team. Yeah, the acquisition.

Yeah, yeah. So, so there's kind of like 3 things going on there. Like, first of all, GPT-5.5 is our best model ever for like general work or knowledge work. Yeah. And so it's really good at using computers. Um, the next thing is, well, what harness do you give the model, or like what information is the model seeing, and like what tools does it have to use the computer? Um, a lot of early versions of giving a model access to a computer were just giving it screenshots of the computer. Uh, but there's a lot of secret sauce in our implementation where the model actually gets text representations of what's on screen from like frameworks like Accessibility. Sure. Uh, and so the model is like much more efficient when it has access to all this information. Oh, that's interesting. Um, and then the last bit that I, I honestly think— well, I don't know if I would say it's underappreciated because I feel like people really appreciated it, but it's the level of craft that was put in to how it feels when the agent is using the computer. So for instance, you could totally just have the agent like click around on your screen and just have that be invisible. And, you know, you wouldn't even— it's just like the computer is updating as clicks happen. But the team put a ton of care into like exactly the animation that this like mouse cursor takes as it goes between the different click positions. And yeah, it was really fun to talk about that and jam on that with them. Like we actually made some interesting trade-offs. Like having this animation actually slows down how quickly the agent can work by just a tiny bit. But it means that it's so much easier as a human for me to understand the system and therefore to trust the system.

Yeah, that's interesting. That sort of goes back to, I remember the initial ChatGPT app launch on iOS had haptic feedback as the tokens streamed in. And so it felt like it was typing, and that could have easily just been generate the following full response, give you a little waiting wheel, and then boom, it loads like a web page. Like when I go to, you know, the New York Times, like the whole page loads. It doesn't stream in. But that streaming made it feel more like a conversation. Those little, like, UX cues help, like, create this more, like, interactive motion back and forth. That's very interesting.

Jordi? [Speaker:JORDI HAYS] Someone's trying Codex for the first time. You've got 1 minute to explain the 3 things that they should do to get the most out of it, what do you tell them? Okay, cool.

Is this person an engineer or are they like a knowledge worker?

Let's say not an engineer because I feel like—

well, no, let's do, let's do both. Sure.

Yeah. Let's start with engineer.

Okay. Start with engineer. If someone's using Codex for the first time, I just say download Codex, you know, attach, connect it to your project, whatever you've been working on. And then ask it a question about your codebase, like a hard architecture question and it's just going to give you an amazing answer. Then the next thing you do is you say, cool, like give it a bug that you've been trying to track down and ask it to solve the bug. And like, I hear this all the time, like on Twitter, et cetera. Like people will be like, hey, I was trying to solve this bug myself, couldn't do it. I gave it to like all the other coding agents. They couldn't do it, but Codex could do it. And then maybe as a final thing to like experience a little bit of the magic that has shipped in the last week or two, if you're working on like something that's like has a web view, like maybe a website or something like that, ask Codex to make a change and then like watch it just like iterate on that change by opening the in-app browser and sort of observing its outputs, clicking around, and then like naturally just like fix and improve the change that it made. It's like, it's super magical because you just realize when you see it like work in that loop, you just realize how powerful it is now.

Okay. Speaking of loops, explain Ralph loop, explain /goal.

Okay, so basically, um, we've had this interesting thing for a long time where people— this is like months ago— people would tell us that they knew the Codex was super powerful, but it felt annoying to work with because they had to like constantly tell the agent, you can do this, keep going, right? So you have this like brilliant model, but you have to like encourage it constantly. Um, we then shipped a feature called queuing, which you can use to give the agent a message. That it will like receive whenever it thinks it's done. And people use that to say, do this, then do that, then do that. But actually a lot of people started using that to just say, keep going. So like they would just queue like 10 messages, like keep going, keep going, keep going. Keep cooking. Yeah, exactly. And so, you know, now at this point we have these amazing models that if you know how to use in the right way, you can have them do hours of work or even days of work, just like independently, autonomously. However, You know, the average person doesn't necessarily know how to do that. You have to do like all this contrived harness setup. And so with goal, which is a feature that we shipped into the command line interface and will come to the app soon, we wanted to make that super easy. So now you can basically describe to like, hey, like I want you to keep going until you achieve this goal. And then Codex will just take care of working for however long you need until you can do that. And so internally, this is something that people have been really excited for for a long time. When we do a lot of long-running work. So for example, when you're babysitting a training run for a model, you don't want to just tell Codex to do something and then have it return in 5 minutes. You actually want it to pay attention all night. So that's what goal is. Do you want— yeah, go ahead.

I was wondering about the— what is, for the non-engineer use case, the thing that gets you to move from ChatGPT to Codex? Like the first, the first, the first like more complicated query, more complicated project that you would recommend someone say, oh yeah, like ChatGPT can do a decent amount of research, but for this you should spin up Codex and start working with that as your primary workflow. Like what is the, entry point? What is the, the, the appetizer into the Codex workflow for someone who's non-technical, not writing code, wouldn't mind if code was written behind the scenes, but really just is interacting with typical knowledge work suites. So email, spreadsheets, Word docs, generating graphics, research. There's a lot of stuff and ChatGPT satisfies a ton of that. And so when are they jumping over to Codex?

Yeah. So I kind of think about the way that you can do general work when Codex— there's maybe 3 categories of work. There's like just very easy tasks. And that's not to dismiss those. Actually, most of my usage of Codex is like easy tasks. And those are the first things you should try. Then there's like hard tasks that are like really cool demos. But actually, if you start trying to do a hard demo, you might, you might just like waste a lot of time. And then there's automating. So everyone should kind of go through this progression. It's like easy, hard, automated. And if you think about it, it's kind of like a human teammate, right? Like if you hire someone onto your team, Uh, you know, you don't just usually, well, maybe at OpenAI we do, but you don't just say like, hey, just like figure out what you want to do. Maybe what you do is you say like, hey, like, why don't you do this like small starter task or something? And then from there, give them a harder and harder task. And then eventually you say, okay, just like go figure it out and work automatically. So, okay, easy tasks. The thing I always recommend people do is like wherever your company works, at OpenAI that's Slack, maybe for you it's Teams or it's email, connect Codex to that tool and then just ask it like, Hey, like, am I missing any urgent, like, do I need to reply to anything urgently? Just like draft some answers for me. Or maybe like I get tagged on these really long threads all the time. It's like, hey, just summarize this thread. What am I supposed, what am I being asked? What should I, what should I answer? You know, just ask these like these questions like, or maybe someone mentioned something you don't know what it is. And you're just like, hey, like search all my company information and tell me what is XYZ, you know, the thing. So those are really basic, easy queries that Codex is amazing at, and I have found that even though those may not sound that hard, people just get hooked on doing that all the time. And then from there, once you're hooked and you kind of become fluent in using this tool to answer any even small question, use it for harder things. So for example, one of the use cases that I do a lot that tends to like really get other people excited when they see it is I, you know, we, we don't like meetings, uh, try to keep meeting attendees to a minimum. Here. So often instead of adding people to a meeting, I'll post in a channel and say, if anyone wants to talk live about this, just reply. And this is something I used to do, you know, from before. But now what I do is I, I tell Codex, post in this channel, see who wants to join, and add anyone who wants to join to the meeting. Sure. Super basic thing, right? Yeah. But the fact that you can ask an agent to do a thing and then monitor that post, mm-hmm, and then like just keep it up to date. Sure. It's actually really powerful. And then when people reply to the post and then the agent replies back saying like, cool, you're in the meeting, it's at this time, here's the link. That always mind blows people. Yeah. And then you start, people start trolling, you know, people like add Sam to the meeting. Oh, yeah, that's hilarious. The agent knows not to.

Yeah, because that's just like functionally what, what, if you're lucky enough to have an EA, you get something like that where it's like, hey, I need to get this thing done. I'm going to kind of set it. I'm going to, I want to set off the process, but then someone else monitoring it, kind of backfilling it.

Yeah, I think any, any work that requires like a lot of work, like a lot of time from you, but it's not hard. Those are good starter tasks. You know, same for like managing tasks, you know, tracking issues coming out of a bug bash for like pre-launch. We do this thing where I'll have Codex monitor a bunch of channels I'm in with internal and external users. And I'll just say, hey, if anything comes up, put this into, in Linear, which is where we track our bugs and like make sure it's deduplicated. And then let people know that we're tracking the bug. So, you know, very, very simple task, but actually incredibly useful time-wise. Last thing I'll say on this is we have an executive assistant kind of like plugin internally that I would love to ship externally. We should get on that. And it's really taken off internally. Like people basically have this thing that is like keeping track of all their information and just like helping them organize their day and like stay on top of things.

It's pretty cool. So then, yeah, talk to me about automation because I understand from integrations and computer use how a single prompt could go into Slack or go into your email or pull some documents together, write code, do whatever it needs to on the internet, APIs. I understand all that, but then what is the workflow to get Codex to do something every morning at 7:00 AM to prepare me a digest or take the same actions to really get rid of that rote work that I've demoed and I've seen the results, it's working. And now I just don't want to think about it ever again, forever.

Yeah, I mean, it's honestly super simple. You just tell Codex, hey, do this every morning. There we go. Right? Well, like in my case—

Yeah, I mean, that's kind of an interesting thing with these kind of products that have effectively unlimited potential is like, where's the line between like a feature and just a prompt? Yeah. And when should, as a company, when should you just launch something and make a big moment out of it? Or when should you just let people just understand the full potential of it? And then when they want something, they can just ask it, right? They don't need to like request a feature. It's just like literally just request it.

Totally. With the agent. It's really interesting as we designed the product, Like, I like to think that we need to keep sort of all the rules and like sort of the heavy UI and stuff around the agent to a minimum because it kind of constrains how much better the product can get when we have a smarter model. So the more things we can kind of like let the model decide, you know, it's like it's AI soon AGI, the better. And then sometimes you just need UI so that people can learn that this thing exists. But even that, the model can suggest things. So I'll give you the most powerful, I think, example of a use case I heard recently was Um, this person on the growth team needs to figure out like what experiments to run and then need to write code to run the experiment and then they need to analyze the experiment. And it turns out they were using, they started using Codex individually for each separate thing. So they would have it like run a bunch of analyses and they would sort of interrogate the data, you know, they would just talk to Codex about the data. Um, and then they would pick an experiment and then they would like write, ask Codex to write the code. Then they would run the experiment and they would ask Codex what the results of the experiment were, and then they would like produce a deck, right? So all steps they were doing individually, and they didn't start by saying, I'm going to automate this entire thing, because that's like hard and scary, right? They just started with using Codex like to accelerate themselves, individual productivity for each task. Then they started basically connecting all these things together into a giant skill. And then one day they just said, hey, why don't you do this every morning? And they gave it a name. It's called Lord Bottleneck because it's like solving the bottlenecks of like friction for new users. And basically, this is the thing the team does now. Every morning, Lord Bottleneck evaluates past experiments, looks at data, looks at new things, proposes some experiments, and offers to the team like, hey, like let's run these experiments. The team like picks like let's do this one or that one. Then Lord Bottleneck's like, okay, cool, here's some code or whatever config that needs to be done, runs the experiment. And at the end, they like, they go and do the same loop the next day and analyze the results. And so that's actually like really serious value in automation that's like, I forget the numbers, but it's produced like, you know, significant company value just automatically through Codex.

How are you thinking about game development? It feels like all of these tools are super useful and could really accelerate game development.

Especially with images too. Totally.

I think there was a time when like everyone had an idea for a meme but not everyone knew Photoshop. Now everyone can go and use images in ChatGPT to sort of make the exact meme that they want very quickly, very easily. And there's this takeoff in like everyone can create the thing. Now everyone can vibe code a Python script pretty easily. And it feels like the next thing is like going and shipping something. And we're starting to see these like simulators and these little one-off like web games. It's still a little bit, prosumer, I would say. But is this something that you think just is like a natural outgrowth or something where you might actually spend some cycles thinking about that as a particular workflow?

Yeah, I mean, I think I'm really just excited about how easy it's becoming for everyone to bring their ideas to life. ImageGen, you bring up ImageGen, that was actually a really big deal. In a way that I still think people don't really realize. Like, having an agent able to just generate images as it's working is massive. Yeah. I mean, ImageGen 2 in itself is actually really popular and is great even outside of Codex. Like, I think, um, let me see here. Yeah, like, I think we are pretty— like, people are making like 50% more images with ImageGen 2 just like a few weeks after launch than they were before. So it's like really taking off, and it was already popular. But then with ImageGen, we see all these cool use cases where people will have the agent, let's say we'll take a development and a non-development use case. For development, they'll have it like build a bunch of sprites. Sprites is like a term for like the image that you use in your game. Yep. So like one cool example of this is like in the GPT-5.4 blog, we have some example prompts to try. And one of them is like, I forget, it's like build like an art roller coaster simulator or arcade simulator or something. And you can just copy paste that. Prompts into Codex now, and it's just like even better because Codex will like create all the assets for the game using ImageGen and then like build a game using it and then play the game for you to test it in the in-app browser. So yes, this is happening very naturally, and I suspect it would be quite interesting to lean into it to make it even better. Yeah. A cool like non-coding use case is we're seeing people use ImageGen to create slides or like slide templates or assets go on slides and then produce the slide deck, which again, you can iterate on just inside the Codex app. So yeah, images are massive for us.

Yeah. Yeah. That makes sense.

Jordi, anything else? No, this was great. Appreciate the update. I'm sure there will be many more, a ton more news by this time next week.

We're joking about like how many Thursdays can we ship in a row?

You know, like, yeah.

Well, the more you ship, the easier it is to ship. So it's That's good feedback, Luke. That's fantastic.

Thank you so much for taking the time. Great to see you, dude. Great to see you. We'll talk to you soon. Have a great weekend. Goodbye. Should we show everyone the game that we made? Should we show everyone TBPN Simulator? Do we have that pulled up? Can we play that? Would now be a good time?

It's late at night. We might not have enough time.

We can have them pull it up while we read through the—

Let's pull up this packaging of— The Neo robot over at 1X. The Neo robot over at 1X. Have you seen this? Oh yes, I did. They got a suitcase.

They got a suitcase that is remarkable. Let's pull this up.

This packaging is so cool.

This warm tone, it's welcoming.

I got to say, it's really cool until it's 2 AM and you hear shaking in your house and you go and you turn on the lights in this little, like, you know, suitcase thing shaking and then Neo pops out and goes Terminator mode. I wouldn't know what that actually is like, but I'm gonna watch Terminator this weekend and report back.

Ridiculous. Wow. Amazon, Google, Microsoft, Meta collectively are spending more money than the Manhattan Project every single month. That's insane. Wow, wild. Um, anyway, in the mansion section, uh, there is a record-breaking home in Los Angeles with a question: will a Los Angeles area mega mansion become America's most expensive residential property? $400 million, they're asking. 70,000 square feet. The, the guest house is 30,000 square feet. It's absolutely insane. 39 bedrooms, 8 acres, 3 pools. It has an X-ray machine. This is in the mansion section of the Wall Street Journal. Fascinating, because Tyler was unimpressed. And I was wondering, like, what could possibly be better than a $400 million mega mansion in Los Angeles? But you found something. What, what absolutely— I was comparing it to, you know, some kind of palatial manor, right? So clearly it's like Versailles. Versailles, you know, 10x bigger. It's like, it's like 700,000 square feet. 700,000 square feet.

That's just the indoor space, obviously, with the gardens, it's like millions of square footage.

Millions of square feet. Kind of just getting brutally mocked. What about on bedrooms? This one is 39. You certainly can't go higher than 39 bedrooms. I think it was, yeah, 2,300. 2,300. Oh no, no, that was just rooms actually. That was rooms. Okay, okay. So not all bedrooms, but I mean, if you need a buddy who crashes, like they can crash in any room, right? Like, you know, have you ever slept in the kitchen? You've never had, uh, We don't have any room. I've ever— you gotta sleep in the— you gotta sleep in the living room. You gotta sleep in the dining room. Lucky.

Uh, let's pull up the trailer for TVPN Simulator.

Oh, we have a trailer? Yeah, we got the trailer. Oh, let's go, let's watch this.

I think TVPN needs a game.

Yeah, we definitely need to build some sort of game. This, this project has evolved so, so much to the point where you can You can move the goalposts. I don't know that you can interact with the horse. You can definitely hit the gong, and then any team member is now playable. Okay. Oh wait, so there is a mechanic. You have to gain the most subscribers. Interesting. Yes, you have to puzzle piece everything. You can watch the stream in the game. You can explore the UltraDome. It's a full backup replication. Tbpn-simulator.com. That was ImageGen 1. ImageGen 2 would never do that. But can we pull the actual game up? I hope so. Hopefully. Yeah, here it comes. Didn't pop up. Um, if we can pull up the actual, the actual game, uh, let's, uh, let's see. There's a whole bunch of good stuff on here. Okay, so this is tbpn Simulator, the latest and greatest version. A remarkable— how, Ben, how did you actually get this, like, the geometry in here? Was this all described via prompt? We've got intern Ben here to talk.

Yeah, so basically the whole process is kind of like, I'm like, okay, look at that wall right there. Yeah, I need that to be taller. Okay, place a desk right at that wall. So the whole idea was that I placed every single object, so I was able to like Is there someone else in the game?

Wait, it's multiplayer? It's multiplayer? Wait, so we're getting other people in the game right now? Oh no, people are coming in! There's Mark! gbpnsimulator.com. I can't believe it's multiplayer. I can't believe you created a multiplayer game. Okay, so we got the bathroom. Okay, this is incredibly— no, no, no, no, get out of there, get out of there. Okay, so, uh, there's a chessboard which was gifted Ring the gong. You can ring the gong.

To the actual studio. Okay.

Oh, gong cooldown. And then those are the photos that we take. And there's the UltraDome.

It's a 1-to-1 LiDAR scanned measurement. Did you—

wait, did you actually LiDAR scan it? I LiDAR scanned it. Wait, no.

Yeah, the measurements. And you could play as any character. So yeah.

Yeah. Okay. So what does it take to make it in TVPN Simulator?

Camera cut. Yeah.

Okay. So you're watching the show and you need to cut between— oh, so you get a current cue and you have to cut between different cues on at the right time to produce the show live. Okay. We have Tyler, we have Mark in TVPN Simulator. Okay. What else can you do here?

How much activity?

Like 35% of the— Is this from yesterday? Oh, this is yesterday's show. Is playing currently. So they're using the wallet. And if you're Jordi Hays, for example, like, you play as, uh, you have to do a mini game which is Guitar Hero but with soundboard, basically. Is that the game? Let's hear. Overnight success. So you have to, you have to hit the soundboard at the right time as the blocks come down. Okay, things like— I think I was doing some ad reads for a while. There's a gong, of course. We're really packing the UltraDome. We'll see if this thing has high throughput networking. Uh, you can play as the guest. What do you play as the guest? You just see what it's like to be a guest on TVPN. Okay, you get— someone's moving the goalposts. Okay, we got a new definition. I'm moving the goalposts. These are general intelligence.

I didn't realize.

Yeah, the, the chat has joined the the game. Oh yeah, someone says make it a Roblox game. This does have some Roblox aesthetics, but this was not— this was vibe-coded. And so this is creating the daily newsletter. You have to stack the different blocks, the strips. It says perfect alignment, published clean. That's Brandon Garel's job. Nick has to, of course, fix the schedule, get everyone to come together. We're watching some Tetris. You're trying to calendar Tetris How are you going to get Patrick Collison and Brian Chesky on the show with Alex Karp and Mark Benioff all at the same time? How will you do that? Paul Coogan. There we go. And then this is Tyler organizing the show, figuring out how to get different posts into the show. Okay. Oh, here you go. You're moving the different— oh, you're jammed up there. You're jammed up there. You got to keep working on that. What else is this? What is Jackson up to? Making cards? What are we doing here? What? Oh, whack a clip. Okay. The timeline sniper. There's breaking news. You got to turn. You ramble. You don't want to cut that. Okay. You want to, you want to hit the good hot takes at the right time. That is fantastic. I cannot believe the number of minigames here. This is insane. What a remarkable— Oh, this is me. Okay. So I have to read different phrases. It's a, it's a typing game. I can— because I have to read without a teleprompter. Live from the TVPN UltraDome. Uh-oh, someone's typing incorrectly. Phrase deck. Okay. Well, we got a whole bunch of people. You can go enjoy it. Hopefully people are playing TVPN Simulator all weekend long. Yeah. Yeah, I expect people to put up— I mean, if you're launching a TVPN type show, this is a great way to get your reps in.

Get some practice.

Get some practice in before we go live. Until we get our mastermind. Yeah, it is the path to financial freedom for a lot of folks. So we recommend it, we endorse it. Get your reps in on tbpnsimulator.com and we will see you on Monday. Anything else, Jordi?

Is it actually live? I think it's live, yeah.

Wow. Oh yeah, people are in there. They're playing.

We gotta change the slop images. Okay, we'll work on those. But a lot of this—

Incredibly well done, Ben. Our first video game truly shipped.

Truly hilarious.

And people are asking for a wave pool.

That's true, we do need a wave pool.

We don't have a wave pool in here. Well, that's not a simulation, but we—

if, if we make it in the simulator, yeah, it's probably increases the likelihood.

I think, I think the key is you have to be able to sit down on one of those laptops and play an even rougher version of TVPN Simulator 2 Layers Deep. That's how you simulate Ben. That— oh no, yeah, yeah, yeah. When you simulate Ben, it should go into a 2D version, top-down, low, even lower-res graphics, pixel art. Yeah, yeah, just a pixel art version of TV Ben simulator, 2 layers deep, and we can see how deep the rabbit hole goes. But will you break out of the matrix? We certainly hope you do this weekend. Go enjoy your weekend, enjoy your Friday. We will see you on Monday. Goodbye. Love you. Flashbang!