Windsurf's Wild Weekend, SpaceX Invests $2B into xAI, Zuck's AI Data Supercluster | Jeff Huber, Scott Wu & Jeff Wang, Carl Pei, Garrett McCurrach

You're watching TVPN. Today's Monday, July 14, 2025. We are live from the TVPN Ultra Dome, the temple of technology, the fortress of finance, the capital of capital. We have an amazing show for you today, folks. One of the best shows. It's the best day to be a journalist. It's the best day to be a live streamer.

It really is.

It's the best day to be a commentator.

That's right, yeah.

It's not a story.

We're in the commentary business.

Commentary. Commentary is. Yeah. That's not a slur.

A lot of people asked us to not yell into microphones. We cannot help it. We just get a little too excited sometimes.

But massive weekend. As soon as we got off the stream on Friday, the news broke that Windsurf was being sort of pseudoaclotic.

It was a trade deal.

It was a trade deal.

It really was a trade deal.

I believe this broke at 3pm Pacific, which is just after market close.

Just after we went off air.

Yeah. Which one were they targeting? That's the big question. I think they were more worried about the TPP and coverage than whatever the market had to say about the deal. But the deal was that Demis Hassibis here has a post. He says. Very excited to welcome Windsurf AI founders Mohan and Douglas Chen and some of the brilliant Windsurf engineering team to Google DeepMind. Excited to be working with them to turbocharge our Gemini effort on coding, agents, tool use, and much more. Great to have you on board. And so this. This was received sort of as like.

Oh, okay, so here's kind of the timeline, as far as I'm concerned. So immediately, our initial reaction, we were here at the studio, we were off air. Our initial reaction was, okay, another one of these quasi acquisitions.

People call them shellcountres.

Yeah, everybody's gonna take care of. It's fine. And immediately, a lot of people started kind of freaking out on the timeline, saying that the employee. Employees had been screwed. Our immediate reaction was, this doesn't make sense.

Why would we. Stupid.

Why would all the parties from the founders to the investors to Google, tolerate something like this? And then call it 24 hours later, I was still getting messages from employees that thought they were getting total zeros. And at that point in time, I started kind of reevaluating, reevaluating what may have happened. And there was, I think, to be.

Clear, the current employees, they saw the news and they didn't get a check, they didn't get a wire transfer, and they also didn't even get an email saying, hey, we did all the calculation and based on the share price, you'll be expecting this much money. It might take a month for it to clear, a couple weeks to clear, but here's your expected payout. There was nothing like that.

People apparently were crying, crying in reaction to the news. Well, here's the thing. This was a unique company, right? So Windsurf launched In November of 2014, the product, or sorry, 2024, the product. But if you had joined a few months prior to launch, worked on the lead up to the launch, launched it, seen this incredible growth, and then scaled to what you originally thought was a $3 billion exit to OpenAI and then you find out that you're actually, you're actually going to continue working at the company while dozens of your best engineers and your founders go and leave effectively to then go and compete against you right in Cogen. And that roller coaster is just absolutely insane. And so really unique dynamic. That was different than a lot of these other the character, AIs, the scales, et cetera, where in those situations many of the employees had been there long enough that they had vested a meaningful amount of shares. They had certainly reached their cliff in this situation. You could have joined in August of last year, not hit your cliff, not.

Hit your one year cliff, but been there for the entire run of Windsurf as a product, like from zero to what, 40 million ARR was reported. So yeah, I mean, you watch the launch and you're like, cool, I got it right. I picked the correct, I picked the correct startup, I picked the correct rocket ship. I'm on board. Yeah. In Silicon Valley, you know, these options, these stock option contracts, they're kind of like lottery tickets. But my number came up. And then you go to the gas station or something to cash it in and you're like, wait, we don't know how much it's actually worth. And it's not just like a tax issue. And so, yeah, a lot of people very upset. And we'll get into kind of some of the debate around this. So what's interesting is that I have Varun Mohan, who is the CEO of Windsurf. He has not posted at all still. The last post he posted was on June 20th. So I think maybe it was a little bit more recently.

Yeah, to be clear, he was completely muzzled.

Yes. Yeah, I mean, that's what we believe to be true, but it was weird.

If he's not muzzled and he's just.

Chosen silence, he's Terrible. Y. The last time he really engaged with anything on X, at least was when.

Wait, John, I have to interrupt because it's now live. Cognition has signed a definitive agreement to acquire Windsurf.

Let's go.

The acquisition includes Windsurf's IP product, trademark and brand and strong business. And above all, it includes Windsurf's world class people.

No way.

We've privileged to welcome our team. We are also honoring their talent and hard work in building Windsurf into the great business it is today. This transaction is structured that 100% of windsurf employees will participate financially. They will also have all their vesting clips waived.

What?

And will receive fully accelerated vesting for their work to date.

Let's go.

At Cognition, we have focused on developing robust and secure autonomous agents, while Windsurf has pioneered the agentic ide. Devin and Windsurf are a powerful combination for the developers we serve. Working side by side will soon enable you to plan tasks in an IDE powered by Devin's code base. Understanding, delegate chunks of work to multiple Devins in parallel, complete the highest leverage parts yourself with the help of Autocomplete and stitch it all back together in the same ide. Cognition and Windsurfer united behind a shared vision. It makes so much sense for the future of software engineering and there's never been a better time to build. So there's a video up between Scott Wu and Steve over at Windsurf and we're hoping to have them on the show later today. If we can coordinate.

That'd be awesome. Yeah, that'd be great. Cool.

So anyways, this. Apparently Scott got to work Friday and not only negotiated, I mean, to do this kind of deal during all the chaos of this other deal that's happening, and then put together a deal that delivers a win for every member of the team. And the other thing that's interesting here is historically Cognition had, I wouldn't say over invested in engineering, but they were heavily, heavily skewed towards technical talent. And the people that got left behind at Windsurf were less economical sometimes.

Yeah.

Enterprise account management, gtm. Right. And I, over the weekend I'd heard other teams that were kind of circling, wanting to hoover up their GTM team, but the fact that they were able to get this deal done in just a couple days is fantastic.

This is incredible. Wow, what a fantastic end of the story.

Or a beginning. New beginning.

Yeah, new beginning, New beginning for sure. I don't even know if we should go through the full history here. We can kind of go through whatever. But I mean, you kind of know the end of the story. But I think we should take you through some of the debate that, that rose up and there's still the question of like, even though it feels like everyone got the good ending, there still is this question about like, we're in this weird zombie structure era where you can be left in limbo. What is the cost of that? What is the, what is the risk to that? Do we need new employee contracts? Do we need new understanding of how, of expectations? Because it's one thing for a CEO to say, hey, I'm, I'm building, I'm trying to build something really cool in Silicon Valley. I'm going to build a tech product and I want you to join me as a salesperson, operations person, an engineer, whatever. And we're going to go on this journey, we're going to go really hard and we're, we want you, we all want to be aligned with like thinking long term. So we're doing a four year vast with a one year clef. Pretty standard. And the, the expectation at least the history has been like, like you will have clarity throughout that process. And now that we're in this weird, weird regime, like you're getting like, okay, the last, the end, the end of the, the last chapter of the book could be really dramatic, but we got the good ending. And the question is like, has there ever been if this, this deal kind of exposed the risk of like the bad ending and that it is possible that that could happen? It didn't in this case. But the big question I think people are debating is like, how much of this is, is due to the chaos. How much of the chaos is a function of FTC and, and the antitrust regime. And Ben Thompson talked about that a lot versus just the differences in communication styles of different mag7 companies. That's interesting. The experience of CEOs, like you look at the way scale I handled it. Alex Wang is, he's a generational communicator. He's been on the Ovan.

Yeah, different Dynamics. Great guy, but suck lead and acquisition. Totally having that sort of founder mode confidence. Totally fully in control of the company. And everything we've heard from various parties says that Varun was muzzled. But again, everybody involved should have been saying how do we avoid making it so that hundreds of people don't feel like they've been completely screwed even for 24 hours?

Yeah, and the evidence was overwhelming at first.

I was, people were messaging me Friday and Saturday morning saying the Windsurf Employees got screwed. And I was messaging them back and being like, you're misreading this. This is the standard deal that they've always done. And then the overwhelming amount of more information from high quality sources saying, no, they're screwed. Yeah, it just flipped. And I said, okay, maybe. Maybe they are right. And the key. So now we don't. We had this whole other kind of tangent that we were going to go down, which is like, what happens to the ghost ship? Right.

Yeah.

It could potentially be this, like, lord of the flies dynamic where there's now hundreds of people that are still working on Windsurf, but the competitive pressure is insane. You're losing a lot of your top talent and your founders. Are you going to be able to keep momentum? Or is Churn going to be crazy?

And.

And then do people try to stay a long time so they can get, like, a distribution? Or do some people leave because they want to pursue other opportunities? And now we don't have to debate that, because Windsurf is home with cognition and Scott Wu and, you know, the team over there. What's that, everyone?

All those employees who are now onboarding to Cognition.

I know exactly what you're gonna say.

They're getting Ramp cards, baby. Because, Scott, time is money.

Save both.

Save both. Easy. Use corporate cards, bill pay, accounting, and a whole lot more all in one place. Go to ramp.com to get started. And, you know, cognition has a huge lineage with the Ramp team. And so we're very. We're very excited for Cognition. We're very excited for Ramp, and we're excited for all the Windsurf employees who landed well. This is funny because there's so much that we should. That we could go through. I think we should go through a little bit of the timeline of, like, what actually happened, how people, how the debate evolved, because it is very interesting. So the context of, like, why is Windsurf selling at all? And I think this big question, the first one is, let's just say, Hypothetically, you owned 100% of windsurf last week, and let's say a completely financial investor comes to you. Not if there's no even, like, big tech dynamics. It's like Warren Buffett, he calls you up and says, I want to buy 100% of your stake.

Warren needs. He's like these agentic ideas. I need it.

I need it.

I have to have one.

I have to have one. Yeah.

I want to buy a great business at a fair price. Let's do a deal.

So that's the question. Is like, if you owned 100% of Windsurf Jordy, would you sell at $2.4 billion last week? And I think for most people the answer has increasingly become yes. And there's a few reasons. So the last big post that Varun, the CEO of, of Windsurf, had interacted with on X was this post by Technium who says, Windsurf wrecked. And it's a quote of Nick saying, breaking Anthropic just pulled the rug on Windsurf. Windsurf will be cut off direct. Will be cut off from direct API access to Claude in less than five days. 3.5 Sonnet 3.7 Sonnet 3.7 Sonnet thinking windsurf never even got direct access to Claude. 4. Anyways, and so this has been the question of, you know, how much, like, if you are in this, like, rapper business. I think it's, I think it's a great product. I think it's, I think it's great. I think I'm actually very excited about, about wrappers. But there is this, there is this anti, there is this competitive dynamic with the underlying foundation lab and Anthropic does seem to be more aggressive about where the value accrual will happen. So, you know, if it's, if it's 50, 50 by default, all of a sudden Cursor and windsurf start pulling 80% of the gross margin dollars from what they're adding on top. Well, then the underlying foundation model is going to say like, hey, let's pull this back. Let's make sure some of those credit cards live in our Stripe account and you're not just paying us big API fees. And so there was a big debate about this. Cursor had to go through it, Windsurf had to go through it. And Varun responded to this and had a good analysis and was basically making the case that, like, we can get through this. And I think that's true. I think there's totally a world where it gets true. But you were saying earlier it is an extremely competitive market. They were not the number one product in the market that's Cursor and by most accounts. And so they were in not a super tough spot. It's a great growing market. It's developer tooling. How many database software, enterprise SaaS products are on the public market? There's like, there's like, like the seventh one is probably worth 10 billion. Right. So it's not like they couldn't have gone higher than 2.4, but 2.4 billion feels like A great deal for a company for a product that's 1 years old, 1 year old for a company that's 4 years old and a hyper competitive space with product that launched less than a year ago and your direct suppliers are trying to come for you and getting aggressive. And so this went on Hacker News. Not sure how much you should read into it, but they said Web Serve is absolutely dead in the water. And while Cursor hangs on for now, all value is going to go back to the models, says Jake. In my opinion, other than the Microsoft IP issue, I think the biggest thing that has shifted since this acquisition was first in the works is Claude code has absolutely exploded. Forking an IDE and all the expense that comes with that feels like a waste of effort considering the number of free slash open source CLI agentic tools that are out there. Let's review the current state of things. Terminal CLI agents are several orders of magnitude less money to develop than forking an entire ide. Cloud code is dead simple to onboard use whatever IDE you're using now with a simple extension for some UX improvements, Anthropic is free to aggressively undercut their own API margins and middlemen like Cursor in exchange for more predictable subscription revenue and training data access.

Yeah, so, so, so the, the other context here is this year Anthropic, you know, windsurf went from zero to somewhere around a $40 million run rate. Anthropic this year has gone from the beginning of the year to 1 to 4 billion dollars. And so they, you know, however you want to say, they've been running, running away with the market. And Cursor's, you know, revenue is equally incredible, although starting from a smaller base.

And so this was some of the like, okay, things might be getting tougher from here on out. Yeah, I don't know that I fully agree with that. But, but people were saying this and then we covered this last week, but the meter team did these evaluations where they ran a randomized control trial to see how much AI coding tools speed up open source developers. And what they found was that developers said that they expected products like Cursor and Windsurf to improve them by 20%, but they were actually 19% slower. And so you have this weird dynamic where okay, maybe the future is like, so I don't know how real this is. I think that we could totally be in a world where CURSE is incredibly valuable for less experienced developers working on more vibe coding projects than like pushing the frontier of really advanced projects. And Then also just the next version of these tools could be really cool and solve this problem. But there was some worry that maybe developers were going to take a hard look at their AI coding IDE stack and say, wait a minute, like, I'm kind of using this too much. I'm leaning on it too much, and maybe I should just go back to the old way. So there were some, There were some. There were some reasonable points around. Like, maybe now's a good time to exit, right? Then. So, so after. So after that, the question becomes, when you sell, you have. The board has a fiduciary duty to shareholders. And the big question that I think everyone was actually debating was this concept of was the windsurf deal with Google a Pareto improvement for all stakeholders? Which means in a Pareto improvement, basically, you can think of, like, the baseline is how everyone's doing. In a Pareto improvement, some people might go up a lot, other people might go up a little in value, but no one is going down. Not a single person is going down. This is the goal of, like, all political legislation. You want to create things that are not just positive sum in the sense that, like, you know, I tripled my net worth and you only lost half. Like, that sucks, right? You don't want that. Pareto improvement is like, I tripled. You doubled. Yeah, I won. But we both won and we're both happy, right? And so in a good acquisition deal, everyone is. It's Pareto improvement. Everyone gets more than they, than they, you know, had before. Everyone gets paid out. Then the flip side is something called Caldor Hicks improvement. And a Caldor Hicks improvement is where some people go up, other people go down, but when you sum them, it's still a net improvement. So very few people were saying, like, you know, the VCs, even when you add up the VCs and the employees and the founders, and it's a net loss for everyone. Very few people were making that argument. And to be clear, the employees got hosed.

It seems like everybody would have done better on an open AI. Everybody would have done better on an OpenAI acquisition. And so when the 2.4 billion number came out in on the evening of Friday, it seemed like, okay, not as great, but still, everybody's going to do well. And then 24 hours, 24 hours later, it was, oh, wow. Like, this might have been structured in a very funky way that, you know, people were. I won't report the number, but people were messaging me information on how much Varun personally made, and that was still in the context of, I mean, you.

Can just do the rough math. Founder stake, couple rounds of dilution. How much do you think he made? Right? Like it's not rocket science to figure this or to at least ballpark it to the right order of magnitude. Right. But you build a great product that gets some traction, that solves some problems and you build a great team and big tech wants you like you. You know, I'm not upset about it, but the question was like, what's fair for the employees? What's fair for, you know, all the different stakeholders? And so a lot of people kept coming out with really interesting stories. So William Allen here has one about Scott Belsky and Behance going to Adobe. We should read through.

Yeah.

So William Allen says, I don't think he's ever told the story, but it's worth telling. When we were selling Behance to Adobe many years ago, Scott Belsky made a spreadsheet of every employee, 32 of us at the time, and personally negotiated each person's title, salary and incentive structure and made that part of the overall deal terms. I'm going to choke up. It's so virtuous. I heard the phone calls where he went to bat for each individual. He not only, he not only didn't have to do this, but it actually complicated some of the other factors in the deal. Of course it changed the trajectory of so many people's lives, including my own. Two years later, 100% of that original team was still at Adobe. Even today, a dozen years later, many of the core members are still there building. I was inspired by it then and I'm inspired by it now. And so I think that was a good, a good case study in like, the founders do have a lot of hard power in the sense of board control and actual voting shares. Almost every founder can veto a deal, which gives them leverage to build that spreadsheet and actually decide who gets what. And then on the flip side, they have a ton of soft power because you can't really do a hostile takeover of a private VC backed company where the founder has board seats and stuff. It just doesn't work. And so it's incumbent upon the, the CEO to actually take that step. And that was the fear that Varun didn't do that, that he didn't go far enough.

Yeah, there was another, there was another example from Parsh Ilya Sukar's company. So one of his former employees says, we got news our startup was being acquired five days before my one year cliff. I thought I'd get nothing. Our shares were accelerated to 100%. It was a life changing event and I'll always appreciate that our founders did the right thing for us at parse and said Ilya. Ilya quoted it and said 100% acceleration for everyone but the active founders. We went backwards and re vested our vested. Glad someone remembers.

So.

Very heartwarming.

Yeah. And so people were, people were debating, you know, did like was the core team taken care of it feel. It felt very rough. But then pretty quickly it came out that so we were going back and forth on this on Saturday and the core debate, like your post went pretty viral. You got 2 million views, 4,000 likes. And it definitely struck a chord. Folks were debating this one line. From what I've heard, the employees are getting screwed regardless of their vesting status. And it's like.

And that was because people had. People were on a four year vest.

Yes.

But they hadn't hit their cliff. Even if they had joined 11 months ago and had again they had worked on the product up until launch, launched the product, scaled it, thought that they were getting this three massive outcome in which again some people would have not been able. I'm sure OpenAI wouldn't have kept everyone right.

Totally.

I'm sure that they would have made a bunch of people effectively revast.

There's always layoffs post merger acquisitions. Like that happens all the time. That's not a crazy thing. So certainly reasonable. I mean the other question is just like, you know, if you worked at this company and you were like I like working at this company, I really would love to work at Google. And you don't get carried along like that is getting screwed. Like not getting a job at Google in some ways is getting screwed. Even just the ambiguity over the weekend is a little bit of like a screwing. And so I still will defend your initial characterization, but I thought that there were things that could potentially shift the narrative. And so I quote tweeted yours with the Steelman argument, which is one.

And we tried to get a knight in shining armor suit. We were getting the steel man suit ready but then we knew we didn't have to.

Production team will get the steel man suit of armor eventually. We need to order from a prop shop. We take, yeah, it's on the way. We're working on it. Yeah, okay, we're gonna. And just make sure it's great when it comes. And so I had three points. One was that employees who haven't reached a one year vesting cliff don't have that strong of a Claim around I built this with sweat and need a liquidity event. And we don't know the 10 years of all the left behind employees. So there is an argument that like if you joined three months before, it was already like really big. Like should you really get accelerated four.

Years and like, oh, and I never would have made that argument totally. But it was a fact that you could have worked there for 11 months and gotten zero. And it's like, no, you're getting a ride on the ghost ship now. Yes, ride the ghost ship into the sunset. And that was in the context of the founders and this is why people in group chats and DMs were getting so angry because it was in the context of somebody that worked for you for 11 months was potentially going to get zero. Now that's not the case now. And then you're walking away, effectively defecting and taking a multi hundred million dollar payday.

It is a weird dynamic with these startups because clearly I don't know exactly what Windsurf's headcount was. I think everyone was saying it was around 500 at the time of this deal. And a year ago, pre pivot, what do you think their headcount was? Couldn't have been more than 50, right?

Definitely not.

So you're looking at a 10x increase in, in employees who haven't really been like they've been on this rocket ship. So I think, I personally think they totally deserve liquidity and accelerated vesting, which it sounds like they're getting and it sounds like they're going to do great. But, but, but, but it is like not the nature of the initial contract, which is like one year cliff for your best. Like if it, if it came out that it was like, okay, they're getting paid out to the tune of, you know, how long they've been there. I don't think anyone would be like that upset. But then somebody was making a good argument that basically like you, when you sign up with like a four year vest, you're saying, I'm going to keep working for four years if you sell the company and put me on some zombie ghost ship. Like I didn't bail. You bailed.

Yeah.

So you, so you should, you should accelerate me. And I actually think that that argument.

Yeah. And the other, the other dynamic here was so when character AI, when, when the character AI founders and part of the core team went back to Google, the character AI became a ghost ship. But it was cool in the fact that they had 20 plus million users, they also had a Hundred million on the balance sheet and it wasn't this like insanely gnarly enterprise market where and I know people on the character team and they were like it's pretty cool. We have a hundred million dollars, we have tens of millions of users and we're just kind of going to experiment to try to find, you know, figure out how to monetize this. And I just don't think that that was, I don't see the code generation market playing out in the same way.

Where I was trying to think about that. Is there a world where if the ghost ship had stayed a ghost ship and they'd pivoted, is there some world where they niche down? There was this rumor that they were going to niche down into B2B. Could they have niched down windsurf into some niche category and actually had a pretty durable business we were talking about like could Jeremy Giffon make it work if he acquires it and like turns it into some cash flowing business? I just wonder like is there, is there a niche where like, I mean we've heard, the example I would give is like we've heard that like these code gen tools are like particularly bad at like rewriting old programming languages like Fortran for example. And so if you created like Cursor just for like rewriting Fortran it's like that's not sexy at all. But that's not going to be hot on Twitter X. But it could be really valuable in like you know, the bowels of like legacy banking and like you could pay, you could charge a bunch and you wouldn't like the TAM wouldn't be very big but the business could be quite good. And that might be a weird outcome, but it could be a silver lining. I don't know. Well anyway, the second point that I made was that, you know, I said the second the real culprit might be FTC antitrust. This is a hyper competitive market and Google still feels like they can't just do a normal acquisition. And I thought that was crazy because it's like this is textbook. They should be able to acquire this company. Right? Like the cursors out there, they're not acquiring the market leader. Yeah, OpenAI is clearly going after this market and to the tune of what's their valuation? Like 350 billion or something like that.

Like, like we haven't, we should try to get an update there. But like I think the Masa, I'm not sure the full moss, it's not.

Even 1/10 of it's greater than one tenth of Google's market cap. Right?

Yeah.

And so you're looking at like. And then is Microsoft going to compete in this? Absolutely. They have.

They already do.

They already do. And so yeah, it's like you're, you're basically like, I don't know, second, third, fourth. In this market there are four very serious players. Like this should be textbook rubber stamp. Good, good to go do the deal. No problem.

There's also a, there's also a long list of founder led companies that were, that are maybe, that were maybe less hyped than Windsurf, but have incredibly talented teams. There's like poolside.

Yeah, poolside.

There's. There was that one that Nat Freeman and Daniel Gross did. That was magic.

Yeah, that was the magic.dev they were found. Apparently they were training their own foundation model for a while and that was kind of like, you know, very capex intensive and maybe slow down some things in the product side, but still like, clearly it's competitive market, everyone's working on it. Also we're missing out. Like there's probably a ton of YC teams they're working. There was actually one YC team that got in trouble for forking like cursor.

Too hard or something a little too hard.

Yeah, they crossed the line there. I think they did fine and I think that they wound up like cleaning it up. But people were upset about like when you fork something you have to abide by the MIT license or whatever the underlying license is. And so you can't just like steal open source code and not abide by the, by the open source rules. And then, and then I said three new facts might come out, which they did. And, and the FTC thing was interesting because there was a little bit of pushback on that. People were saying like, oh, like Lina Khan at my lunch, like Lina Khan's like responsible for everything. And I wasn't, I wasn't necessarily saying like lina Khan was 100% responsible for how this was, was, was structured or like how this played out, but she did kind of like warp the road a little bit. That made it a little bit wonky to drive.

Yeah. Basically we made hardcore capitalism illegal. So we nerfed capitalism. And then, and then capitalism ended up coming in and saving the day.

Yes. Yeah, yes.

With Scott Wu.

Yeah, Seriously. Seriously.

So the solution to capitalism.

Capitalism is more capitalism. This is the Brad Gerstner point. Yeah, totally. And so I still think that there is a way that even with the weird new FTC antitrust regime, Google and the Windsurf founding team could have probably been a little bit more aggressive, taken a little bit more FTC risk and avoided this whole PR dust up, but that would have been more risk for Google. And I think that the way they did this here, where they, I mean.

It'S a great outcome for Google. It's like we didn't buy the company. Look.

Exactly.

They bought the company.

They bought the company. Exactly. So it really is a great outcome for Google. Even though I'm sure that their comps team was up all weekend calling people. Yeah, Like, I have to imagine that.

This was like more room for them. It was like serving, you know, it was the best possible story that the New York Times could have ever asked for.

Yo, totally.

It was like big tech.

Yes.

Buys hot company that was competing, you know, not. Not necessarily. Like the whole point of Lina Khan's thing was that, you know, big tech is like squashing competition by, by, you know, buying these companies early. And so you had the story that that was served up to them, which was, you know, exciting company in Cogen, a lot of traction. OpenAI almost buys them. Google says, okay, we're going to buy them. That's Google. Google and OpenAI are arguably, you know, you know, the two competitors that actually, like, matter in search right now. And so it was just the perfect story. And then the, you know, I think Scott over at Andreessen had a good post that I, that let me pull up here for a second. He was basically saying, there is an irony that Lina Khan's activism has resulted in deals that are far worse for everyone except the people she was targeting.

Wait, who tweeted this?

Sorry, Martin Casada.

Oh, yeah, Andreessen.

Yeah. And yeah, it's like the people with the lowest.

It's crazy.

It seemed at the time that the people with the lowest leverage, the employees that had joined recently, were getting the worst outcome out of a deal that should have benefited everyone.

If you're looking for code review for the age of AI, head over to graphite.dev Graphite helps teams on GitHub ship higher quality software faster. You can get started for free@graphite.dev so my question is Google, it feels like PR nightmare, accidental PR nightmare, because they had done what character just fine. It was not that big of a deal for them. Not a big. And then scale had.

Well, and it was very clear that that situation was interesting too, because Google has delved into social in the past.

Hasn'T had the best time not beating the ChatGPT script.

No, that was Intentional, of course. No. But they've messed around in social. You could tell that Google doesn't want to own the AI. Totally, totally. They just don't want the associ.

I mean, there was a rumor that Gnome didn't want to be in that he kind of like, was just like, I want to train a great model. I want to train a great model. I want to do AI research. And then that became the use case. They found product that fit. And then he was like, I don't really like.

And even today, you see what Elon's launching today with the. I don't even. I won't even say it on the show, but his new launch today basically signals that that is a very important.

He's interested in that market.

Yeah, he's interested in that market.

Yes. Anyway, my question for you is so. Feels like a PR nightmare for Google. Feels like a couple stressful days.

Feel like it's over the weekend.

And that's my question. Do you think that they are. That they were intentional and happy about releasing the news after market close? Because all of the facts come out over the weekend and it was such a hairy deal that more deals needed to be done even after the story came out and it couldn't be super clean. So. So you have to do it late on a Friday. So then Saturday, Sunday, people can talk, deals can get done, and then announcements can go out Monday, and then it resets the whole news cycle. So you kind of skip the crazy news cycle, I think.

I think the world's too online, so I think it helps that it wasn't a weekday. But I mean, to me, this whole thing, from everything that we've kind of triangulated, this could have been avoided by managers taking it upon themselves to basically call each individual person at the company and say, before you have a negative reaction, I need you to be patient. I know you've been patient through this deal with OpenAI. It's not the deal that we wanted, but we are working to make this a positive outcome for anyone. And. And don't cry yet.

Yes, don't.

Just don't cry.

And it's really hard. And it's particularly hard because if you, if you're in the scale situation, a lot of those people had been there for four years, and with Windsurf, most of those people had been there for less than a year, almost certainly. And so the strength of those connections socially outside of work is weaker. So there aren't as many. Oh, yeah, like my boss. Like, we are already on signal with Disappearing messages. So we just talk back channel all the time. It's like, I mean the crazy thing.

The crazy thing, three months ago the argument was the new company has a strong balance sheet.

Yeah.

And they're going to be able to just do a distribution or pivot or like that was effectively the suggestion.

It was intentionally ambiguous.

And at the same time, the employees left behind on the ghost ship were already talking with other companies. Other companies were coming to them, trying to recruit them.

Coining ghost ship because everyone involved in the deal would prefer you to use the term remain company. Remain company. But that's not as sticky as ghost ship.

So I think ghost ship without a ship without a founder.

I don't think this will be the last ghost ship we see. I think it's like a new pattern and we're just going to see this more and more.

No, but so then you have this dynamic where these talented people, specifically I talked to a founder who was talking with a lot of people on the go to market team being like, these are very talented people. They took a company from zero to tens of millions of dollars in ARR and a bunch of enterprise contracts in a very short period of time. I want to scoop them up. And those people were entertaining conversations because it was just totally unclear to them what was going to happen.

So yeah, I mean I still, I think some of our friends hold The FTC like 100% accountable. I'm probably under 100, but maybe over 50. But my key point is that never lose faith if the FTC is giving you trouble. Dylan Field didn't lose faith when he was dealing with a messy situation with the FTC at Figma. And also you should use Figma. Figma.com Think bigger, build faster. FIGMA helps design and development teams build great products together. You can get started for free@figma.com Anyway, there was another side of the stakeholders that was interesting that this, that this guy Publius said in your, in your, in your, in response to your post, which I don't think anyone was talking about at all.

The entire weekend is cool.

And this is something people don't really think about. I think we're in a good spot with these folks now and I think they'll actually probably be in a better spot, much better spot. But let's read through it. So says there are more stakeholders here that have not been discussed. There is a whole network of software resellers that invested millions into building partnerships with Windserf. I was leading the team at one of those partners. We have customers that we have been selling to for years, who we vouched for. Windsurf, too. I am sending out a note to our 300 customers on Monday that we are stopping all partnership activity with Windsurf, effective immediately. I hope he didn't send that because you gotta recall that email. Don't send the email. Cursor has no channel partners, and this will cost our firm millions. Oh, interesting. I didn't realize that Windsurf had a channel strategy and Cursor does not. Very interesting. So. So Windsurf actually has resellers who go and sell it into their partners because they're doing something on. You know, we do software development consultancy. We come in and help your team run more efficiently. We help you pick the right tools. You can imagine, like, McKinsey, I don't know where this guy works, but, like, you can imagine a consulting firm coming in and giving advice on what tools to use. They get a kickback from Windsurf, they don't get a kickback from Cursor. So he's going to lose millions of dollars. Hypothetically. Hopefully not. But my reputation and good name is worth more. With Google taking the best engineers from Windsurf, our large enterprise clients are screwed. When they purchased Windsurf enterprise licenses, they were also purchasing an ecosystem that they assumed would continue to improve. This is disastrous for any startup tech company with goals of building an enterprise channel. It's interesting. I don't think people have any idea how much this just killed channel sales for any AI company. So the rest of the AI companies out there who are like, oh, channel channel sales is really interesting.

Basically, if you're, if you're. If you're a top company in your category growing super quickly, you might turn into a ghost ship.

Yeah, yeah. And so if I'm trying to resell that type of stuff, then I have to go to the CEO.

They're good, but they're not so good to get these maxed out. They're not getting a trade deal.

Yeah, yeah.

You can count on them to not get.

Yeah, yeah, yeah, yeah. You're having, you're having dinner with the. With the CEO of some AI company is talking to you about, like, yeah, we have the best software. And you're like, but you don't have any, like, actually great AI researchers. Right? Like, you mostly just like to partner.

With companies with, with like, medium to.

Yeah, yeah, great product. It works, but not too well because we don't want to lose you.

You want to keep working with them for a long time.

I keep working with you. And so yeah, there were a bunch of other reactions. Shell quires. Let's see. See, I'm not an M and A guy, but something feels really broken in Tech M and A. Confidentiality is out of the window. Deals get done and called off on a whim. Lack of rigor and discipline on both sides. Curious to hear what practitioners think. People are going back and forth on this, but really quickly, let's tell you about Vanta Automate Compliance Manage Risk, Improve Trust Continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or managing a complex program. And so Will Menidis was particularly blackpilled over the week.

Friend of the show, incredibly blackpilled.

But we're so back. Will, you're good. Will Manitis said the reason that tech has been able to grow so quickly and create so much wealth that it is rich, that it ritualized a set of norms around corporate governance that are very distinct from what the law actually requires. The second someone defects, the whole ship goes down. He's so upset, but basically he's talking about like the handshake deal protocol is not a legally binding contract. Even term sheets. You get a term sheet from a VC that's not a legally binding. That's not a legally binding contract. Like a VC can write your term sheet, everyone can sign it and then you can just literally says it literally it says this is not binding. And then you can just wake up on the wrong side of the bed and be like actually I don't want to do the deal. And or it could be like you found something in due diligence or you met a different competitor but the game isn't over until the money's in the back. The experienced entrepreneurs know this but we run on this like basically you know these society these like social constructs and these like social contracts that say hey, if you're the type of VC that constantly rugs on term sheets, on signed term sheets like you will, there will be a bad vibe around.

I can think of two that are widely known.

Yes.

And are still in business.

Yes.

But they it becomes very hard to be an effective VC if your brand is. You just pull term sheets.

Yes. Yeah. And some of those, some of those folks have been banned from YC Demo Day. Other people have had like viral Twitter posts about them every once in a while. Like they, they do that to a founder and then the founder becomes very successful at post economic and is like Yeah, I still have a grudge. So I'm going to tell the truth about X, Y or Z. And so the thing is that now where we are now is I don't think the whole ship goes down. I don't think this is a true defection. I don't think this reflects poorly. And I think that the question of like the big question here was will this cause more new startup employees to want to just go straight to Google? Let's say that you have the option of working at Google or big tech broadly where you know the paychecks are coming on time, you know the RSU are liquid and you can just cash out as soon as you get them or whatever their lockup is. And you know your managers are going to be there and you know the company's going to be around.

Somewhat accurately predict what your comp will be in five years if you really work hard.

Exactly. And then, and then on the flip side you have startups which are in some ways lottery tickets but also you get to work on game changing technology with really cool people. You get to move a lot faster. There's a lot less overhead, it's a lot more edifying. You get to work for people.

Potentially the next Sundar.

Yeah.

Instead of Sundar.

And, and the economic upside can be a thousand X. It can be huge. And the question is if that, if that deal was broken, if that, if that deal was okay, yeah, you can go work on the cool thing, but you might wind up penniless and not cash out like you were promised. That changes the equation and that could pull people away from startups. But I don't think that's going to happen based on this because I think the lesson from this is that everyone got taken care of and the social contract still still holds, although it was very messy. And I think going forward every company that's doing a deal like this needs to think how do I get the scale AI outcome and not the windsurf outcome? Because just two days of pain is not worth.

I think one potential solution to these deals is exactly what we just saw happen.

Yes.

Which is big tech company does basically an acqui hire of the team that they want and then you simultaneously sell off the underlying asset immediately and you generate a liquidity event for everybody. So everybody's happy. It doesn't have to be like life changing. Right. But it has to be something. Right. So that it's.

And I mean the very key thing is that they left enough money on the balance sheet because the money on the balance sheet and this was, this was Hari Raghavan's analysis. He clearly, I mean, the guy's, he's a good friend. He's. He's obsessed with cap table math.

I know he was.

Yeah.

This was capital J journalism.

Yes, thank you.

This was investigative journalism.

Yeah. And so, I mean, his post is so long that I need to just go to the tldr. But the TLDR is that he says, I spent hours last night breaking down the cap table and deal dynamics from public info. I realized that the unvested Equity is around 80 to $100 million. Awfully close to the $100 million the founders and board negotiated to leave behind in the bank. Why shouldn't management just.

That was unvest, which is a key detail unveiled. So they, so they're basically saying, we're, we're leaving enough as though you vested your full four years.

Yes. And so, and so if you, if, like, it's, it's, it's still a leap of faith because I don't believe that Google and Cognition were in communication at all. And I think if they were, that would be more like antitrust chaos and stuff. But just economically, you know that if you acquire, if you leave a company there with that much money on the balance sheet and those many and that many, like invested options, somebody is just going to be like, wait, so I can get that for basically free. If. And I can make all the employees whole and I get all the employees and I don't have to pay really like anything to get that thing. It's just going to happen immediately because it's just like you've created, you've left this, you've left this, like, you know, $100 on the floor and someone's going to pick it up. Like, the free market will just work.

I think, I think the question that I had and that I think a lot of people probably had is what is the governance structure of the ghost ship? Right. So Jeff, Jeff, crazy question, Jeff Wang, who's coming on the show at 1 p. Oh, he is. Jeff and Scott will both be coming on at 1pm for 15 minutes.

Okay. Can we communicate that to the other guests?

Yeah, yeah, yeah. And so the question is like, okay, Jeff, you now run a company with hundreds of employees that are angry about what just happened, that probably, like, maybe they want to keep building Windsurf, but maybe they want to go and do something else. And what happens like, Jeff is basically, I would imagine, like has to make a bunch of calls.

Yeah.

It's like, do you just keep Running the business. Is there pressure to keep running the business?

First question, Jeff. Did you ever feel like, is it Frodo Baggins throwing the ring into the fire? It's like, pay out the hundred million dollars in bonuses. After all, why shouldn't I keep it? Just throw the ring in. Throw the ring into the Mount Doom. Yeah.

The question was like, were you ever tempted? Could the new leader. The leadership.

Yeah. Like, was it ever was that.

Well, could the leadership have decided, you know what, we can run this business super lean.

Exactly.

We could actually run this business with like 10 people. And we're going to run it for a few years and we're going to. We're going to use the advantages that we get from our own product to run a super lean team. And then you would have employees, like, effectively burned again.

Potentially.

But that didn't happen. I'm very happy about it.

Yeah.

And the question. Another question is, like, if you're.

If you're big tech and you're going to do one of these, like, zombie acquisitions, you got to leave enough for the unvested shares. And if you're founder and you're going to do one of these zombie acquisitions, the ghost ship's got to be full of plunder. Full of plunder stacked with treasure. Because then.

That's right.

Then the rescue divers will go and lift the ghost ship off the. Yeah.

And if big tech comms are telling you you can't tell anybody that they're getting taken care of, you have to risk it. Tell a little birdie conversations. Tell a little birdie to go fly around.

Yeah.

People know.

You got to be. You got to be having, like, quick calls, not. Not saying anything too specific, but saying what you said. I think you worded it perfectly. And it just. Don't cry yet.

Or varun. Just because. Because he was getting dragged. Because. Because he apparently, he. Apparently, he. Apparently people were messaging me like he joined like the. The all hands for like 10 minutes.

Yeah.

You know, and it just seemed like he was being like ultra, like kind of like a psychopath being.

Totally.

See you guys. Yeah, it was a good run. And it's like, go, go have fun building.

You know, this would not have happened if he had said, no, I need an hour. I need to explain everything. Would that have killed the deal? Almost certainly. If that killed the deal, what's the outcome for the employees? Probably worse than what they got.

You think it would 100% have killed the deal?

Almost certainly. I think one of the key deal points in the negotiation with Google was We do it via Google's comms and legal strategy because Google does not want to get hit with some crazy FTC lawsuit that's going to not just be some multibillion dollar settlement, but also a huge headache for years and bad press and, and stuff like that. So Google says, hey, we want to do this our way. It's going to be awkward, but we're going to take care of everyone. And yes, we will leave enough money on the balance sheet to pay out all the unvested options. But in exchange, you got to do the comms our way, which was rough.

You have to create a tower, you.

Have to be scapegoated by the entire. You get hazed.

Yeah, you're going to be hazed for 48 hours.

You're going to be hazed, but then.

And it's going to blow back on us.

Google didn't look good either that weekend. It was rough. It was rough. Anyway, let me tell you about Linear. LINEAR is a purpose built tool for planning and building products. Meet the system for modern software development, streamline issues, projects and product roadmaps. Start building at Linear app.

And if you are launching a TBPN derivative product, derivative work get on Linear. It's the tool we use.

So what you would do is you'd put all the glasses of tap water that you need to drink and then you would just check those off at a time. Yeah, every time you like. We only have one rule about copying us, it's the derivative works clause and our terms of conditions. And in there, if you create a derivative work of tvpn, you are legally required to drink a glass of tap water on.

Every time you go live.

Yes, every time you go live.

And so yeah, again we don't have.

To make slice fee.

It's about being accessible.

We did invent tv. We invented news. But we, we will open source it. It's an MIT license type of situation. Modified MIT license.

Yep.

The only modifications that you have to drink tap water. Yeah.

Anyway, it is a small price to pay. Anyway, in other news question, what do you think? Why do you think Google did this?

Why?

Yeah, I mean.

Oh yes, yes, yes. That is, that is. The real question is, is. We've talked about the role of the VCs. I think based on what happened. Never talk down on first ballot hall of Famer Neil Mehta Never talk down on the future for first ballot hall of Famer Everyone was saying, oh Neil Matta, Green Oaks they did this sloppy deal screwing over the employees. Employees are great. Neil Matta potentially Goaded. I don't know.

I agree, I agree. No, it's well played. No, no. But the question is, why did Google feel like they needed to do a Meta style trade deal in.

Yes, it's a good question. So windsurf in my. I don't know everything about all the employees. They brought over. They brought over 50 engineers. I don't know if any of those are like insane AI researchers with like crazy citations on like foundational papers. I don't know that it's the same. I don't know that they're trying to build a super intelligence team to really bring their models to the frontier because they already have that. Like, yeah, Gemini is, is at the same, like, level of like, AI research quality as OpenAI anthropic by most benchmarks, by most people that use them. Like, it doesn't feel like they're way behind there. Llama 4, on the other hand, was a little bit behind and so. And it felt like Llama 4 was specifically behind, not on the product side, but on the research side. Like, they just weren't figuring out what works in rl and so they had to bring over a ton of people to just be like, how do we actually train this next model? Like, behemoth couldn't. They couldn't tame the behemoth. Like, they got to get some behemoth tamers in the building. And it seems like that's why they acquired all the different researchers who know, okay, like training on 100K H100 GPUs gives you this and then you need to use this data and you need to hold back this data and you have this, this reward function and this reinforcement learning technique and you need to do all these different things. And so you need AI researchers for that. But Meta has great product people. So like, they're going to be able to take what they build and then vend it into Instagram and vend it into all sorts of different places and use it to optimize the ads 1% and make a trillion dollars. Like, they're going to be fine on that side. Google is great at AI research. DeepMind is amazing. They have so many reasons. They invented the transformer. They're not behind on the quality of the foundation model. They are, they are behind on product and they are, they are behind on like, user experience. And everyone complains, oh, they have this amazing VO3 model, but like, I had to log in seven times to get access to it and then go through an upgrade thing and it was on wrong Google account. And Then I have to trigger it and then I get.

So here's, here's the, here's the, the.

Steel man for the short sellers.

No, here's what's, what's kind of interesting about the way this is developed. So at Google's last developer event, they launched Jules, which is an async development agent for code Gen.

It's a Claude code competitor and OpenAI more.

So it was like a dev. Yeah, dev and AI competitor.

Devin competitor.

And so now you have. You imagine Google has more ambitions around code gen. Right. They've seen anthropics. Very few businesses actually like will work at Google. Right. You need a billion dollars of revenue or you're going to get shut down, most likely. And so codegen is an area that I'm sure they're taking very seriously. I haven't heard a lot about Jules since the event. I haven't heard about people that are, that are using it. I'm sure it's, I'm sure it's powerful.

Yep.

But you now have a dynamic where some, you have cognition. And Devin, who got, I can't say cloned by Google, but again, like they were the first, like serious like enterprise coding agent and so. And the Windsurf gtm, you know, sales marketing team versus the former Windsurf founders and top researchers who are now at Google. And so I'm just incredibly excited to watch this play out. Pavel was posting earlier, he said, I hope these guys smoke google@AI coding. Imagine the chip on your shoulder after this if you're a Windsurf employee. So, yeah, everybody's got to be very fired up.

Okay, really quickly we got to. I want to track some poly markets around Google. Google forced to sell Chrome. Remember we were tracking that a while back. There was a question of whether they need to split it up because the ftc, we've been talking about the FTC a lot. I will pull up this. Google forced to sell Chrome. It is down at 1%.

Now.

No one thinks that they're going to have to sell Chrome. And of course the, the browser war is heating up completely. And I was checking the. They're not tracking this Sundar Pichai thing.

Sundar pitch AI.

Pitch AI. He has been pitching AI very successfully. But the, the question of like, competition is like it's heating up a ton in browsers and then now it's heating up in this and like the FTC should totally be sitting back and just watching the gladiatorial games play out. And yet like, even though, I mean.

Look at look at, look at the vibe coding market right now. Is competition not thriving?

It's insane. It's insane. Yeah, no, I completely agree. And so, like, I think, I think there's a lot of people that are making the argument that like, oh, you can't keep blaming Lina Khan. She's not even there anymore. And it's like she was there for long enough to like, really like lay out a new pattern and just instill a bunch of behavior in the big tech companies that results in these, in these crazy deals. Anyway, let me tell you about numeral numeral hq.com. sales tax and autopilot spend less than 5 minutes per month on sales tax compliance. It's. It's like an AGI for sales tax.

That's right. Should we talk about Zuck's new super cluster?

Oh, yeah, let's go through it.

Okay, so there's some new reporting from the Times as well around their strategy, which we'll cover next. But why don't we start with Zuck Hit threads this morning.

He hit threads hard.

With a thread.

With a thread. It was very cool. He had a great visualization. I hope we can pull this up. I'll put it in the tab. But basically, bangers. There we go.

This Jarvis send link to Bang.

So basically he has this cool visualization of how big the Hyperion data center is and it's just the full size of Manhattan. And somebody was replying to this being like, he's building the cube. Oh, Sam d' Amico from. He's been on the show a bunch from Impulse. He's building the cube because there's this old visualization of like if you put a billion people in one building in Manhattan, it would just be this cube. And so everyone who's like extremely pro building housing is like, let's build a cube. Let's build a cube. Anyway, the news is that Mark Zuckerberg is giving more details. He's announcing that like the super intelligence team is real. It's been only rumored reported he hadn't. He hasn't actually said, hey, we're doing all this. That was all like leaked reporting. He came out and posted on threads and he posted on Facebook too, saying like, this is real. And here are the details. He says he's going to invest hundreds of billions of dollars in capex, which is, which is what they're doing like 60 to 80 a year now. And so I guess, like, it's more just like he's going to continue that for the couple years he's not giving Specific guidance on where the rank 2024.

EBITDA was 84.

What was the CapEx?

85 billion.

Oh, CapEx or EBITDA?

That was EBITDA.

Okay. Yeah, look up the CapEx.

CapEx was between 38 and 37 billion.

Okay. And then some of that goes towards like reels, and then some of that's used for AI. And then the AI stuff kind of slips over into reals and just other, other, other data center uses. But in general, you know, he's going all in. He's continuing to invest billions. What's interesting is that he claims that Meta will be the first lab to finish a 1 gigawatt supercluster. So he's basically racing against Stargate and thinks that he can bring Meta's next data center online before Stargate. And he's calling his 1 gigawatt supercluster Prometheus, which is a hilarious name. The guy loves the Roman Empire, apparently loves Greek mythology too. But Prometheus is a very, very edgy name to pick in my opinion, because Prometheus is the guy is the Titan who stole the fire from the gods. And it's like, I get that. It's like the seed of like the new. It's like the seed of humanity. It's the seed of this new thing. It's a great name in many ways, but also it's like if God's real, like, he might not like that. And that could be a problem.

It could be bad.

But, you know, Zuck seems to like, like Icarusian names. Yeah. Because Orion, he likes flying.

I mean, he objectively likes flying close of the sun.

He does.

That's kind of where he feels at home. He does the warmth of the sun.

The warmth of the sun. And you know, the hot take on Icarus is that, you know, everyone talks about, oh, like, you know, don't fly too close to the sun. You don't want to be Icarus. But you know who, you know the person who was like, telling Icarus to do that.

No.

Surprise, surprise. No one remembers his name. His name was Daedalus. And Daedalus was like, oh, don't fly close to the sun. Who do we remember? We remember Icarus.

That's right.

And so maybe hot take. It's better to be Icarus than Daedalus if you want history to remember your name. Even if you fly too close to the sun, even if you have a great fall, you will be remembered. That's right. So maybe it's a, you know, Icarus gets rich Daedalus sounds smart. Situation. Daedalus sounds smart. Icarus gets rich.

That's right.

I think that might be the what's.

Going on mental model.

Anyway, I mean we. I was also on one of the first shows we ever did. I about. Was talking, talking about how the name of the new VR headset Orion is kind of odd because Orion is like this hunter, but was notoriously like over his skis with the hunting and was killing all the animals and was smited by the gods and you know, and basically like it was like a textbook case of like hubris and overaggression. Like the moral of the story of Orion, who is the hunter in the stars, the reason he's in the stars is because of hubris. And so it's very funny to name your product like a tale of hubris. Maybe it's a reminder. I seriously want to talk to Zuck about this because I'm very interested. How many layers deep does he go? Because I mean the guy's like, clearly studies this stuff. He's like a student of the Roman more than me. I just google it. When a new name comes out and look at the summary. I imagine that he reads it much deeper and I imagine that he gets the joke and is doing this almost ironically, which I think is cool. Anyway, the other news is that they're aiming to scale up Hyperion. So Prometheus will come online in 2026. So that's probably one year away, maybe a little bit more than one year if it's toward the end of 2026. And then they're building Hyperion, which will be able to scale up to 5 gigawatts over several years. And that feels like simultaneously like crazy insane ambitious. Amazing. I love it. I'm super excited for the capability coming out of that. And also like the biggest glass of cold water poured on the AI 2027 folks like of all time. Because like I feel like everyone in the AI like is going to be self reinforcing and just ripping is like, well, yeah, like the data center capacity will 10x every year for the next 20 years. And it's like. And Zuck's like, we're gonna 5x over the next three years. I don't know how many several years is I imagine like three or four something. But it's just funny to be like, like it's, it's incredibly ambitious. It's the biggest data center ever. I believe it's like this massive project. It's so cool. It's amazing. And it still doesn't match the rhetoric of, like, the really crazy AI people at all totally. But, you know, it's fun. Anyway, that's the story of Zucks. Any other takeaways from the Zuck supercluster?

Congrats.

We should ring the gong. We should ring the gong for Zuck.

Let's hit the gong. So there was some new reporting from Eli Tan over at the New York Times around Meta's new superintelligence lab is discussing major AI strategy changes. And so there is a picture.

What does that mean? Going closed source or.

Yes, I'll read through it. So Meta's newly formed superintelligence lab has the discussed making a series of changes to the company's artificial intelligence strategy. In what would amount to a major shakeup. Last week, a small group of top members of the lab, including Alexander Wang, Meta's new chief AI officer, discussed abandoning the company's most powerful open source AI model, called Behemoth, in favor of developing a closed model to people with knowledge of the matter. Said for years, Meta has chosen to open source its AI models, which means it makes the. I'm going to. I'm not going to read this. Any move towards a closed AI model?

What is software? Can you tell me?

They were about to say, the New York Times wrote, they're about to make the computer code public.

What is a computer? What is code? Can we explain this for our audience?

Yeah, we should really break it down. Anyway, so Eli writes, any move towards a closed AI model would be a philosophical change at Meta as much as a technical one. Meta has won over developers for open sourcing its AI models. And one of its top AI executives, Jan LeCun, has said that the platform that will win will be the open one. So, yeah, I think this is obviously still very much in flux.

Build Mecha Churchill.

Mecha Churchill.

If he builds Mecha Churchill, instant win. Instant memetic win over Elon. They're clearly battling it out. They're trying to fight in MMA wars and they never, they never got that fight down. Now they're fighting it out in AI Build Mecca Churchill. It'll be such a good meme.

Be great, something there.

Whatever you build Zuck, you got to sell it to people. You got to use ADEO customer relationship magic. Adio is the AI native CRM that builds scales and grows your company to the next level. You can get started for free. Tailwinds of the last decade are the are now headwinds, says Buco Capital bloke. He says higher rates AI labor, compression and Available roles, salaries, business model maturity, managing margins vs growth crowded knife fights vs greenfield less liquid, less liquidity because companies stay private longer slash forever worse M and A environment due to regulation PBD technology enabled outsourcing. So little bit bearish from the Buco capital bloke. Shield says feels like a good time for everyone to learn about single trigger versus double trigger acceleration of shares. So single trigger is if there is an acquisition, all of your shares vest. I don't know if this would. The question is like, how do you define trigger? Because these zombie acquisitions, you're on the ghost ship. Who knows if you trigger the trigger. But double trigger means you have to be acquired, change of control and then fired. And so double trigger, it's very hard.

To negotiate that if you're just the 300th or the hundred.

Almost never happens. The people who negotiated are the CFOs who come in. Yeah, CFO comes in. Doesn't matter if CFO is employee number 700 or number three. Like CFO is going to get single trigger. Always. Always. I don't know, maybe not always, but like, like the people who know know. But you should know about it. You should know about single trigger and double trigger. The other interesting thing that I was digging into that we didn't really ever get to the point of like understanding was for a long time people were saying like, hey, you can debate all of the stuff about Windsurf, but the employees on the ghost ship, the ones who are in the remain co, the ones who were left behind, they would have vetoed this deal if they could have. I think if you go to them today and say, hey, you're working at cognition, you're using Ramp now. They're gonna be like, yeah, yeah, I'm happy, I'm happy. And it's not like, it's not even like they landed at a boring company. I was saying like they could land at like, you know, some old line company that just picks them up. Like, well, what was that company that like humane landed at like HP or like they're like doing like smart printers now? Like that's not exactly like a, like a, like a upgrade in terms of status in Silicon Valley.

That's not exactly a airborne moment.

Exactly.

You got to put your, you got to put these on, experience the audio.

Audio. But, but, but going to Cognition seems like a great upgrade in terms of just like, like, you know, being at another cool company. You're still on the startup roller coaster. You're on the ride. You're know, you, you have a Lot of unbounded upside.

I mean, I don't want to take too strong of a position here, but Scott and the team at Cognition are some of the best. And if you just lost all your engineering talent and you're confident in your own abilities and you want to work with world class engineering talent, really great, really great place to end up.

Yep. Really quickly. Let me tell you about Fin AI, the number one AI agent for customer service. Number one in performance benchmarks, number one in competitive bake offs, number one ranking on A2 on, on G2, go to Fin AI. But the, but the question that I was running through and I was getting to was single trigger, double trigger, on like that, that does matter. That is important in acquisitions. But another thing that potentially could be relevant in a future acquisition if we get sort of like, you know, the windsurf situation, one of these zombie aqua hires with the bad ending where even years later you're like, I would have loved to have vetoed that. Like, I really didn't. I never, I wasn't happy with the ending at all. And I haven't changed my position for years. Not just. It wasn't just a few days of chaos. It was like forever. I was like, I wish that hadn't happened. I picked the wrong company. The question is, if you exercise your stock options, would you have a different outcome? And so essentially, when you get a job offer from a startup, they give you a stock option contract that allows you to exercise and pay to buy stock, actual stock in the company for a lower price. And the question is, your legal rights as a option holder are different than as a stockholder. So an unexercised stock option makes you technically a contract creditor, not a stockholder. And so you do not have the bundle of statutory and fiduciary law rights that attach to stockholders. So if there was going to be like a class action for the Ghost Ship employees and they really did want to block it, or they wanted to sue Google or they wanted to sue the founders or something like that, they would say, this is a breach of fiduciary duty and I'm a shareholder. You had a duty to me. But they didn't in this case because if they had, if they had an exercise, I'm sure some of those employees had early exercise. It's a thing. But it's just an interesting piece of the puzzle that we fortunately didn't even have to go down this road. But I do think that people might want to re have this conversation because I mean, the chatgpt summary says to preserve more robust remedies that stock law makes available. Inspection rights, fiduciary duty claims, derivative suits, appraisal, exercise first, complain later. Exercise first, complain later. And so you can do a partial exercise, but if you don't own a single share of the company, then you don't have the right to actually sue the. Or you might have less of a claim to sue in a scenario like this. So it's a potentially rough scenario. Should we go through Elon? Let's switch gears.

Elon is leading a massive $2 billion investment from SpaceX into his other company, Xai.

Yes, we've talked about this. Nothing better than one hand washing the other.

That's right.

This is what we like to see.

Yep.

It's so efficient.

It really is.

How else do you wash your hands if not with one and the other?

Yeah, try washing.

It's impossible.

Don't.

Yeah, so try to wash. Elon is taking 2 billion from the SpaceX balance sheet, which is what, at 400 billion now? And they're raising billions pretty regularly and doing tenders and stuff. So SpaceX is, I think, the largest. They're kind of neck and neck with OpenAI in terms of private market companies. It's a 20 year old company, very stable business. Starlink's print, throwing off cash. There's like so much that's. That is mature about that business. Obviously still got to get Starship to work, still got to get to Mars, still got to go to the moon regularly. But, you know, exciting road ahead, but still very solid business. And he's taking 2 billion off the balance sheet of SpaceX and putting it into XAI. And so the map of Elon companies is getting crazy. Yeah, it's Koretsu. It's a Koretsu. Everything owns everything else.

So the question that, when I saw this news come out.

Yeah, What'd you say?

My first reaction was that Mecca is going to love Mars. He's going to love it up there. It's going to be the first, potentially the first AI to get to Mars. My second reaction was I actually think that if you add up the valuations of every Elon company, Tesla taking it number seven on the mag 7.900ish billion dollar company SpaceX, 400ish billion dollar company Xai, which is maybe getting valued at 200.

It's a 200, right?

Yeah, something around 200 now and then Neuralink, which is what, at like 12? Somewhere between like seven. Roughly around like double.

Almost a. Almost a $10 billion company. Might as well be.

So somewhere in that range.

Yes.

It's actually interesting to think of, of if you rolled all of these assets up together.

Yeah.

What do you think it would trade at a premium to all of their standalone valuations?

I don't know enough about. In the public markets or private markets.

Yeah, in the public markets.

I don't know. I don't know enough about, like, because I've heard a lot of, a lot of public markets investors say stuff like the market just really wants a pure play in this.

More pure play than Elon Inc. Just Elon Inc. Because a lot of these companies still, even at various points, end up being founder bets. Like Tesla is still a founder bet. Will Elon figure out autonomous vehicles?

And I do think that there's this.

Underrated narrative of like an Xai too. At $200 billion, it's like, I don't think investors are underwriting the new AI.

Companion today, the profit stream or even revenue stream from the API business.

Yeah.

No way.

Yeah. They announced a $200 million government contract today. They were under writing Elon being able to get deals like that done. Right. And even. Remember Ben Thompson, we asked him about this and he basically said, I can't do analysis on Elon companies because they are detached from traditional business reality. And so my question is, like, if you roll all these things up, Is this a $3 trillion company?

Yeah, I know. I mean, the interesting thing is, like, it does give you, as a retail investor, like, you can pull so many different narratives. So like right now, just within, just, just within Tesla stock, if you're a, if you're a Tesla investor, you can look at what's happening with Tesla deliveries because of the political backlash or what's happening with the, with the tariff war, the EV tax credit. And you can be like, that's bearish. But then you can simultaneously tell yourself a story about, well, Robotaxi is going to be so big that it'll make up for everything. And then if that doesn't sit with you, you can be like, Optimus is going to be so big that, that it justifies a multitrillion dollar valuation. So I'm still a buyer. Right. Because you have these, these call options.

And really it's all about the rare earths on Mars.

Yeah. So imagine, imagine that. So right now with Tesla, I see like four stories and they're kind of like too bearish, too bullish, potentially. Like, the bearish ones are, are EV tax credit and, you know, not being able to sell as many vehicles. The Bullish ones are like, maybe he figures out Robotaxi, maybe he figures out Optimus, right? And that's why you might choose to.

Buy or sell Tesla or Neuralink chip in your brain with the xpi.

When you put all them together, it's like on any given day you can. It's so much easier not to be a panic, right? You don't need to be panicking because you're like, well yeah, like Neuralink got pushed back by the FDA today, but SpaceX landed Starship. So like I'm buying or like oh, the starship blew up but you know, Neuralink, you know, made a monkey play counter strike. So like I'm going long, right? There's like always some good news to latch onto when you see the, the Elon portfolio as a whole. But there's this fascinating chart here. I don't know if we can pull this up, but it's the links between all the different Elon companies. So, so SpaceX buys, SpaceX buys 5 million annually in vehicle parts, energy systems and in the Model X Astro van. So that's money flowing from SpaceX to Tesla. Stainless steel alloy technology, Starship materials for Cybertruck go from SpaceX to Tesla and then the Boring Company purchased $300,000 annually in energy systems and uses the Model 3 for the LAS Vegas loop. And so they're buying from Tesla. Starlink has a T mobile partnership connectivity for vehicle Internet and superchargers in Teslas. And then you have all these other things where like Neuralink might link into Xai. XAI did the all stock acquisition of X Corp. And then there's a study for robotic arm control and future mind controlled limbs that links Neuralink to Optimus. SolarCity was acquired by Tesla. FSD is a separate product. So like all these companies are, are overlapping and the interesting question is like, is like should, should they just be that Karatsu like and what is stopping that? Should they all just be under one Musk Holdings, Musk Inc. And if it was, I think it's that, that would actually be extremely uncomfortable in the public markets because there would be all these crazy shareholder lawsuits where so like you know, if you were, if you were, if you were a shareholder in, in like Musk Inc. And Musk Inc. Is public and owns neuralink.

You can, there's also, there's beautiful, there's also this beautiful cottage industry of people that are effectively resellers of Elon shares.

Yes.

That own a lot of these companies and wouldn't necessarily want this to happen because that kind of stops the chain gravy train. I had somebody message me yesterday and said it's funny how investing in Elon Cos is not too dissimilar from buying watches from ads because if you have a great relationship with your ad in this case which is Elon, you can get direct access to the companies but if you don't, you're on the secondary market just buying SPV.

You're going to get bezel.com but don't worry because your Bezel concierge is available now to source you any watch on the planet. Seriously, any watch.

Great transition.

Stop worrying about your ads. Head over to getbezel.com anyway. If you're looking to buy Tesla shares the one Elon Musk company in the public markets, head over to public.com investing for those who take it seriously. They got multi asset investing industry leading yields. They're trusted by millions. Anyway, in other news, is there anything else we should talk about with Elon ripping 2 billion into Xai? I think we're good there. Let's move on to Casey Neistat. Some personnel news baby. He's joining Mod Retro as chief storyteller. That's not a real title and it's not a real job but I am excited to be a part of the team. I love the candor. The Candor. I don't know, I've mispronounced that but he put out an 11 minute video talking all breaking down the Mod Retro thesis, all about how planned obsolescence is destroying great consumer electronics devices. You buy something and it's in the trash in a couple years, it's collecting dust. And he goes through some of the products that he's had when he was young that still work as intended. They didn't need some crazy software Update. Wasn't an IoT toaster. It was, you know, the things that he loves, the Walkman, the original Game Boy. And he breaks down how Palmer's thinking about building and the Mod Retro team is thinking about building a you quote unquote like Game Boy that's different from the other companies that are also creating similar like emulator products. He's like I got this $50 one and it works like you can actually play it now you can't play it at a high level. Like the frame rate is not like the quality of the product is not at a level where it's like truly competitive. But also it's, it's kind of designed just like as like you know, temu level junk basically. And it's just like, how can we just like slop it up for a little bit? Whereas Paul, Mr. Lyle artist.

Just getting to the quality level of the original Game Boy color. I have a Game Boy color from my childhood.

Oh yeah.

And a while back I just put in. It had been two decades. I just put in new batteries and just turned it on and just immediately booted up and worked.

And that just doesn't happen anymore with.

Modern electronics because after Tyler updated to glass, we made.

How is it? Give us a thumbs up. We're missing a microphone for Tyler today, so give us the thumbs up. It's still good.

We got a thumbs up.

We got a thumbs up. Okay. Well, he will be back on the stream shortly. We will, we will buy another camera. We will buy another camera and we will also buy another mattress from Eight Sleep Mattress Cover. Go to Eight Sleep.com get a Pod 5 Ultra. They have a 5 year warranty, 30 night risk free trial free returns, free shipping. Shane Copeland over at Polymarket says one day there may be more AIs than humans. Creating, pricing, trading millions of polymarkets on everything. A unique benefit of on chain prediction markets is the ability for AI agents to plug in multiple groups, quietly print on Polymarket with agentic workflows. As AI ushers in super intelligence, Polymarket ushers in collective intelligence. It's happening. I'm using the best AI models. Oh, interesting. Didi is doing something over there. I asked it to use modern portfolio theory. Bet sizing may calculate that. Yeah, very interesting to see the, the interaction of AI and, and, and like Web3 or like you know, just crypto, generally crypto powered protocols because it's like.

This is one that actually makes sense. But there was, there was a, there was an era of, of basically meme coins that were just like, I'm not a normal meme coin, I'm an AI meme coin. But this is actually bringing together and them bringing them together in an interesting way. Josh Kushner, friend of the show, had an absolute banger. He said the goal of life is to be excited to go to work and excited to go home. I love it and I couldn't agree more.

That's fantastic. Put that on a billboard, Josh. Go to adquick.com out of home advertising. Make easy and measurable. Say goodbye to the headaches of out of home advertising. Only Adquick combines technology out of home expertise and data to enable efficient, seamless ad buying across the globe. Josh, I want to see that. You know, every VC fund has some, you know, occasionally trite Catchphrase saying on their homepage. This is actually a great message. You should take out a billboard in New York City, Thrive Capital or Thrive holdings or Thrive.

Our goal for life is to be excited to go to work and excited to go home.

It's beautiful. It's beautiful.

This is, this is.

I think it should be on the billboard.

Yeah. This is pronatal.

We just made a transition.

This is sustainable pronatalism in tech. It is, it is being excited to go to work and also excited to go home, be with family.

It is. Should we do Marque Andreessen talking about O3 Pro. This is fun.

So one of our buddies got one shotted by this.

Yes.

So Matt Schumer said, ask O3 Pro based on everything you know about me, reason and predict what the next 50 years of my life will look like. And Mark said, do not under any circumstances do this.

I would have told Mark is like, you will take your fund public, you will run a massive firm, but it will be a ton of headaches.

It'll be like, you know that little knee pain you have, well, you're going to have to probably get surgery eventually. And you know, I think the issue is like ChatGPT knows an incredible amount about, about the things that you're working through in life, depending on how you use it, but the things that you're working through in life.

And, and I still think it knows, it does know a lot, certainly more than many other AI or many other software systems that I use. Like it probably even knows more about me than Gmail, but it's still only getting a slice of the picture. And I think that I need to do this because I imagine it will be like, you will, you will continue to succeed as a train conductor. Because months ago I was trying to, you know, prompt engineer it into giving me stronger, like more in depth data about trains. And so I was like, I am an expert in trains, I own many trains. Tell me all the different brands of trains and the prices don't dumb it down for me. I am a train expert. And then, and then I noticed that once they rolled out Chat GPT memory, it kept being like, you know, someone who knows everything. Yeah, train expert. I was like, no, I was just prompt engineering. I'm sorry, you do this, you do this.

I'll be like, yeah. And yes, John, you'll finally get that height right. Reduction surgery. I know you're always complaining about people saying, oh, I didn't know you were so tall. I didn't know you were so tall.

The Funny thing about that is that, like, that's, that's just something that I would never go to ChatGPT with, right? Like, discussion over whether or not being 68 is like, good or bad or like, I, I, I, I just, it wouldn't have that conversation with ChatGPT. So I, I don't know. I, I should run this prompt. I think everyone should run this prompt. It's kind of, it's kind of interesting just to see, like, the state of AI, right, and the state of these models. Like, how clear is it? But I actually quote, tweeted this and said the line from Terminator, no fate, but what we make, which I know you haven't seen. God, Terminator 2. You got to see this movie. James Cameron, come on. It's one of the best ever.

Someday.

Someday. But in that the whole theme of Terminator is the idea that they are able, that they invent killer AI, Then they invent time travel, they go back in time, they're trying to prevent the AI from taking over in the second run through. And there's this debate about, and the refrain of the main character is, no fate but what we make. No fate but what we make. We have power over our future. And I believe that, I believe that to be true. And so I think no matter what 03 tells you, go do what you want. Carve your own path. Make your own kind of music.

So should we pull up this video of.

Yes. Impulsive Podcast. I absolutely love this. This is from Kyle Harrison.

I'll be right back. But we should play this.

Yeah, I will try and pull it up. So Kyle Harrison is. I'm putting this in the tab. Let's see, let's see. So Kyle Harrison is clipping the Impulsive Podcast with Palmer Luckey, talking about Kyle says. A strong aficionado of conspiracy theories myself, Palmer Luckey has given a perfect, direct answer to anyone who believes the moon landings were faked. And I thought this is very, very fun. Because as, as it's become more, more open to just talk about conspiracy theories generally, there's been a weird meme where people have gone from like, I believe in one conspiracy theory to I believe in all the conspiracy theories. And if you believe in all of them, you're not really thinking critically, you're not actually thinking independently. And independent thinking is what's the most important. So how are we doing on the video? Can we pull that video up? I would love to watch this and react to Palmer Luckey on the Impulsive Podcast full screen. If we can be great let's see. It is loading. But he gives a direct answer to. It's a very funny reaction. Working on it. There's a quote, tweet of Mike saying topic at lunch today. What is one? What is the one conspiracy theory that you have the most conviction in being true? Moon landing being fake? Was the consensus of the group. Mike, who you hanging out with? Canvas outdoors. You got some wild men in there. It's crazy. I like the conspiracy theory that. That John Wilkes Booth was a British agent and killed Lincoln. This is a very interesting one. Have you guys heard this? No. You've never heard this conspiracy theory? Yeah, there's a conspiracy theory, like, everyone knows about the JFK conspiracy. You got to go deep to get into Lincoln assassination conspiracy theories. But there's all. There's very little information about this. And I don't know if we'll ever get to the bottom of the.

There was no Reddit.

There was no 4chan at the time. But. But if you look at, like, some of the incentives at play at the time, like, it's completely possible that, like, someone was pushing John Wilkes Booth to do this, and potentially it wasn't this reaction to the war or something. I don't know. It's just an interesting theory that, like, I don't think many people have gone down. Anyway. Do we have a chance to pull this up? Yes. No, it's not working. Okay, we'll skip it. We'll skip it. Let's move.

Anyways, just. Just to give people kind of the takeaway, the Palmer was basically, Logan said, I don't believe the moon landing was real. Palmer says, the reason that I think it's real is the astronauts left these effectively reflectors on the surface of the moon. And you can build your own laser system to beam them up, wait for the light to travel and then come back to you.

And it will not bounce. Like, the laser won't bounce back if you just shoot it at any part of the random moon. You have to hit the mirror. And they put a mirror up there so that you could do this. The real question, the real conspiracy theories, the real, real conspiracy theorists ask, how deep does this go? Who put the mirror there? Maybe it was aliens. Yeah, because it could have been aliens. And.

Or it could have been a rover, you know, or rover. Like. Like a non human. Like a non human landing.

Oh, rover. I've never heard this term. Alien rover.

No, like the Mars rover.

Oh, oh, okay. So humans never went, but NASA still put the mirror there.

Yeah.

Oh, that's funny. That's no, but.

But the real. But the real.

We can't get humans.

The real.

But we can get a mirror there. So let's do that and let's. So it's like the moon landing was like 90% or let's just land a.

Mirror on the moon.

That's hilarious.

No, no, but the.

That's amazing.

But the. But. But the real. The real theory that Palmer very clearly did not deny is that the moon. He didn't say the moon wasn't made of cheese.

Geez, I knew you were gonna go to cheese. Yeah. Well, hopefully we'll have Palmer on the show soon, be able to press him on the moon landing further, get to the bottom of it, see how deep the rabbit hole really goes. But we have our first guest coming in the stream. Welcome to the stream. Jeff, how are you doing?

Hey, guys. Good to be back. Thanks for having me.

It's great to have you.

For some reason, I'm missing it. Let me see. I think I lost it.

John's not getting audio.

Are we going?

Are we going? I'm not getting anything right now. No. Can you.

Georgia, you got me? I got you loud and clear. How are you doing? You had a busy. Busy. We haven't put up a traded graphic for you and the team yet, and I'm almost surprised, but I really should publish an address.

I think they keep sending to the wrong spot, so that's okay.

Yeah, yeah. What's going on? Give us the update. Since the last time you've been on.

The show, I was checking it, and three months ago we chatted, and actually the question of the day was like, what's going on with Zuck? What's going on with llama 4? Where is this going? And I don't know. Not sure if Zuck is a follower of the show, but I did go back and I said I wouldn't bet against Zuck at $100 billion of free cash flow per year. And I think that's panning out to be pretty true.

Yeah, that was a great prediction. How did you react? Did you catch the news this morning? That first hit Threads and then it was being shared over on X as well around the new data centers.

Listen, I mean, it's hard to ignore a giant blue box on top of Manhattan. It's pretty freaking epic. The technology pros love to see it.

So it's a good graphic, right?

It is, yeah.

Very relatable. Yeah. I thought it was great visualization. I guess the. The. The other question is. Yeah, just like, what was your reaction to the windsurf news. We were chopping up on it. Did you go through the emotional roller coaster? Did you never lose faith?

Was super excited to see what's. Yeah, I'm super excited to see what Scott has to share here in a few minutes. Yeah, I mean, you know, obviously all kinds of weird distortions in the market from kind of former FTC policy. Obviously. Hope that gets cleared up quickly. I think it's a big get for cognition. I think it makes a ton of sense. They kind of needed a horse in that race. And while yes, the gentic coding is part of the long term play here, you can't go straight to streaming. You kind of want to start with the DVDs. And so with cognition picking up windsurfing, I think they're kind of buying that DVD business and going to allow them really leapfrog to the agent stuff in the future. So makes a ton of sense. Excited to see what they build.

Okay, the like where we left it as like the potential risk of these like sort of zombie ghost ship aqua hire situations is that if a deal happens and we get kind of the bad ending where there's 80% of the workforce, 90% of the workforce, couple hundred startup employees who've been working hard for maybe a year and they don't get a good outcome and they're not happy about could have a chilling effect on startup hiring. And those folks could wind up just saying, hey, I'd rather just a stable bet in big tech as opposed to a lottery ticket that even if the numbers hit, I might not be able to cash it in. And so my question for you, you have to start, you compete with hiring.

Yeah. The question I have is do, do some of these later stage or even, you know, you know, big tech companies themselves start using that as a recruiting tactic to be like, look like, you know, it's very possible that this company gets acquired in the next year. And it's unclear, you know, I mean they already do.

Like these RSU's are liquid. Like they'll just show up in your account. You can just cash them in. But. But what is your take? Do you think it makes harder for you to pull people over from big tech into the startup world?

I think the true believers in startups are oftentimes, yes, they're buying a lottery ticket. They sort of understand that. But ultimately the analogy I think of is ships in harbor are safe, but that's not what ships are made for. And I think the best early stage founders and hires just can't they have to live a certain way. They kind of have to hit the high seas and see what they're made of. That being said, I think obviously as startups start to scale and they're hiring the 20th person, the 50th person, the 100th person, the 400th person, that math really starts to make a lot of a difference. And yeah, absolutely. I totally believe that the recruiters at the big tech companies are already weaponizing this to bring people in the door.

Yeah. What about the. I want to know more about how you're seeing the hyperscalers pick different parts of the stack to plug in as like, as they build out, like the AWS dashboard. Because it feels like with like, I wouldn't have necessarily expected like IDEs to be one of the places where potentially like AWS needs a play. They have Microsoft Azure has Copilot, now Google. What was the one that they launched that was like, Jules, Jules. But now they also have the Windsurf team or some of them, and like, they might launch something that's. That looks like, you know, like a piece of software that's not at the API level, above the API level. What. What other categories of like, AI tooling wrappers, that type of stuff do you think big tech is looking at? Or what do you think they will, like, build internally? Like, I feel like you have some experience here.

Exactly where the value capture ends up landing is a big open question. Everybody's nervous. You know, the SaaS companies in the middle are nervous. I think also the model labs are also pretty nervous about this. You know, nobody at this point wants to like, have any two principled of a stance. They'd rather have sort of an egg in every basket, and however it hatches, they'll come out on top. I think that obviously, yes, coding has emerged as one of the largest use cases of AI today. And it makes a lot of sense. It's a highly deterministic, highly controlled environment. Sort of makes sense that people who are using these tools at the edge also are the ones applying them to more software. And so it's a great use case for AI. In retrospect, it was sort of obvious, I think, that like, it was going to be a really big deal, you know, where the value capture really shakes out over the long term is anyone to say. But I think like, you know, you can in some sense, you'll never regret owning the top of the funnel and having a direct relationship with an end user and being in the default in a given category. Like, that's Lindy that goes back, you know, a millennia. Right. You want to be the default, you want to control the direct relationship with the end buyer, and you want to own the top of the funnel. And so in that way, you may be pretty bullish about your models. But does that mean you shouldn't have an IDE play? Probably not. You probably still want an IDE play because you're not going to bet the farm on the models winning the value capture at the end of the day.

Yeah. How bright is the line between AI engineer and AI researcher right now? Because it feels like we've. The original narrative around a lot of the trade deals was like, AI researchers can get a $500 million training run unstuck and that can immediately unlock billions of dollars of value, save you tons of money on inference. Like, the math just works out. So. So clearly I feel like we've always had this divide between. I mean, to some degree we've had a divide between DevOps and product engineering, but the bigger, brighter lines have always been between engineering and sales or engineering and operations. But is there a new dividing line between who's an AI researcher versus who's an AI engineer? And are companies gaming that yet? I always see the member of technical staff and I don't really know what that means. I assume if you work at OpenAI, you're smart, can do a lot of stuff, but how niche is the skill set and how much of a dividing line is that?

I think the skill set's still quite different. If you're working at a large lab on a research team, you really are concerned about primarily training. Whether pre, mid or post training, you're primarily looking at loss curves, you're primarily thinking about what benchmarks you're going to be exceeding at. You're thinking about what data sources you can capture to continue to train and fine tune on. You're experimenting with new approaches to reinforcement, learning or other to try to eke out incremental gains and beat out your competitors. If you're an AI engineer, you're really thinking about how do I apply GPT4 to legal? And it's really kind of a different shape of both skills and also foci and obsession. So that's not to say that in the limit these things won't converge or won't merge to some degree, but I think today there's just a pretty bright line between the two.

Yeah. What about the. I guess the question is like, DeepMind seems very different from the llama organization in the sense that the Gemini models are truly frontier but the problem with Google's AI strategy feels like it's on the product side, where not that many people are downloading the Gemini app, or a lot of people are. It's like millions and billions of hundreds of millions of people. But, like, you know, market share on. On chat apps and, like, who has it in their home row. Like, ChatGPT is so successful, kids just call it chat. And Gemini is not really at that level yet. And so, Jimmy. Yeah, the question is, like, is like the problem that Google's trying to solve feels like a different problem from the problem that Facebook is trying to solve. And the question is, can you acqui hire and acquire talent that actually solves that problem, or is there some sort of underlying structural problem of how big Google is that is actually slowing down product development? My fear is that you get some great product engineers in, but then you're still hamstrung by Google scale. Google's HR department, their legal department, how they think about, oh, well, like, we can't launch this unless it's internationalized on day one because we're Google. And so all of a sudden you have these great people who've been able to, like, go zero to one on a product, get it really viral, and, like, create a product that, like, people love, but they can't do that within the confines of Google. So what do you think about that?

I mean, the potentially more specific question is, are you betting on Google + Windsurf or Cognition + Windserve if they're going after the same developer sort of engagement, Right. If the products effectively converge.

Conway's law, the idea that you ship your org chart remains quite stable and true, and you can kind of take that even one step back, which is like, you ship your culture as an organization. Obviously the org chart is one manifestation of your culture. I think that if Google leadership does not give this new Windsurf or half Windsurf team a really long leash and a ton of autonomy, there's gonna be some level of a immuno response and they're gonna throw them back up again. I think you really have to give those folks a long leash. And that has to come through the top. Has to come from the top, right? It has to be either founder or C suite level where they're getting the green flag to really run hard and ignore stuff like internationalization. And, you know, maybe you still have to get the legal sign off, but, like, you get a dedicated team of legal people, you know, at your beck and call to unblock you whenever you need to yeah, yeah, I won't. I want too early to, you know, bet on the horses here, so I'm not sure I can bet on the horses yet.

I mean even, even like the, like the ADA rules, like accessibility rules, like when you're a startup you move pretty quickly and you're still, you still have to do a lot of the accessibility work. But if you're like 90% there, it's like if you mess up and the screen reader breaks on one page, you're like looking at like a $5,000 settlement to just say, hey, sorry, like we messed up, like one person couldn't use this and like we feel bad and we paid you and we're good. But if you're Google, it's like there's going to be an army of lawyers who are like, how can we, how can we squeeze them on this, get them to settle. Like there's just more economic power and that, and that just naturally slows things down, down. Yeah, it's very interesting.

How are you, what are you expecting out of open source? Broadly open source models. Broadly. In the next couple months we saw OpenAI delayed their launch, just basically saying like the takeaway was effectively not quite ready. Need a bit more time, want to ship something that we're super proud of. At the same time today there was some reporting and unclear if it's true yet, but reporting that says maybe LLAMA will shift away from open source in general. So I'm curious, generally excited for OpenAI's open source model, but what are you thinking?

Yeah, it's a great question. I mean China obviously continues to bring the heat on open source, regardless what we think of the CCCP and their motivations. Just yesterday over the weekend the Kiwi model came out. It's already doing incredibly well on benchmarks. A bunch of developers and founders that I know are pretty excited about that model. And so I think that at the end of the day, if there is some belief that if you have a very good model that it could sort of run away and eat an entire market, ipso facto you're fighting against everybody else who doesn't want you to be able to do that. And so I do think that to some degree the margins will be competitive way. I think that more things will be open source in the future. And I think that also if you think about the workloads, we don't use supercomputers to write our email. We just need basic at home stuff. It works just fine. And the same models, same idea can be extended here. While the Frontier models are going to be developing and figuring out new science and hopefully letting us don't die. We'll see. I think that, you know, for a lot of, like, business use cases and consumer applications, honestly, the intelligence bar is just, like, not that high. And the capability overhang that's already in the models today continues to be quite large as well. I think there's a lot of, you know, people joke there's $100 million inside your laptop. You should figure out how to unlock it while it's $1 billion inside these model weights at least. And, you know, we've got to go unlock it.

So. Yeah. What's your, like, temperature check on just like, overall progress, stagnation? Like, are we. Are you feeling the acceleration? Are you feeling a deceleration? DWAR Cash last week updated his timelines, kind of pushed them out. Still very aggressive timelines. But what was your reaction to that? It feels like I'm seeing a few memes about, like, no one knows how to scale RL or like, you know, yes, we're seeing impressive stuff on Arcade GI from Grok4, but keep in mind, we went from like 7% to 14%. And this is a test that kids can do. And so it feels like everyone's saying, like, we need new research, we need new breakthroughs, we need new paradigms. What's your overall temp check on timelines? And just like, yeah, I mean, listen.

I think that, like, AI continues to be really spiky and it's really hard to intuit where it's strong and where it's weak. And it's, you know, we want it to be strong in all the ways that human intelligence is strong. We also want it to be strong in all the ways that human intelligence is weak. You know, sort of project all of our hopes and fears and dreams, like, onto these things. And, you know, I think in many cases they're still not there. Just today, Chroma released a technical report on the topic of context rot. And so the question is basically like, okay, this model's got a million tokens, but does it? Like, what can you actually do with those million tokens? Needle and a haystack is certainly one benchmark, but we demonstrate across a sweep of other approaches that actually these models, like, they can start to fall apart really early, like way earlier than you might think. You might see reasoning performance drop off even around, like, 10,000 tokens or earlier. And 10,000 tokens is not that much. And so what I want as a builder is I don't want A model that has the 10 million token context window, but kind of maybe sort of work sometimes. I want a model that has 60,000 tokens, but is perfect at paying attention to that model, to that context, and perfect at reasoning over that context. And so we hope this report will both help developers understand what's possible to build today, but we'll also create, like a new set of benchmarks to the model Labs will care about, because I think this is ultimately what people who are building care about.

Okay, so, I mean, that seems like victory left for you. Give me the update on Chroma Rag. And, you know, because it's. Yeah, just to kind of set the base case. Like, there was a lot of, like, oh, like, oh, rags going to get one shot of by just huge context windows. I saw the million tokens and it was like, that's a lot of stuff. I feel like I could just stuff everything in there. And then I was talking, talking to Dwarkesh about can we go even larger? Can we get to a billion token context window? And. And he was making some of those points, but he was also just saying, like, the, like something about, yeah, the inefficiency, like it scales quadratically. And so just getting to a billion tokens is not a thousand x harder. It's like a billion x harder or something like that. Yeah. But give me the update on, like Rag as a research effort, what Chrome is doing in terms of product and then some of the actual, like, use cases when this works, where this breaks down, like how it fits into the overall puzzle. Yeah.

The big question has been, is long context all you need or not? And, you know, maybe it will be someday. So I've not. I don't have a time machine. I've not been to the future. But I don't care about someday. I care about now. And I care about the community developers that we serve and giving them good information so they can really reason about, what is this stuff? What is it useful for? How can I mix it together to create value? That's what we really care about. And so, I mean, the research was not really commercially motivated.

Right.

I think no good research is commercially motivated. You have to go in with just like a question and a thesis and you'll just see what happens. You can't go into it with a bias. And I think really the bias from us came from intuition, from us building stuff. We noticed that actually once models start crossing these certain thresholds, stuff really starts falling apart. And this is what you talk to anybody building in the space, they'll tell you this firsthand. This is their intuition as well. Yeah, like million tokens maybe, but like it really starts to fall apart after a certain number of tokens. And so we wanted to like really just try to quantify that and reason about that and put numbers around that. And so you start to think about like, what are the solutions there? Well, one solution, solution certainly is to use retrieval. Obviously, you know, we are biased in that way. We think Chrome is a great tool for that. We're also seeing increasingly people use re ranking in the loop. We're also increasingly seeing people take like a large context window, say like 10,000 or 100,000 tokens and breaking it into many small LLM calls because the fewer the context, the better the model can reason. And so that sort of process of doing sort of a mapreduce of like splitting it into many pieces and then having these like huge army of parallel, small, fast, cheap LLMs like process that context is also, I think emergingly a best pattern, a best practice. But it's still early. I think that we didn't really want to go into the technical report either with here are all the answers. We don't know all the answers. We want to motivate. Here's the problem as we see it.

Yeah. Speaking of other research and technical reports, what was your interpretation of the meter report that suggested that code tools like Cursor and Windsurf were actually potentially slowering down a system set of developers? It wasn't a huge study, but it seemed like a shocking result in the sense that it was. People were expecting a 20% increase and they got us 20% decrease roughly. And one of the developers who was actually in the study chimed in as well and was like, I think I know what's going on. I'm like leaning on this tool too much. I'm doing a lot of waiting for stuff. And so maybe some of this just naturally is cured by speed. But would love your take on, on that report if you read it.

I mean, we as society are going to figure this out the hard way, right? We're going to figure out like, how much of our sort of reasoning should we externalize to AI and how much of that reasoning that we then have to integrate to process what it thinks. Right. And there's also news over the weekend about, you know, OpenAI making you dumber, making you dumber kind of headlines like this and of course the exact claims. You always want to dig under the hood and look at the actual study design and their actual claims versus the salacious headline, but I think it just speaks to how early all this stuff is. Obviously there's like a ton of excitement. People want it to change the world. And you know, I think exactly how it gets integrated into our workflows over the medium and long term. Like you know the classic thing where like over the next year it probably won't look like that that much, but over the next 10 years it's going to look totally different. And so yeah, we'll just wait 10 years and see.

But yeah, yeah, talk about just, just condensing down information in the context of, of RAG specifically, but also just the other, the other techniques for this. What I'm interested in is like Andrej Karpathy recently was kind of thinking through what the future of reinforcement learning could look like and he draws on this very interesting story that I think most people will be familiar with, which was that for a long time if you went to any LLM and said how many incidences of the letter R are in the word strawberry? It would get confused and it would give you the wrong number. And this was because of tokenization. Each piece of the word is broken down so it can't really see the letters. And so the solution was that they kind of baked in some sort of like system prompt apparently that said if a user asks you to count letters, split it into individual letters and iterate through it one at a time. And then you just kind of like figured out the cheat code for that particular type of problem. The problem of course then is that like humanity job is unbounded and there's a billion different versions of that that come up all the time in the context of work. And like that's where you have all these little edge case failure modes. And that's probably why, you know, I still have to book my own flights or something. Or you need a human to book your flights because using all that stuff there's just too many edge cases. So his, his solution was like we need to develop a reinforcement learning paradigm that goes and finds those hard won lessons and bakes them down and saves them. And that feels like something like compressing of information because you're not just going to want to stuff all that into the context window again, you're going to run into that problem. So, so just talk to me about your reaction to Carpathy's post and then just kind of how RAG might interface with the next models, whether on the training side or on the inference side.

Yeah, for sure. I mean I think learnings from the system running can be integrated Both into the data systems that underwrite that system can also be integrated into the weights. Now, whether that's through RL or different means is sort of going to be dependent on the use case. One way that we've been thinking about it is we're calling this the inner loop and the outer loop of context engineering. So if your context needs to be tuned to do a good job, context fraud is real. It motivates the need for context engineering. And this is the job that many application developers are doing, have been doing now for years, is like solving that problem of what should be in the context window right now. That's the inner loop. The outer loop is how do you design a system that will get better at the job of the inner loop over time? So how do you capture signal from your users, how do you capture feedback from agents, how do you capture all that data, how do you analyze all that data? And then how do you bring that data back into that inner loop? Problem? That could be obviously, number one, massaging and changing the knowledge and data your model has access to, as known as retrieval, event generation. Or it could also be applying it to weights in different stages of the pipeline, including the LLM itself, the embedding model, re rankers, all the above. And so I mean, this concept we just sort of put out, like a week ago we gave a short kind of conference at Ramp's office in New York on context engineering, nyc.

Let's hear Ramp.

Listen, you got to do it.

You got to do it.

And this is the first time I've really seen this posit in such a clear way. And so we don't have all the answers, but I think this is the question people will come to ask in the future is not only how do we build a system which works today, but how do we build a system that gets better over time automatically, and that is the outer loop of engineering.

I want to get how you're thinking about Grok. Last week was an intense week from everything, from the fiasco. I think it was on Tuesday with the Grok bot just going wild. Maybe it brought more attention to the launch Wednesday, which was extremely impressive. But then later in the week, people were realizing that the reasoning model, or part of the kind of reasoning loop was like kind of making sure that it understood the boss man's views by what is Elon saying?

My question find the coverage and then reinforcing that wrongly and not so.

As I went into the weekend, I was thinking, this will be interesting to see how Grok can do in the Enterprise. And then this morning they announced like a $200 million deal with the government. So I think that's a good stamp of approval. But I'm curious what insights you have around some of the basically larger potential customers and how they're even thinking about model selection for different products and applications.

Yeah, I think you should probably wait. I think actually again, Ramp Plug has the graph of Anthropic and chatgpt, or Anthropic and OpenAI revenue growth. You know, let's wait to see when GROK makes a dent in that. That's not to say they will or they won't. I don't know. You know, there are these headwinds that you're talking about, which is, you know, you can either see as a feature or a bug, depending on which way you look at it. I mean, you know, there's things to like, there's things to not like. It's kind of hard to predict the future here.

I guess what I want to know is like, what is the, what is the flip side? Maybe we should just talk to ARA at Ramp about this, this. But what's the flip side of that chart? We know who the sellers are. OpenAI, Anthropic, Gemini and then Grok. And there's like kind of a long tail. But who are the buyers? Because I see individual like prosumer, knowledge retriever. I would pay $200 for a better Google types one of them. Then there are the developers who are paying for developer tools. But then what is like the third bucket of like buyers for tokens Right now it seems like it's a lot of companies with like kind of point solutions. Little. I'm stuffing an LLM in my enterprise SaaS, speeding something up, cleaning up some data. But do you have any insight into like kind of what's the third wave that's coming that everyone listening to this should start investing in now?

I think you're right that a lot of the spend, while a lot of the spend is consumers, you know, a lot of the spend is also businesses, enterprises are using it for really boring stuff. Process automation, broadly. You know, those buyers you should expect to be like way more conservative about like the tool they pick. You know, the adage of you never got fired for hiring IBM definitely also applies here. And so sort of reasoning about like, who is the IBM of this space and why? I think we'll like help you determine where that spend is going to go.

Yeah, that's a great question because from a, from like a, from like a being on the Frontier as a brand, I would say OpenAI, but then from like a, you know, scalable infrastructure, I would say GCP and Gemini. But then from like a model agnosticism view, I would say Microsoft. It's a tough question. I don't know if the IBM has emerged. Maybe you, maybe you have a strong opinion, maybe Jordy does. But overall, I mean, great, great, great catching up with you as always.

Yep.

Anything else, Jordy?

No. Congratulations on the, on the launch of the paper.

Yeah, I mean, yeah. Very exciting results. So thanks for, thanks, thanks for funding it.

Good.

Have a great rest of your day. We'll talk to you soon.

Cheers.

Take care. We need to tell you about wander.

Find your happy place. Find your happy place.

Book a wander with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning and 247 concierge service. It's a vacation home, but better folks. And if you're looking to take a vacation to the Goodwood Festival of speed, you might run into Dua Lipa. She says, I love speed. I love it. She's really a driver. Like this is not a large.

She wasn't driving the car, but she's in, but she was in the car.

She's not just hanging out.

And the delivery was absolutely fantastic. Just to give everyone some, some context. So we have Scott Wu from Cognition and Jeff Wang from Windsurf joining any minute now where the studio crew is getting. We're ready for everybody set up. We are ready whenever.

Bring them in. How you guys doing? Congratulations. Thank you.

Round of applause. Did you guys have a relaxing, relaxing weekend?

Yeah. Really chill weekend?

Yeah.

Yeah.

Easy, easy, easy. Nothing too much. Honestly. We, we both haven't really slept.

Okay.

So go easy on us.

Okay. Well, you had a lot of, you had a lot of people to take care of and honestly, when, you know, we, we heard the news, you know, a few minutes before we went live and we were fist pumping here in the studio because it's such a, such a fantastic outcome and I'm sure everybody involved is grateful for the work you, you guys put in to get this done.

But give us, yeah, kick us off with the high level. What's actually the partnership look like? What do you, how would you describe it in your own words of like what's actually happening? Yeah.

I mean this is a real acquisition.

Okay.

We are being acquired by cognition.

Okay.

Must feel great to say that too. You don't hear that much.

It was a tough Friday, I would say. I don't think anybody would want to be in my shoes on Friday. But we talked to a lot of teams out there. There's a lot of AI companies, foundational companies, and after talking to Scott and the Cognition team, it was done. In my mind, there was a done deal. I don't know if you guys know this, but Cognition was probably the only other team that we thought was smarter than our team, actually.

Yeah, no, and that's what we put. We put that together just on the show, seeing the announcement. It's like you take this incredible, you know, go to market sales machine that you guys had built, and you combine that with one of the best, you know, research and just hardcore engineering culture. Perfect overlap. And it's a. And it's a perfect, you know, match.

Perfect overlap.

Yeah, absolutely. And on the Windsurf side, I'll just, you know, obviously there were all these stories breaking out on Twitter and everything going on, and people were saying, well, you know, the thing that's left is just the shell. And we looked at it and we said, well, actually, I don't know if that's right. You know, it's the. There's. There's the code, there's the product, there's the customers. Most importantly, there's the whole team, obviously, and it's an amazing product. And there's a lot to do and a lot to build together. And so I think as we started talking about it, we found that there were just so many natural ways to partner, you know, both in terms of, obviously, the customers and customers would love, you know, are super excited to have a solution that basically combines both the IDE and the agent and then the technology itself, which is, you know, I think we. Devon, at Devon, you know, building Devon, have always focused on this kind of remote background agent as the thing that we have done. Whereas, you know, Windsurf is really focused on the agentic ide. And I think there's a lot of kind of things that play in there all the way down again to the kind of the fit of the relative teams and the team strengths and so on.

So it's almost like you predicted this because, like, the, like, these types of mergers are really hard because you have two brands that are popular. But Scott, you've been managing Cognition, and Devin, like, you've had the corporate structure and the product structure for a while. And so, like, it doesn't seem nearly as complicated as, like, where does this go? Do you have to rebrand one or another? I don't know if you have thoughts on it yet. I'm sure this is down the road, but it seems like you're actually fairly set up to. To integrate these two companies in a way that's not super chaotic to either, which is great. Yeah, yeah.

I wouldn't say it was intentional. I think on our end, for example, it's probably more that just we tend to over focus on software engineers because.

It'S just a lot of who we.

Are and how we think about it. And in particular, kind of just like these product and capabilities things. Whereas I think on the windsurf side, I think Jeff and the team, the whole team built out a really great kind of suite of all the different kind of functions.

Right.

You know, sales, deployed engineering, enterprise work, infrastructure, marketing, operations and so on, you know, finance, etc. And, and, and yeah, just turned out to be a really, really great fit.

Yeah.

Jeff, Friday, end of day, did you think you would be running a process so quickly? I know, you know, the obvious move would just be, you know, continue to run and scale the business. You guys had a lot of momentum, but what was going through your head Friday night? Because I imagine at the time you're also balancing. You were all of a sudden managing a team of hundreds of people that I probably all wanted your time and attention as well.

My immediate priority was just to get a lot of options on the table and to have a lot of paths forward. I had to tell the team, this is the path forward immediately because this is happening now. And then outside of the meeting, you know, I was thinking like, okay, who do I need to talk to? What is the best use of my time? I can tell you. I was on the phone for pretty much 24 hours nonstop on my phone. After, after the. All hands and me and Scott, we worked really fast. We met at our office the next day. He even brought in everything on a piece of paper to sign. He even brought.

He took the. There you go.

You obviously had to talk to a lot of. But that was. It was really. It was pretty dramatic, Scott. But it was cool.

It was cool.

Yeah, it was fun. And naturally, you know, because it has to be this way. We got things signed like one or two hours before we were able to put out the announcement. But, like, as soon as we started talking about it together, I think it was kind of clear for both of us, like, the way that we really make this a success is, you know, obviously customers kind of want to understand what's next. The team wants to understand what's next. You know, the whole world is talking about it. We want to be able to come out together as soon as possible and say we're going to do this. We're going to make this really amazing. We're going to make sure we take care of all the users and all the customers. We're going to make sure we take care of the team. And there's just a really great product here at its core. And that's what we want to focus on and that's what we want to build out together.

What is crazy, we talked pretty late on Friday night as well. We were comparing notes and there was almost no overlap. I mean, we talked about the team already, but even the product, the product is missing exactly what the other product is offering. And the ability to combine the products together is going to be so amazing for our users. It's going to be really cool.

Yeah, I think there's going to be.

A lot of fun stuff around. You know, you can have, have an agent work and then review the code. You know, locally in your ide. You could be planning tasks out and then kind of sending them off for work. You can be, you know, you can do a first pass with an agent and then come and look at the work and then kind of finish that up and do the finishing touches with, you know, with tab and all these other features and so on.

Right.

And we're really excited to build that out to together.

How is, how is the team feeling right now? I imagine this is the craziest 48 hours of their career.

Give us a preview of the slack emojis that are ripping right now.

Actually, we, we announced it at another all hands. All hands. And then we put another Monday all hands. And very, very different.

They're like, it couldn't possibly get worse. So it's only up.

We made it dramatic too. So me and Scott and the team, we really went out of our way to, to make this a very generous acquisition.

I'll just say that.

And I think I, you know, the team probably felt like, you know, their, you know, their career was over or something. You know, like it was very sad on Friday, but I think today, I don't know. How long did they collapse, Scott? They kind of cheered for like standing ovation or something. So I think the sentiment has shifted. The team is even more fired up to go. The whole product is still there. All the, all the things we had. And the GTM team is still there. And now they also have Devon to sell as well. So everybody's fired up now.

That's amazing. Yeah. Every time I talk to Scott, it's always because, like, he's done something really incredible and aggressive like like, and it's like, okay, like let's do, let's talk about this right now. And before I was making like documentaries. Take me a month to put out a single video. Now fortunately I have a dance daily TV show. So I can just have you on and we can chat about it. I do have some, some random questions I don't know if it's worth, I'm interested to know about, about that go to market team and, and channel sales specifically. There was somebody in the comments of one of Jordy's posts talking about how Windserve was growing revenue and, and, and working with channel partners. Can you walk me through how like what would you give advice to somebody who's structuring a, like a successful channel partnership strategy for a product like this? What incentives do you have to align? How big of a piece of business was that? Do you expect it to be a continual source of growth? I'd love to know about just that as a particular strategy as opposed to say Google Ads or going viral.

Yeah. We made the decision to go 100% partners back in February. The thing about a partner channel kind of like segment is like, you have to like, you have to like kind of feed them like you kind of kind of build belief with them in the product, but also show that there's an economic incentive too.

Sure.

And it did take some time actually from Q1 to Q2 to like get the top priority partners. Make sure those partners can easily sell on your behalf as well. And it really took like about four to five months to really see this like kind of take off. So some of our partner partners have already started bringing in seven figure deals even without even trying the product. They just go to their customers, they ask them, you know what, what they want and they just say it's Windsurf. And, and now they, they can say Windsurf and Devon.

That's amazing.

Yeah, I, I was getting multiple messages from people Friday and Saturday being like, I want this team, I want this team. And that was what. And I think that it is, it is just this whole deal and everything is really a testament to the caliber of the team. Which was one of the major disappointments Friday is if you were on the team that was staying behind at Windsurf. There's this sort of almost like signaling right. Of like I took this company, let's say you joined last August and you brought the product to market, you launched it, you grew it to tens of millions of dollars in revenue and then don't get to kind of continue the Mission and it's. I'm just really happy for everyone involved.

Yeah. And now. Oh, sorry. Yeah.

I was just saying that, you know, now we're all one big team and it's. I mean, we're kicking it into another gear. I think we're all so amped up to. To take this and just go, you know, we're going to go full speed basically for. For the next while and we get to play on offense now and go for it all.

So are you guys already making plans to get under one roof? What you want to see?

Yeah, we have a lot to figure out, but yeah, maybe get some sleep first. Getting to the point, there's kind of the product collaboration, there's the go to market collaboration, and then there's the literal kind of like team situation. A lot of work to do over the next couple months.

Well tell our customers like we got there back now and the product is going to get so much more.

Cool. That's amazing.

Well, you guys turned something which was in many ways, you know, a disaster. I was just waiting for. I was waiting for the New York Times piece on Monday. And getting to see the New York Times push out this news and turning this whole thing into a win is incredible. And you guys have done a service to the team and the industry, so thank you.

Yeah.

Honestly honored to get to be a part of it, to work with such a great group of people. And I just want to see you guys. You guys have no idea how things went with all the legal and everything last night in terms of figuring these things out. Are we going to make it there in time? Are we not going to be? It was. We've been through a lot and I think it's just the start of things.

So just so you know, Scott was like, we are announcing Monday hell or high water, and we just did not sleep for the last like 72 hours. Yeah, it's been a crazy.

It's been crazy.

I mean, we appreciate you copping us the stream. Do you have anything else, Jordy?

Should we. No, you guys deserve. Probably. Probably too early to sleep, but you both deserve a power nap at the very least. Thank you for coming on. Anything else that makes sense to share now or should we let you go?

We're all good. Awesome.

Yeah.

Stay tuned.

We're gonna have a lot more announcements soon. Thanks for having us.

Amazing, guys. Great work. Cheers.

Talk to you soon.

Congrats.

Bye. And next. Next up, we have Carl Pei. No, not yet. One second.

That is amazing.

Yeah. What a great story.

Absolute dogs.

Absolute dogs.

That is the perfect cemented in SV lore.

Tech lore for sure. For sure.

Didn't know. I mean, I knew Scott could fundraise. Yeah, I didn't know he was this kind of this caliber of deals guy.

Absolute dog. Absolute dog. Well, we'll read through some timeline. What else do I do? Figma has such a homey feeling. I spend all day and I really enjoy it. Feels cozy, safe, good to come back to shout out. Zach, giving Figma a little shout out. Let's bring in our first.

There he is. First studio walk in.

First studio walk in.

Let's go.

Welcome to the stream. Carl, how you doing?

Welcome.

This is fun. Welcome.

Thank you.

How you doing?

It's great to have you here. You haven't done this much before.

No, we've had a few people in studio. Done a few. Usually when we're in person, we're on the road, but good to have you come here. Would you mind kicking us off with an introduction on yourself, the company?

Yeah. My name is Carl. I have a company called Nothing. We're the only new smartphone company to have emerged in the last 10 years. Yeah, I felt like tech got really boring.

Yeah.

Like, I was a big, big tech fan. I got the latest iPhones all the time. I sat up to watch the keynotes. I was based in Europe, so the time zone wasn't always the best. But in the recent, like five years, I haven't done that anymore. Like, every iteration is pretty much the same as the previous one and I wanted to do something about it.

So what'd you do to spice it up? What was the first. The first feature? And then walk me through all the other features. Please show us.

So we're in a very big industry, right? Like 1.2 billion smartphones are sold every year.

Yeah.

So we're not trying to reimagine the category. We're trying to find our niche. So we started with industrial design. So if you look at our products, they look pretty different.

Totally.

Before actually starting to make products, we made a design book. We took inspiration from movies, from architecture, from furniture design, and we created this little mood board and then we based all our designs on that. So if you look at our different products, these are our headphones. This is a smartphone. You can immediately tell, even if you don't see the logo, that it's from the same company.

Totally.

Yeah.

So design is how we started. If you ask our users why they buy our products, design is the number one reason. And then lately we've been working more on the software side.

So it is funny. A cell phone should be the ultimate personal accessory. Yet the that. But yet they've all converged onto the same exact form factor. It's so funny because John and I have the same exact, same, exact same phone, and we get mixed up all the time. I grab it, he grabs mine.

Yeah. I mean, how much do you think we are at, like, the end of history for differentiating on, like, quantitative specs? Because for a long time it was like, okay, I need to go with this one because it has just a faster processor, but that kind of disappeared and then there was software lock in. You want the blue bubbles or whatever. But if we truly get to the end, then we're going to get it. Like every watch tells the time the same people spend a lot of money on certain watches that say something about them. Right. Same thing with cars. But what's your view on, like, just staying on the frontier. How stagnant is the frontier?

We're pretty much at the end. It's a hardware innovation. Like, we all have the same suppliers, we buy the same screens, same processor, same battery.

And you don't think that there's going to be any sort of, like, crazy breakthrough where, like, this happened with the, with the Apple Vision Pro where, you know, Mark Zuckerberg was certainly, like, doing everything to stay on the frontier. VR headsets. And then Apple allegedly does this crazy thing with their support supplier where they're like, you know that thing that you're making in the benchtop that costs like, thousands of dollars for this screen with this pixel density? Let's just pull this forward and just like, eat that cost to just leapfrog everyone and have the best screen now. They made a bunch of other mistakes, so the product didn't take off, but they at least nailed it on the screen front. And everyone agrees that the Apple Vision Pro has, like, the best resolution, at least of the modern VR headsets. So you don't think that there'll be like a, oh, wow, the next iPhone has twice the battery life. They figured out something crazy.

Not anytime soon. And I would probably say that the first phase of the smartphone wars are over.

Yeah.

And Apple and Google won have the biggest platforms. IOS has like, a billion users. Android has 3 billion. But I believe that the next war is just beginning to happen, and that's going to be all about reimagining the operating system. The previous generation operating systems are very mature already, but now with AI technology, I think we can build something really different.

So, yeah, how are you thinking about partnerships? I mean, John's probably said this 200 times on the show, but all he wants is better transcription. Siri.

Oh, you know what we did? I don't know if the people there.

Can see it, but hold it up, hold it up.

If you hold this.

Oh, wow.

Now we're actually recording this meeting.

Okay, okay.

Oh, you see the waveform there?

See the waveform? See the little red light blinking?

That's a nice touch.

And after the meeting, you can just have AI summarize the notes.

And my question is, is that using a transformer based transcription AI or is it using the old thing that gets every other word wrong?

We're using a model Whisper. I think we're using Whisper plus Gemini.

Okay. So yeah, it's going to be fantastic. Amazing. And for some reason no one's thought to implement that in the old stuff. And it's getting to a point where I'm like deeply frustrated with it because my standard workflow for I need to send someone a text message this long is I open up the ChatGPT app, I click the Whisper button, I talk to it, then I copy it, then I, then I paste it into my, into my text message box because I don't trust the standard transcription in iOS by default. It's very, very frustrating. Anyway, I want to talk about the second, the second smartphone war that's coming. Who are going to, who's going to be in that war? It feels like Apple and Google are going to continue to compete. You're going to be there. It feels like Jony I've and Sam Altman and the OpenAI team want to play in there. There's probably going to be other folks.

I feel like they, in a lot of the kind of teaser content, it's really around saying like, this is an additional device in your stack. They're not going as far as to say we're going to fully replace the phone.

Yeah, I guess why not just do try and do like an add on or like a fourth device, third device. I mean, most Apple people have like Apple watches, phones, the iPad, the computer.

You know, for me, like the smartphone is one of the best devices, especially for applied AI. Yeah, because it's got all the distribution. 1.2 billion are sold every year. We have it in our pockets every day. We charge it every day. We use it for four to six hours every day. Yeah, it's got the distribution, but it also has the data. We do so much on our phones. So if we're going to build a really intelligent operating system, we need the distribution. Scale, but also the depth of data to create a really personalized experience for each consumer. I do believe that in the next 10 years we will all be carrying another sensor as well, because sometimes the phone is in our pockets and we need another sensor to capture data for us. I don't believe it's going to happen that quickly though.

Other sensors, glasses. Something we've been debating on the show could be glasses. One of the challenges though is like you're walking around a lot of these. You know, the Meta Ray Bans are like actual glasses. Sunglasses. Are people going to be carrying multiple glasses that they're using differently throughout the day?

I don't know what the form factor is going to be. It might be glasses, it might be a pin, it might be a necklace. But I'm not super bullish that it's going to come in the next two, three years.

What do you think of the idea.

Of which is a challenge, just to be clear. If you're a seed stage company and you need to generate traction sales now, so to pick a new form factor, you guys can almost from your position be like, well, we have the supply chain. We know we can build hardware. We can in some ways kind of see how this next device.

Exactly. We built this machine right, because making hardware is very tough, especially smartphones. Like from having an idea to actually shipping the products across like 40 countries, servicing the consumer. It's a machine. We're building the brand, building the fan base. So I think whenever a new form factor comes out, it's quite easy for us to quickly follow and have distribution worldwide.

What do you think about putting a camera on headphones?

We've seen patents of other companies doing it. I'm just not sure what we would use it for.

It seems like Meta Ray Bans are like, we want a camera on your face and we will give you headphones for free via speakers that you can kind of listen on someone's call. But for the most part it's so. It's so close and so localized that if someone's taking a call and the meta Ray Bans back there, we won't really hear what the other person's saying. And then Apple, I think, is the company you're referring, you're like alluding to potentially has a patent and people have been kind of rumoring that, that the next version of AirPods Pro might have a camera on there. And so then you could ask Siri, take a picture of that, tell me what it is. Are you bullish on like the idea of visual intelligence because when I've walked around with the meta Ray Bans, I like, I like having an LLM at my fingertips, but I'm very rarely actually saying, what is this can? Because I can look at it and it says Yerba mate. It says Mata. You know, it says Mango. Can you. I don't need to do that. What I do need an AI to do on my glasses is tell me what's the history of Yerba Mate? And I could just ask that without taking a picture. I can give it the, like. I'll be walking around and some, some idea will pop into my head and I'll be like, I need to know the history of the state of California. Tell me when it was founded. Just give me the data. And it's nice to have an LLM at my fingertips, but I don't actually need the visual intelligence. I see it on my phone. I have the new. I have the new button where I can pull out visual intelligence. I'm not a daily active user of any visual intelligence product. But what are you, Are you bullish on it?

Generally, long term, I'm bullish. I don't know exactly how it's going to play out. But for your use case today, we already have a ChatGPT voice integration in our earbuds.

Yeah.

So if you connect your, these headphones to our phone, you can just click a button to make a phone call with ChatGPT voice.

Yeah, that's good enough.

Yeah, Good enough for now.

Yeah.

And do you think that the technology industry broadly just so badly wants there to be a major platform shift that they just talk that in some ways. And it makes sense in the context of like, Zuck. Right. Zuck has spent the last decades like, basically being like, you know, very, you know, the platform layer just kind of messing with it. Right?

Yeah, it's been a hassle.

It's been a hassle. So, like, it makes sense to invest tens of billions of dollars to try to make sure that you have a real horse in the next platform race. But at the same time, I think if you love technology and you grew up the way that I think the three of us did, you have this desire for there to be. Because it would just be fun and exciting. Even if you look at the phone, you're like, okay, this might be the dominant platform for the next 30 years still.

Yeah.

I'm super excited to see what the industry builds. We need people to try new things for us. We're still going to be focused on, on the phone because I believe there's so much cool software we can just build on the phone.

Yeah.

Of course we're thinking about what comes next and we're prototyping, but that's not where our main focus is.

Sure.

Was the changes to imessage a catalyst for you guys? They allowed like read receipts and delivered.

Oh, the rcs. I see it as a positive for us.

Yeah, it has to be a positive. I'm curious if it was. Did it make a material difference insert?

No, because I think it's not just about the color of the text message or it's not just about the read receipts. It's actually about the color.

It's about the color. The color is more important than the tag.

Way more important.

Really? Really.

And that hasn't been solved.

So yeah, they need to add that to the protocol. What color message do you want this to show up? And then all the Android users, there's.

So much social pressure of having the blue bubble. Like, my friend told me that at school, his daughter's teacher is like handing out homework through airdrop.

Yeah.

And his daughter was only one who had to raise their hand and say, I don't have an iPhone. Sorry.

Yeah, that's rough. Yes. The social pressure is big on the, on the, the. The blue bubbles. And the idea of like tech is almost a status symbol and a personal expression. I was joking earlier that, you know, if. If OpenAI wants to build a phone, they're going to have to contend with the blue bubble issue. They should do like gold bubble or like the Trump phone. I think Trump was launching a phone. Love your take on that at some point. But, you know, is there a hierarchy of colors that you should be thinking of? Is blue intrinsically more welcoming than green?

I think because of how the ecosystem has played out. You just have to go blue if you want a chance.

Sure.

Yeah. So we're on our phone business. We're not targeting the US very strongly just because of this reason. Like, we can't see a way past the blue bubble situation.

Okay, so yeah, walk me through the landscape of your non US competitors then. Who's actually in the market? What are the different markets that are interesting? Like, where's the biggest opportunity?

If we're talking about building the next generation operating system, I think we got to put China aside because China will have its own ecosystem with its own players, and the rest of the world will have its ecosystem. I believe you need data and distribution to be able to build the next os. So of course the smartphone companies will be contenders like Apple, Google, Samsung and Us, there's only four smartphone players outside of China today. But then there's other companies with a lot of data as well, like Meta, for instance, or OpenAI. So it could be any one of them.

Have you ever done like a postmortem deep dive on the Facebook phone?

I had it.

You had? Yeah. What? I didn't even know it existed.

Yeah.

Wait, it was actually a real thing?

It was real. It was built by htc, I think.

Oh, interesting.

Okay.

Yeah, I think Chamath worked on that project. An interesting, like, bit of Silicon Valley lore. But I wonder, I wonder what the. Like is the lesson there. Differentiate more on design. Is that what you learned from that? Or is it more like price for performance? What are the different levers that you think you need to pull to pull something?

Probably like pmf.

Yeah.

People have certain expectations from a smartphone, like great cameras, great battery. You can't underperform any of those things. You got to be competitive with the latest iPhone, the latest Samsung, and then you can add your flare on top of it. But your flare cannot be significantly different from what people are used to. I believe that in the future, for instance, that apps are going to go away. There's only one app on the phone, it's the operating system.

Interesting.

But we can't ship a phone without any apps. People love their apps, so you can have a future vision, but it's going to be like a step by step process to getting there and you got to bring the consumers along the way.

Yeah. Talk to me about how do you.

Think I can imagine we're a world in which more and more of the apps that you use day to day are just eaten by the operating system. Something like weather, for example. I don't love my weather app, but I will open it up from time to time. And there's a different scenario where I just, you know, transcribe what, what's, you know, how's the weather going to be this afternoon? And it just knows where I am and it just pops it up and says, you know, it's going to be 72 and sunny. But then for other things, like don't you think there'll be some networks that endure? Like for example, like something like a Strava, somebody's really into running. It's hard to imagine that, that just merging with the operating system.

So I see it slightly differently. I think broadly there's two categories of apps, like waste time apps and save time apps. Waste time apps are more like entertainment apps and the save time apps are just utilities I think the utilities are going to be really easy to replace in the beginning, but when it comes to Netflix and Spotify and those like entertainment apps, it's going to take some more time. But eventually the OS will partner with apps. And apps can still exist in a, in a container, just not as a separate thing you launch.

Yeah, yeah, yeah, that makes sense.

What have you learned from the history of like bloatware on smartphones? I worked at a startup years ago, 15 years ago or something that was, it was running a Siri competitor and part of their business model was getting like BlackBerry to pay them to pre install their app. And it was like a pretty good deal for BlackBerry. And this was, in this case, it was a pretty good product piece of software. And so it just, it just obviated the need for the customer to go and Click, pay for $20 for this app. At the same time, customers have notoriously been like, why do I have 75 Verizon apps on my phone? And they get kind of, they get kind of frustrated with that. What's the good condition for that? What's the bad case? And then where does it go?

The smartphone market is very big. So on the lower end everybody has a ton of bloatware and, and I think that's okay, ish for the consumer because they get a better priced product that they couldn't afford otherwise. But if you're operating on the high end and you have a ton of bloatware, then it's kind of weird because people are already paying a lot of money for your product, then have ads inside of the product. So that's kind of how I think about it.

Yeah. What are the key breakpoints for phone pricing Internationally?

Even internationally, probably around $300. Like that's the main, main category.

Yeah.

Here in the US it's very different. It's like a barbell shaped market where the ultra premium takes a lot of share and the low, low end, like burner phones take a lot of share.

Oh, interesting. How much are burner phones in the.

U.S. i don't have the numbers, but probably like 20% of the market.

Okay. Wow. Yeah. I meant in terms of price, like, are we talking like 150? Yeah. So really, really cheap. Interesting.

I think how much capital would you need to deploy to be able to make a cheap phone, high quality phone in the United States? Are we talking like, like, you know, manufacture locally? Yeah, yeah, yeah. Like would you need like $50 billion.

To set up all this Capex or opex problem?

Well, yeah, the Question is, like, if you have to basically like, re. Not just do the assembly, but like recreate all the sub suppliers. It's. You start adding that up and. And it gets.

I don't really know because we partner with a bunch of factories all over the world. So if we wanted to manufacture in the US we wouldn't know who to go to to partner with. We do a lot of our manufacturing in India.

Yeah.

And there we've seen that the government has been pretty successful. I think, you know, India is now the second site in terms of volume after China.

Yeah.

And they have incentives to, like, if you manufacture locally and export locally, you get some cash back. And I think that model really works. So if the US Is serious about this, it's more of like the structural incentives that need to be set up for this to run.

Yeah.

Interesting. What would it take to. Is there anyone trying to build like a Tata Nano of phones? Like something that's like smartphone level, not burner phone, but at like 75 bucks a pop?

They already exist.

They do, yeah.

It's like for Africa.

Okay. Are they delivering on the promise at all?

Yeah, they're doing a great job in Africa and other markets where people need cheap phones. You can get a smartphone for like 50 bucks now.

Yeah.

Interesting. What about. How do you guys think about. Have you integrated wallets for stablecoins? Is that something that should happen at the operating system layer for a phone? I'm sure you've been.

I haven't thought too much about it. For me, like, there's no real use case for crypto yet.

No real use case for crypto. What about store of value, bro?

We just haven't seen anything with real PMF yet.

What about bitcoin? That's pmf.

Yeah.

That's like the. One of the only things.

Yeah. You're probably not trying to sell that many phones into sanctioned countries where the citizens really want to get dollars on chain.

Can you walk me through the relationship between manufacturers and telecommunications companies? I remember, like, years ago.

This is the absolute last question because I know you got a heart out.

Oh, yeah.

Just got to hit the. We got to make time for the gong.

Yeah.

Every market is different, but in the US there's basically very little relationship between brands and carriers because there's only three carriers and they're playing a user acquisition game against each other. So their main way of acquiring users is not through selling unique hardware. It's mainly the bundle. Right. How much data do you get? Do you get free Spotify? Do you get Netflix? So for a small brand like us, there's basically no play with a carrier unless we build a groundswell of support with our community or have like really, really innovative products. I think our products are becoming more and more innovative and fresh, but we probably need a couple more years before we have a product that's really different for the U.S. cool.

Very cool. And then we'll let you go.

Well, do we need a number? Can you give us anything? How big is the company? How many employees you got? How many phones have you sold? Sold?

We're about 750 employees.

Wow.

We're 0.2% market share globally.

That's actually huge. Market is. Congratulations.

Thanks. We've done more than a billion in revenue and we're doing a billion this year.

What? Okay, okay, okay.

Please, please hit the right behind you. That is very well deserved. Give it, give it a solid.

Congratulations and thank you so much for coming by. Really appreciate having you in person.

This has been incredible progress.

Thanks. So.

Great stuff.

Really buried the lead there for sure. Are we hopping straight on with Garrett or. We got 15 minutes. I guess we got 10 minutes until the next minutes. I want to go through. Is there anything else you want to go through? I like this. In the how to spend it section of the Financial Times, we have a review of something we've been hunting for, which is the best hotel gyms in the world.

There we go. Here we go. So if you've been to a hotel gym on chess day, you've probably been disappointed because many. It's hard to find a hotel with. What is it bigger than? 50 pound dumbbells.

50 pound dumbbells. And we believe. We haven't fully verified this, but we believe this is due to a market failure in the hotel gym. Insurance market.

Insurance market.

And so I think that there are two tiers of insurance. One for like a CrossFit gym or powerlifting gym that has free weights and also barbells and you can do hang clean and clean and jerk and all that stuff. And then there's a different level of insurance underwriting for things like treadmills and bicycles and free weights that go up to 50 pounds. And so that's why when you're traveling, you see a lot of gyms that don't have. They won't. You can't do squats. There's a leg press machine. And so they lean on the machines because the risk of injury is lower, but the risk of gains is also lower. So we are going to understand where you should travel because to get Access to the world's best gyms Hotel gyms are a mixed bag. Even in top flight spots. Working out can be jar can be a jarringly subpar experience.

That's right.

Could not agree more. Machinery might be decades old. The water cooler might be might date to the 80s office era should get Rorat. For years, hospitality didn't invest in gyms, as unlike with restaurants and spas, the gym usually doesn't bring in any extra revenue. That's an interesting dynamic I hadn't thought of. So yeah, I mean, if you put in like a Nobu in your hotel, that's going to drive a ton of extra revenue, whereas your gym will not.

So increasingly charge me per set with 75 per pound. Yeah, per pound.

And then you got to put the bigger weights in because like I'm going.

To want to lift heavy profit maximizing strategy so.

But increasingly top hotels are pivoting. It's important I keep the same fitness routines as when I'm at home, says Emily Oberg, founder of clothing label Sporty and Rich, who works out six times a week. Her favorite hotel gyms include the Ritz Club in Paris for the pool and the Il San Pietro in Positano. It has. It has alfresco cardio equipment, outside cardio equipment, tennis courts and views. It's motivating if the space is also beautiful. Gym lighting needs to make you look and feel good while working out, says interior designer Kelly Werstler. Acoustics should amplify without distracting materials. Should perform functionally while still feeling refined. I couldn't agree more.

Have you been to any of the proper hotels? Santa Monica proper gym?

I haven't.

She designed the Santa Monica proper.

Oh, really?

Kelly Werstler.

Yeah, there you go. Okay, so the best concept. These are awards from HTSI in the Financial Times. How to spend it the best concept goes to Power in Ireland, a great gym chain in the lobby of the hotel. Power, Ireland's leading boutique workout provider, has partnerships with the Dean Hotels in Cork, Dublin and Galloway, and the Mason, also in Dublin. In the gyms, there's a functional fitness area with a custom athletic rig and Astroturf track, a lifting area, free weights and a cardio equipment cardio area equipped with technology, Technogym skill run treadmills, rowers, ski Ergs, Watt bike pros and more. There are several 45 minute classes including RUN and METCON. Guests can recover afterwards in the Hyperice Suite, which is stocked with massage guns and foam rollers. Rooms from €200. Not bad. The most scenic is at the Amonzo at in Greece. It's not my own property, of course. Summer escapes are all about picturesque locations, seafront settings, a quaint and quiet locale, and often ancient European history. This Amman property gym offers all the above and then some. The resort has its own UNESCO protected ruins, but it's. But it's the outdoor shaded weights area overlooked.

Let's give it up for hotels that.

Have ruins that is the real cell. According to Tim Blakey, a personal trainer and founder of the workout at Prime Body. We go on holiday to be outside, he observes. There's a slight breeze, the sun is usually out and you can get your nature fix while you lift weights. Rooms from 1500 Euros also not bad. Oh, hey, they have a proper hotel. Did you know that this was gonna be in here? The proper hotel in Santa Monica. Best for recovery. It has an immortal chamber where tech in fused lounge chairs pulse with electromagnetic fields to deliver red light therapy to reduce inflammation and combat jet lag.

It's funny because when I walk.

Testing. Let's go.

Okay, that's cool. Yeah. It's funny though when you walk into a. I mean you're going into a hotel gym, what do you really care about at the end of the day, John? How much weight.

How much weight you can throw around. So and they say at the proper hotel in Santa Monica, don't be surprised if you see Brian Johnson on your way out.

My personal favorite hotel gym, Sensei Porcupine Creek. This is Larry Ellison's basically Four Seasons competitor and they've done a really nice job with it.

Okay, we got to go through. There's a few others. The most exotic, the best for classes, but the most exclusive. This is what I want to know about. The most exclusive hotel gym in the world, according to the Financial Times. Suren at the Emory in Knightsbridge, guests of the all suite Emory have full access to Suren, a comprehensive wellness club with only 140 vetted members. It costs 15,000 pounds per year to join.

I thought you were gonna say per month.

The gym is small, but it has 360 degree mirrors, so it's feels bigger. LA celebrity trainer Tracy Anderson has licensed her name to Siron. It's the only place where you can book her dance cardio workouts in the uk. Alternatively, guests can request.

Is that kind of disappointing though that the nicest like the most exclusive gym in the world is £15,000 a year. Like shouldn't somebody make. Shouldn't somebody make the $100,000 a month gym?

I think you just designed home Gyms in mega mansions.

Maybe I did with an entire team of specialists.

So you can pay the £15,000 a year for a membership or you can book a suite for 1600 pounds per night. So maybe, I don't know, what is that? It's in pounds. So that's like 15 bucks US or something.

It's pretty funny. The dynamic of this is the most exclusive club in the world, but if you just book a hotel room, you.

Can go to it. Yeah, that's a bit silly, but this is where it gets good. So if you book the hotel room, it's suites only. You book the hotel room, it's going to cost you £1,600 a night. They have a weights trolley that you can request and it brings the weights to your suite. It includes dumbbells, kettlebells, skipping ropes, yoga balls, resistance bands and a leather gym ball. A peloton bike can also be installed at a whim. So you can be in your room and be like, I don't want to go to the gym. Bring the weights to me and they will bring them to you on a trolley. That is luxury. I love it. Anyway, lots of fun in the how to spend it section of the Financial Times. Any other news?

Let's do some timeline.

The other. Yeah, some timeline.

I like this post from Dylan Patel said AI Bros. When I tell them Boeing outperformed Nvidia Meta, Google, Broadcom, AMD and Microsoft. This year I got to pull up the Boeing chart on public.com. i certainly do. So year to date, up 34%. Let's give it up for Great American.

Only 34%. How does that outperform Google, Nvidia Meta? I guess they all round trip because of the tariff things.

Nvidia's only have 18 because a lot of the election stuff started getting priced in.

Yeah, yeah, yeah, totally. No, that makes a ton of sense. Well, we have our last guest of the show joining now. I'm going to let you take the intro. I will be right back.

Let's do it.

Can we bring in Garrett? I will say how. I will say what up? How you doing?

What's going on?

How's it going?

It's great to have you here. This is the first time correct? Do I have amnesia? First time. I'm surprised it took this long. It's great to see you. You've got some exciting news today. Hit us with a quick intro on yourself and the company and then we'll get into the news.

Sweet.

Yeah, My name is Garrett, CEO of Pipedream Labs. At Pipedream Labs, we build the infrastructure to make autonomous logistics scale.

Amazing. Give us the quick. Catch us up to speed on the history of the company. I know we first met back, I think it was in 2021. You've been very busy since then. And then I want to get into what you're announcing today.

Yeah, well, likewise, man. Congrats on the show. Thanks for studio look. Sick.

Yeah. Have to come in person next time you're around.

Yeah, I got you guys. So yeah, the announcement is something we've been working on for a long time. The kind of first product we were working on at Pipedream is what we see as fiber optic cable for autonomous logistics. How do you move a lot of things really quickly and really cheaply? So we put underground pipes in cities and then we have these robots that go through the underground pipes and deploy orders to what we call portals. But they're like ATM sized kiosks. It's a modular system made to be able to be retrofit whether that's into a building, an apartment building underneath the road and uses just standard construction methods that cities are used to using for non invasive construction going into buildings and through public right of way. So we got, you know, that's something we were working on. We got that to a really good place and something that we kind of thought about a lot when we were starting was, you know, if we're working on the fiber optic cable, we wonder if there's also going to need to be kind of like the cloud architecture for autonomous logistics. Because as you start to move things really quickly and we've seen this with drones already, like drones are like this magical. They're magic. It's incredible. And they're so quick and yet it's still taking like 15, 30 minutes sometimes. Like I had one that took 40 minutes because it's so hard to pick out of those stores that they're picking from. And they're picking out the stores because those are the closest storage areas to where the customers are. So it's kind of like an inefficient system. You have this really fast delivery method and a really slow processor to get that thing out of storage and into the drone. So that's something we thought about a lot. So me and my co founder Canon, we're huge retail nerds. So we're always diving into Costco and Walmart and Amazon and the history of just logistics and retail as a whole. And what we realized is like, man, it is time. It is time for autonomous logistics to scale and no one's stepping up and building these fulfillment centers that are made to rapidly dispense. So we had learned a lot, worked with a lot of partners who taught us a lot about how to run those facilities and decided, you know what, it's time to go. So we just acquired our first rapid fulfillment center here in Austin. It's going to be what feeds the network here in Austin. And yeah, that was the other announcement.

Just so I'm getting it correctly. So is this set up in order to serve the original Pipe Dream infrastructure and then also traditional drone delivery as well?

Yeah, so the way that we see it is if you have something at a rapid fulfillment center, the goal is how do we get it into the customer's hands as quickly and as cheaply as possible. And a lot of the times that's, you know, you may want to move it through the Pipe Dream network to a closer location. You might just want to put it into a drone directly. It may be a heavier item, it may be a bundle of items. You want to put it into a self driving car. I think for the foreseeable future, I think it's going to be even like.

Really.

Load dependent as well. I think autonomous vehicles and drones are going to be kind of like the GPUs of the autonomous logistics scale up. We're going to be constantly constrained by their availability. So our goal is to be really good one at dispensing really quickly and dispensing into the right modality and constantly figuring out what is that right modality for that address, what is the right modality for that time and making sure that that order is getting into that right modality and getting to the person as cheaply as, as quickly as possible.

So, so what is the, over the next year specifically? What, what delivery form factor are you most excited about?

Yeah, I think the two that we're most excited about is autonomous delivery right now is still pretty cost constrained. The cost per delivery is still really high. So you're either passing that along to the customer or you see a lot of the times it's heavily subsidized. I think especially on the drone side that's going to come down a lot over the next year. And so that's the one that we really want to be the best at doing. We have an internal goal here at the team that we want to, by the end of next year we want to have done the most drone deliveries of any company, which is crazy. It's not like a super high bar right now, but I think that is a resource that's really underutilized.

And then what is the regulatory framework that drone deliveries are needing to happen within right now, specifically in Austin, like I imagine peer to peer, I could just put something, you know, attach something to a drone, take it up, fly it over to a buddy's house and drop it. But then if you're trying to do this at scale, I imagine, like you have to keep, keep a lot of other things in mind.

Yeah, it's right now it's not such a straight pathway to being able to do it city by city. So each company who is doing the drone deliveries has to get their own regulatory clearance. And so it is like pretty, pretty big hassle. But that's changing pretty rapidly. We just had the executive order to make that pathway clear up. I really think especially as you start to get more consumer demand for it, I mean, it's just if you go and look at the reviews from people who have experienced a drone delivery, it is, it's borderline a religious experience. It really changes, like how you think about commerce as a whole. So I think that's going to change for the next couple years. You know, I think there's a lot of companies that have died saying drone delivery is going to, you know, scale in the next two years. They've been saying that for the last like 15 years. So it's a tough thing to time. That's why, you know, we want to get really good at drone deliveries and be really careful and then just kind of learn right now, be ready for that scale and really focus on pickup as well. Pickup is a huge industry. It's one that doesn't have a lot of infrastructure to it. So it's still such a bad experience, but it's still one of the fastest growing ways that people experience retail is ordering ahead and picking up, especially in groceries. So that's one where we saw the ability for us to go into grocery, make it a really, really great pickup experience. That's something that you can do right now, make really good margins on learn. And then as those autonomous methods decrease in costs specifically due to better regulation, then we'll start to add that on as we go along.

Amazing.

Anything else?

I think that's it.

Good to see you again. The last time we talked, or I mean, we've talked a few times, but I interviewed Garrett like two, three years ago or something, all about how Amazon had been talking a big game about drone delivery. And I was like, let me talk to everyone else who's been, been, been working in this space. And it's. It's a fantastically massive market that is incredibly difficult to actually get things to work. And once they start working, it's great. So good luck.

I think the key thing is staying in the game, finding, finding a sustainable business model that's not dependent on, you know, dependent on adoption over a 15 year period versus a 15 month period.

I remember when we were. When we were talking, we were new. Like, it's just such a. It's one of those businesses that you just, like, naturally noodle on. Like, okay, should you go to, like, people that are building new prefab home communities and install a bunch of pipes in the ground when they build the buildings, or should you do some sort of boring company thing where you get really good at drilling into old, old, you know, neighborhoods? Or should you do flying or driving? Like, there's so many different modalities. So very exciting that you're. That you're kind of exploring all of them now. So congrats and good luck.

Appreciate it.

Thanks for having me on stuff.

Yeah, we'll talk to you soon.

Talk soon.

Bye.

All right, see you guys.

In other AI acquisition news, we almost missed it because, you know, Sundar did everything he could to drown this out, but we surface the post and it's about Mark Zuckerberg acquiring Play AI poaching the entire team, and they are joining the Meta Super Intelligence lab next week. Zuck cannot be stopped. He has the Thanos gauntlet there. They completed who's on the.

Who's on the. Who's on the. The who's on the cap table or.

No, it's the researchers. I guess it's the team. I feel like. I feel like we were played a very big role in this meme becoming bigger, but a lot of people have run with it in different directions, and it's been amazing watching it, watching it grow like this. The Play AI group will report to Johan Schalk. Cool name. I'm not sure how to pronounce that. Who recently joined Meta from a separate voice AI startup called Sesame AI. I remember this. That was a controversial one because Sesame is a new, new company and, you know, shouldn't be losing people to Zuck already. But the entire Play AI team is coming over, and I believe it's a YC company because I saw one of the YC partners posting about, hey, this is another hundred million for Gary the Dark Knight. The dork knight rides again. Never go up against him.

Well, in other news over the weekend, Janic Sinner made history as the first Italian man to win a singles title at Wimbledon. Naturally, this is according to Bezel. He celebrated with a rose gold Rolex Daytona oyster flex on the wrist. So thank you for doing the most important work. It looks absolutely fantastic.

Yes. Cover of the Wall Street Journal for this, by the way. In bloom. Italy's Jannik Sinner celebrates Sunday in London after his men's singles championship win over Spain's Carlos Alcaraz. I don't know follow tennis enough to know how to pronounce that. Poland's IGA Swatek defeated blah, blah, blah. And more coverage, but he's looking happy with the trophy.

I love how little attention you play to tennis, John.

You know what I do follow though? I follow hitters, I follow watches. And Jensen absolutely mogged the rest of the tech.

He really did.

With one of the most insane, Richard Mills.

You don't see him in a short sleeve T shirt much.

No, he took off the leather jacket for this. And look at this. Such a come from behind moment. Because we were talking about his, the lack of wrist game and we've been doing wrist checks on Jensen for, for months now.

Karp was, was early and right.

Karp's. Yeah. Oh, for sure. He orange with the orange strap. Zuck's been ramping up, putting up insane FP Jorns really.

Anytime he steps out of the house.

He has a cubitus. He's got, he's got a stack collection. And everyone was saying, even, even Jensen's family relative Lisa sue over at Amen AMD was mogging him with a, with a Rolex we saw. But now never doubt steps out, never talk.

And the market loves this. John, the stock's up. Oh, the stock's up 3% over the last five days. Obviously, obviously because of this.

Obviously because of this. But this watch is, is really, really special. I mean what I, what I love about it is that it's just so thin. Like when I think Richard Mille, I think of the, the silhouette, the outline, this kind of like. Well, it's almost like a bowed rectangle, which I think is a very cool, unique design. I think of the skeleton. It's checking those boxes. But then it's just so thin and it looks fantastic. And Jensen is wearing it very well.

So we need to do it, we need to do a deep dive on RM because the story is crazy. Basically this long time executive who decided he was going to make a luxury watch brand.

Yes.

And to actually just decide, it's easy to say I'm gonna start a watch company. Yeah, very, very much harder to like actually crack the code and be able to sell watches in the.

I'm surprised some of our friends haven't started watch companies given how many companies they're spinning up every two seconds.

Yeah, a certain one in particular.

It'll happen eventually.

It'll happen eventually. The last another post somebody posted RIP McKinsey, you don't need a 300k consultant anymore. You can now run full competitive market analysis using Grok4. And our friend Buco Capital bloke says I'd like to see an LLM provide regular consulting to turbocharged opioid sales, be a trusted partner for autocratic regimes and help young people get hooked on e cigarettes.

Very negative on the consulting firm firms. Who speaks for them?

But we got it. We got it, we got it. Let's give it.

They do a lot of good work, the white shoe firms. They do great work. Yes, some mistakes were made, but overall good farm team for future PMs and big tech.

Yeah.

So in other news, Monumental Labs the has raised some money. Over eight million dollars most recently in a seven million dollar seed round led by Alexis Ohanian. 776. This is the company Stone Carving alert. Yeah, they're carving stones. They're carving.

You heard Monumental Labs, you thought new new foundation model.

No AI robots can already carve stone statues. Entire buildings are next. The stone carving startup, Monumental Labs has raised money. They have a new HQ, almost 40,000 square feet in Greenpoint, Brooklyn.

Right.

So.

So the cool thing here. So they are opening this new facility in Greenpoint which makes total sense to be the home of what will become the largest fine art stone fabricator in North America and possibly in the world when it opens in the fall. Makes so much sense and a lot of people would. This company was non obvious but then if you think about what we would pay for stone statues here in the studio and I think what a lot.

Of people will pay all the Mag7 CEOs potentially, just hypothetically. I don't have like a mood board for that or anything. Yeah, it's not like saved on my phone.

We want to make the mount rush rush. More of the manosphere.

For sure. Andrew Huberman. Yeah, yeah, yeah. It's Mag 7 watching over the manosphere.

No, but I think you know, if they can really croc.

Congratulations to the Monumental Labs team. That's a very exciting company and led.

By Alexis Ohanian, a friend of the show and also he says looks like the secret is out. Can't wait for you all to watch because he is going to be a guest shark. He's going to Shark Tank, the upcoming season of Shark Tank, which will be Wednesdays this fall on ABC and stream on Hulu. He's going to really have to really go shark mode because Alexis is known to just be a really nice guy.

Really founder fairy, very sharky. That's very funny.

I mean sharp elbows, I'm sure. Getting into deals, sure. You know, but, but yeah, he's gonna, I mean, but there's always room for just offering an entrepreneur on Shark Tank a nice fair deal, you know, he's not gonna be offering, you know, like just give me 20% of top line for.

Yeah, I mean as a personality, I think you'll fit in great. I think you'll really add something to the show. So very excited for him. Go check it out.

And final post from Adam Singer.

I was gonna end on this too.

CMO of Adquick out of homemade Easy and measurable says love seeing out of home from the new gen of tech media brands. Great brands do things online and in the real world. Our software makes this really easy for all and I just want to give a quick shout out to Adam for betting on us early. Yeah, bet on us when we just were two guys, two suits and a stack of posts. We will never forget that. And go do yourself a favor and sign up for ad Quick after you're done with that.

Make advertising easy and measurable. Come on people. I don't even need it.

It's really not complicated. After that leave us a five star review on Spotify or Apple podcast, wherever you listen. And I can't wait for for tomorrow. There was a couple weeks of news since last Friday and I'm sure there will be quite a bit more tomorrow.

The other piece of news, the information launched a competitor to us and they had some technical issues. So if you know things about computers, please contact them.

Yeah, tell them, tell them the technology.

Brother sent you and that's our show. See you tomorrow.

Cheers.

Goodbye.

Have a great Monday.