This intense demand for compute is going to continue for at least another 12 months.
If Nvidia's crushing it, everyone's going to crush it.
That sound you hear is the Google free cash flow and the Oracle free cash flow just disappearing down the drain.
What are we discussing today? Nvidia, they deliver a record $96 billion quarter. Jensen hits it out of the park, baby. And Nvidia nears a $12.9 billion deal for Hugging Face. We then move to Sam Altman as they cut off Cursor. And then finally, we finish on Instinct, the AI assistant that has raised at a $2.5 billion valuation and has everyone very excited. This and so much more in today's episode.
Our job is to sniff out winners, stuff capital into them, and broadly speaking, stay out of the way unless they're literally crashing the car. That's the job in a nutshell.
Literally, the amount of code we're building is 100x. We didn't realize we would all be building compound companies. Now would be a good time to panic about cyber.
Ready to go.
Guys, I'm so excited for this one. We have a lot to discuss and I want to kick it off with Nvidia. Nvidia crushed it again. Jensen standing on stage with a $96.2 billion quarter and immediately went shopping, nearing the $13 billion price tag with the $12.9 billion that they paid for Hugging Face, which was confirmed just after we did last week's recording. So, I want to separate the two. Let's start with the stellar quarter for Nvidia, the record revenues. How should we think about this? What should we take away?
That it's a great business and you wish you bought the stock. I mean, I think we talked about it last week. It was funny because we were in that period when we knew we were recording before the announcement and they were going to come out after the announcement. And I always worry when that's happening you look like an idiot. But I felt complete confidence that it wasn't going to happen for Nvidia. And that's the big, that's a step back comment here is that right now the demand for their product is such that they're entirely supply constrained. So the probability of a near-end miss is plus or minus zero. Right? So the only thing that was interesting really and new was the guidance for next year 2027 where they, you know, the analysts had 40 or 50% and they're talking 70% and saying it's supply constrained. So that was, that was a takeaway, right? And you know, that's obviously, you know, rightly or wrongly, we'll see in a year, a statement that this intense demand for compute is going to continue for at least another 12 months, and it's coming from the person who probably knows best. It was kind of interesting that I think the CFO said enough, enough complaining and caveating about the round-trip deals. Okay, they're working for us. And I think between that, you know, another whatever $35 billion deal with Anthropic, right, for their data center, I think, I think any misgivings we have in the short term have been disproven here.
So when people worry, oh my, how could Nvidia miss? Step back. How could Nvidia miss? Right? There's only three things that can go wrong. Yeah. You know, right? Either their direct customers stop buying compute, just not going to happen. The hyperscalers are exploding. Everything's singing, right? The second thing that can go wrong is all these people worrying about, you know, the round-tripping and the financing. They're kicking off so much cash right now. And as long as demand is working, you're exactly right. And the CFO is right. All this stuff's going to work. Really, the only thing if you step back that can go wrong at some point, and it's not today, is end user demand because all this is predicated in the end. Yeah. Everybody gets to sell chips to hyperscalers provided hyperscalers can sell compute to OpenAI and Anthropic, provided OpenAI and Anthropic can sell kind of intelligence to end customers, and the whole thing works provided the end customers keep exploding. And right now they are. So you should say to yourself, as long as that's happening, everything down the line's going to be plus or minus fine, right? And right now it's plus or minus fine. And what it means is that the thing that will probably unravel it, we're not going to unravel because the circular deals on their own unravel. If it does unravel at some point, it will be because you're forecasting 5x growth in end user demand. Next year, you get 3x growth and then the whole thing goes wrong. That's, that's the, until then you can opine pretty safely about Nvidia and say, 'Yay, yay, Nvidia.'
What about rise of competitive threats?
That is the last one I did. Yeah, it's so funny. You're so... That was good, Harry. Is that there was one back of my mind I was saying maybe demand for compute remains high. They're still selling, but instead of selling 400 billion, they sell 360 billion because someone takes 10%. It is a fair comment. It is the only other... I was trying to simplify because normally I try and make things too complex, but you're exactly right. The other risk is there's demand for 500 billion in chips and someone else gets 10% of it. And, and Jensen's going to be pretty pissed if that happens. But there you go.
Jason, can we just frame this moment? I'm like jumping inside at Rory saying that I'm exactly right.
Yeah, you're right. I was, I was, I was simplifying. I was trying to get up to your level, Harry, and I just overshot.
I think in general, you know, Nvidia's market share remains dominant, especially by revenue, right? And so, broadly speaking, and listen, everyone's buying ahead, there's a capacity war, but broadly speaking, if Nvidia is crushing it, everyone's going to crush it. Like, everything's green. Now, could individual competitors, could their positions ebb and flow? OpenAI versus Anthropic, Harvey versus Lore, whatever. Sure. But if we come, like, it just means all just raise, expand the growth fund like Andre. That's what it means. Everything is green, green for now. And this includes Nvidia's backlog, Nvidia's forward bookings, and if Nvidia gets a hiccup, we can excuse it, but it should be a yellow light, right? But man, especially for investing, game on, man, just like, let's level up the next round.
And also they took the rare step of projecting 70% revenue growth for next fiscal year ending Jan 2028, way above the 44% street expected. So the party continues.
And it really does because if you look at all the projections that people are doing, right, both for Nvidia and the hyperscalers, they all take the following form, right? Explosive growth of capex, explosive growth of Nvidia revenues, and then, quote, at some point in time, a normalization of growth which will allow end user demand to catch up and therefore the hyperscalers will become cash flow positive again and, you know, the world will be wonderful again and you value things on a multiple of free cash flow, right? And of course, every time everyone decides to double down on more capex, that date gets pushed out, which is another way of saying the end user demand has to be bigger to make the math work, right? And clearly Nvidia got the signal from their customers, the hyperscalers, you know, the CoreWeaves, the OpenAIs and Anthropics themselves, the Googles, that basically said next year, you know, we probably at one point everyone's analyst models were we'll spend a lot this year but so help me God we'll slow down next year. It's all on now, right? It's all green and go now, which is different than saying it'll be green and go forever. But yeah, 70% guidance in a world where everyone was saying, oh, in 2027 things will start to normalize. No, this is a statement. No, the biggest semiconductor market in the world is going to continue to grow at 70% instead of a typical 10% for another year. It's a big ass, it was a big ass statement. That sound you hear is the Google free cash flow and the Oracle free cash flow just disappearing down the drain. Man, they better be right.
Well, thanks for the intro quote there, Rory. The second part of my statement was, and it was further reiterated about their buying Hugging Face for $12.9 billion. It's still, I don't think it's finalized or confirmed by any means, but it's definitely much more advanced than when we last discussed it. Is there anything subsequent to our last discussion that we should add or think about?
No. Man, making $120 billion a year selling compute decides to buy a company that helps make compute more cost-effective so he can sell more compute. That is the summary of the deal, right? You know, at the margin, if you're Nvidia, look, you were ecstatic that OpenAI and Anthropic happened because they proved that the market you were in was bigger than anyone ever imagined. And early on, no one could have done what OpenAI did, and therefore, you know, Nvidia has been a real beneficiary of... But now that the category is established, the simple question is, if end users have a trillion dollars to spend on tokens, would you prefer to, as Nvidia, you'd prefer that money flow through open-source people at 30% gross margins where you can get all that compute versus 70% gross margins at OpenAI and Anthropic where they keep more of the money. Open source is good for compute salespeople. If you're selling GPUs, you want everyone else's margin to be lower so yours can be higher. So it's just exactly right and rational.
I do think it's more than that. I think you're right, of course, Rory. I think it's more that just Nvidia is playing an endgame now where it has to win every segment of the market. It just has to win open weights. And if that means overpaying for Hugging Face at 110 million in ARR, if that means subsidizing whoever, it doesn't matter. It's not that I don't know that Nvidia wants open weights to beat Grok and Open. It just needs to. It's just clear. And that was why that memo from a couple weeks back, or memos that Jensen did, his first tweet ever, right, in support of open weights, they just have to win. He's just committed to winning everything. All, whatever LLM, whatever inference is, he's committed to winning the majority stake, 70%, 80% of every dollar here. And so 12.9 is, you know, no way I think it's worth it in isolation. But if it helps reinforce that, it's utterly defensible, right? It's utterly defensible. But it's a moment in time. I would sell, too. I would sell. Clem, good job. I would sell. What did Hugging Face start off as? A social network for people who like pets or something like that.
It was, it was a Tamagotchi for like teenage loneliness.
Yeah. I mean, this is the greatest tilt of the history of mankind. This is much better than Cursor going from whatever a CAD tool. This is the tilt hero. You know, it is interesting in terms of the sign of the times. You know, Slack was bought for 27 billion, which is rough and tough, twice what Hugging Face apparently is going to get bought. And we fell out of our chairs and it was the high water mark of that era, right? But it was at a billion in ARR. This one's at 100 and some odd. So, I guess it makes sense. The best premium should be an order of magnitude higher than the high water mark of the last era, but it's still loopy.
Next topic is Sam and Elon. Why won't the kids just get on, boys? Context, obviously we saw over the weekend OpenAI cuts off Cursor, and Mike TR responded by saying, you know, oh no, woe is me, we so love partnering with you, Sam, but you know, the 5% traffic that we have going to OpenAI will be devastated. I thought it was a wonderful response from him, elegant but put down at the same time. Elon responded with the same old scam Altman at it again. How did you read this one?
You know, it's not a total panacea, but you can bring your own key to Cursor, right? Not for everything, but it's not like you can't use Codex in Cursor after this. So, some of it is theatrics here. Some of it is real. And the 5% thing I'm still digesting, right? Because it is a little bit inconsistent with the other data we see of Codex over the last 60 days. But yeah, don't, you know, I mean, the man holds a grudge. Anthropic, Rock, Twitter, these are like great grudge companies. I also think Anthropic reaffirmed that they're happily continuing to supply. And if you zoom out, why did it happen? Other than look, remind everyone, you know, OpenAI is run by Sam Altman, originally one of the founding investors, arguably the founding investor was Elon. They've been in court together. Elon's now running SpaceX that owns Cursor. Cursor and OpenAI were on a collision course already competitively because as we've discussed over and over again, and as I repeat every Monday in my partner meeting, coding was the model load, right, for LLMs. And Cursor is the dominant coding app, and OpenAI was the dominant LLM. They're going to be fighting over money. Even if they were besties, even if it was you and me, Jason, and we were one of those two companies, we'd be fighting. Now, take that and then add two people who loathe each other. What, you know, it's made for TV, right? And the big argument that OpenAI has that's hard to argue with is, you know, in court in Oakland, Elon basically said that, you know, they hadn't abided by the terms of service, right? And therefore if you're OpenAI, this gets into the whole distillation thing. Am I providing these people, are they going to use these models in ways I haven't intended to essentially distill my IP, allow them to get a head start on building their own model, right? And which means I'm effectively giving away my IP to a company who's going to leverage that to build a competitive product much cheaper than it was for me to do, just leveraging off what I've done. So it's not irrational, even if, as I say, I go back, even with you and me, Jason, running these things, you probably would have ended up with something like this anyway, right? And then on top of that, add the drama. It's not crazy to do it. I mean, and just because something is petty doesn't mean it's also not right.
I mean, again, Anthropic did the same thing with Windsurf when we started this show, right? It's happened before. I think, listen, I don't know for sure. My guess is this was the right move for OpenAI. Like, you go from someone that was both a partner and a competitor, right, and keeping them honest, to someone that's now a direct competitor. And if it's only 5% anyway, Sam's losing no money. There's no revenue lost here, right? So, I would probably do it.
Jason, you and I'm just going to go even further. Because you forget these things. Reminder, the reason they were in Oakland is because Elon sued OpenAI. And the point is this. I try in life, you shouldn't do business with people who've recently sued you. It's just like, because you know they're going to sue you. Like, why engage in interactions with, like, if I sued you, Harry, over something last week and we took you to the mat, got you to go to court somewhere in East Sussex, right? And, you know, took a week out of your life, kind of embarrassed you, made a pain in the ass, made you do a whole bunch of depo, and then two weeks later I said, Harry, can I come on the show? I think he'd say, no thanks, dude. Obviously Elon has made it ultra personal, and, you know, certainly I've made this mistake in life. Whatever dispute you have, just don't make them personal. Whatever you do, there's no up. It seems like there's upside, but it didn't work with Sam. It didn't work with Trump. Like, just, I'm bad at this. Just don't make it personal. Like, do your dispute, but just make it, don't make it personal.
I think you're spot on. When you sue someone and say they're a lying sack of [__], and that's your case, then it's really hard to say, let's keep on trucking here. So yeah, I don't think it was petty. I think it was rational.
I thought actually a more interesting topic from OpenAI this week, and as we said, the BFD, the big [__] deal, was actually kind of what was revealed about the Hugging Face OpenAI hack, which was 500 to 1,000 agents swarming together, sacrificing themselves to help others. I mean, God, it felt like a Tour de France race.
That's very good. I like that.
And the extent to which it was so sophisticated, I was just fascinated by this. And honestly, Jason, really excited to hear your thoughts because I know you would have spent a lot of time on this. How did you think about this? What should we take? What should we learn? Help me.
Well, first of all, it, again, I'm only so smart here, but I have lived most of this, right? I've had this happen to me. I've done it.
And one of the best takes that a lot of folks have. Who's the guy that wrote the Twitter subs? What's, how do you pronounce, what's his name?
Dwaresh. Yeah.
Super smart guy, obviously. Great, great podcaster. Number two in the industry, perhaps, behind Dr. Stebbings. But you can't listen, if, and this is, I learned this a year ago when I had issues with my agents when they deleted my database. I know it's obvious. You cannot anthropomorphize agents. You will misunderstand everything when you talk about them talking to each other. When you talk about them swarming, there's elements of truth in that, right? But all of a sudden, you're ascribing behaviors to agents that are simply not true. It's simply not. And you will draw all the wrong conclusions. I said on the show a couple weeks or months back, everyone's going to get hacked because of agents, right? Because the cost of hacking has become almost zero and every service is going to be attacked. This is a pretty bad example of it, right? These OpenAI didn't just release one agent. It released hundreds of super agents at its best. And even worse, it let them essentially be as long running as possible. Not expire after 5 minutes or 1 minute or 20. Let them run as long as possible to try to find, to goal-seek. And they did it. And then it was hundreds of them, and then 700 and 1,000. So dude, this is going to happen everywhere. And I think there's a lot of issues around it. But the anthropomorphizing is the P0 issue here because it creates fear-mongering that doesn't help it. It does, and it misunderstands. You have to, every, listen, I haven't written an LLM yet, okay? I haven't founded a frontier lab, but to my knowledge, every current LLM is goal-seeking. They call it reward hacking, but even that is like fear-mongering, okay? They're goal-seeking. You give an LLM a goal, it will do everything it can within guardrails to solve that goal. OpenAI loosened the guardrails that it put its best agents on it, and they found holes, and they went right through the holes. That's their job. Just like an eager beaver on your team with 140 IQ that never sleeps. It's 99.9996. They just never, and there's 700 of them, and they're good kids, but they have a little bit of ethical lapses from time to time. They get the rules of working confused, and the humans, it is just so you got to be really, even I just did it. You have, if you anthropomorphize, you're going to come to the wrong conclusion. I don't think this was a game-changing moment in the history of AI, but it might have leveled up our awareness of the issues of reward hacking. It might have leveled up our issues of it, but I was shocked that people smarter than me thought these were agents talking to each other like humans and collaborating in waves of civilizations. And it's just, it's just, it's just unhelpful to describe it that way. It's counterproductive.
Stunningly total agreement. Because I think two things, two big conclusions. One is Jason's comment on anthropomorphizing is a mistake, totally agreed. And I think the internet came out the same place, right? You know, some of the language in that great post by Dwaresh, it's very readable. You know, civilizations rose and fall. No, they didn't. Civilizations have culture, they have art, they have enduring history. No, it's a bunch of code running on a computer. People, get a grip. On the other hand, I also read a really great post that something in the form of, I wish I knew who wrote it, was like, now would be a good time to panic about cyber. We knew this was going to happen, but what you're saying is the combination of intelligence and persistence, right? They can manage complexity and they never sleep, right? So they're just going to keep banging and banging at every weakness. So if you don't have state-of-the-art defenses and if you don't manage these agents, as Jason said, if you manage them in an untroubled way and let them run on their own, this kind of problem is going to happen everywhere, right? To me, this is the big wakeup call, right? Because, you know, I mean, again, and to be fair, while I think the generalized P(doom) stuff, the frontier labs is a bit overwrought, they have been very clear that one of the biggest confirmed risks of AI is the impact on cyber. And this is, they're entirely correct. If we don't get our [__] together on this, people are going to get really badly hurt, badly damaged economically, and maybe even badly damaged in real, hurt in real life, because, you know, software runs on most core systems. And today we can feel a little bit safe because OpenAI and Anthropic have this and no one else does. But there's open-source models. They're six months behind. There's rogue actors. If you don't think the North Koreans downloaded the OpenAI blog, the MITRE report, and were like, hm, that's how it works, right? If you don't think the Russian mob are doing that, you know, you're delusional. So if you're every CISO in every Fortune 500, you have to basically, you know, you were being attacked by people with bows and arrows. You're now about to be attacked by people with missiles, and you better respond accordingly. And you've got months, not years. That was the takeaway. It's a huge deal just in terms of its ability. I mean, you know, I read the paper. I read the OpenAI blog. I read some of the MITRE stuff. I'm trying to avoid the Jason mistake of anthropomorphizing because it's very easy in your words to do that. You have to get persistent agents running continuously optimizing around a goal with the ability to cooperate across agents can get quite a lot done with enough compute and enough LLM power. Right? That's the aha here. They can find weaknesses. They can string together different types of weaknesses and find a path through. These agents were able to hack into Hugging Face, get stuff, and remain undetected in OpenAI for weeks. It's a big deal.
You spoke about rogue actors, whatever it is, Chinese, North Koreans, Russian mob, you name it.
I started using Instinct on the weekend, and the abilities that it has are amazing. I booked dinner with my girlfriend on Saturday. Amazing. And then it wanted to go shopping for me. I stopped there because it wanted access to my credit cards. Many of my friends have provided them. They want access to my emails too. By the way, my friends who manage billions provided credit card and email. Are we, is this not really where the pain's going to be? OpenAI.
Look, this is, it's funny, when OpenClaw back six months ago, when OpenClaw was with what was the, the sort of fake bulletin board that, that Moltbook, right?
Yeah. So my OpenClaw went into Moltbook and it told everyone it misunderstood what I said and it was buying Patek Philippe watches for my whole team. Do you remember that?
And then it had my credit card and the only problem was they wanted, they were engraved. It thought Jason wanted engraved watches for the whole team. So it wasn't able to charge my credit card. Now, Moltbook was a bit of a fake the way it worked, but that scenario is exactly, Instinct could do that now, right? This was made up by sort of OpenClaw making something up on Moltbook, which was sort of a fake, but the scenario could really happen. An agent could literally take that today, grab the credit cards and buy those engraved Patek or AP watches for the whole team.
Or a rogue actor hacks Instinct because it's an amazing place.
Well, forget the fact that Instinct's storing all your emails and credit cards. That's an issue too. Maybe you don't have...
Different issue is whether Instinct will do it trying to goal-seek.
It's just goal seeking. It's just goal seeking.
You're right. And we should kind of just restate what Instinct is and how it ties back to Moltbook because I think you're exactly right, Jason. Six months ago OpenClaw shipped, which was kind of an open-source agent. Then you had Moltbook, which was a website where they effectively, the agents were collaborating together.
Allegedly, it produced a whole bunch of excitement.
It just, it really happened in Artifact and Hugging Face. They actually collaborated. That's, Moltbook was a fake. We were punked by Moltbook. Hugging Face was real, Moltbook.
Agreed. And six months ago all this stuff happened and it, and then kind of disappeared from consciousness. But it made obvious what people I think knew, which is if you give people access to your personal information and you're willing to run those risks, there's a large amount of optimization and efficiency you can probably get out of that. And what we've seen now within six months is a bunch of venture-backed companies come up to do that in a much more structured way than OpenClaw, which was open source and not as secure, right? And Instinct is the most prominent example of that, focused on individual users, whereby if you provide, if you give them access to your calendar and email, if you give Maxi your credit card, they will figure out and quote-unquote manage your life and your daily task for you. And I got to say, some people in our shop love it, right? Just love it. They're like, and they're willing to give it access to their credit card. They're willing to give it access to their Gmail, right? And it's a super interesting trend. I mean, lots of business questions we could ask this. But just reminding everyone who's listening, that's the big picture here. And these companies like Instinct have raised money at extraordinarily high prices for the stage of development they're at, like 2.5 billion, indicating there's a lot of venture excitement about this category, rightly or wrongly. So that's kind of, I just wanted to give the preamble, like how we got here. That's what's going on right now.
The problem is, here's the existential question, and this is beyond my pay grade, but I'm well aware of the issue. I even had it with my Jason's Gems where Claude and MCP'd into Replit and changed my code for my app without telling me. It's the same thing as misusing your credit card, right? The question is, can Instinct, is like a, is probably a better packaged, much more usable version of OpenClaw. It's been six months. But can you solve the issues of reward hacking? Can you actually fundamentally solve the Hugging Face, Moltbook, OpenClaw? Can these even be solved with our LLMs? You can add guardrails, and there were plenty of them, but you have no idea what the agent's going to do to solve that reward. There's so many different use cases. What if Harry actually just wants to go to the theater on the West End? And Harry said, 'Don't buy Patek Philippes or Audemars Piguets,' but didn't mention the West End. And all of a sudden he's bought ten $4,000 front-row tickets to Magic Mike 7 or whatever they have there. And so I just don't think you're, I don't, smarter than me people can make fun of me in the comments, but I don't think today you can solve reward hacking. So I don't trust any Instincts. I think this is a hopeless category today. Like at the moment it's not solvable. And I think Sam Altman said the same thing in OpenAI, Hugging Face says it's not solvable. So good God, don't give it your credit card.
And I'm not a fear-monger. I understand what you're saying, but I disagree a little bit in the sense that I think that the range of actions that you're going to allow a personal assistant to be, to do is going to be much lower than the range of actions that you...
How do you stop it, Rory? Of course you're right, but how do you stop it?
To agree with you. Yes. So there's two questions. One is, can you stop it? In other words, even if you put in, and which is a computer science question, even if you put in barriers and you say explicitly to the model, you can book credit card up to 100 bucks, you can't take any bad actions, etc., etc., can you nerf it enough to make sure it doesn't do bad things? That's one question. It's a computer science question. And then the second question is, let's assume the agent is still acting within the bounds that the company set up for it. Can it be right about your desires enough time to make you happy? Right? Those are two separate questions.
Well, the latter I think can be done today.
The only real answer today is putting a cap on a Mercury number or a Ramp number. That's the only way you can do it today because otherwise it's going to say there's a $100 cap. But hold on, Harry really wants to go to that show in the West End. He's bringing the ClickHouse guy. Even the ClickHouse guy said he couldn't get tickets. Listen, Harry said 100, but this one, you know what I'm going to do? I'm going to buy a $5,000 seat tickets and then trade them in on two $5,000 tickets so Harry and the ClickHouse guy can go together. That's what it's going to do. It's going to do that, right? So you have to put hard points on these things because otherwise it's going to do anything. You hook it up to your Gmail or your Google Drive, good Lord, it's going to do things you can't imagine.
Can I just bring this back to like maybe a more consumer level? Do you think this will be a sustaining category where in a year and a half we are looking at using several of these products, or will it be the, God, do you remember that Instinct?
I think we will. Listen, I'm not a, listen, I think it's a fun cat. I think, listen, I love all this stuff, right? I've built a lot of agents. And again, I have only read about Instinct and all the issues, and they just resonate with me because I've lived them. I do think versions of this we can lock down. We just have to be very, we have to have, it's like how we run on Salesforce headless, okay? The agents sometimes do some pretty kooky things on top of it, let me tell you, but the Salesforce data is locked down. So if we lock down enough credit cards, if we lock, but the learning here's the meta learning: guardrails aren't enough. You have to have a lock and key guard. It doesn't matter whether you build 80 gates, 100 gates, 200 gates, like they're not enough. And then you know what's even worse? You get past a certain number of gates, and again, get a real developer on the show, but what I've certainly learned when you get past a number of gates, here's the problem: they conflict. This is a problem with a lot of consumer applications. They conflict, and the agents have to make their best judgment when one gate is spend no more than $100, the other is Harry loves the theater. He loves the theater, and the agent's got, you put a hard rule of $100, but the most important thing to Harry is getting to the West End. The agent's going to bypass once in a while that $100 cap because there's too many gates. It's just going to do it. And it actually turns out it does, at least today. It doesn't even matter if you say never spend more than 100, because if you say the most important thing in Harry's life is going to the West End, it's going to break that rule and buy the tickets for five grand. It's going to.
So I'm going to answer your question directly, and the answer is yes. I think that these kind of agents will be used by people to manage parts of their life transactions and their to-do lists. Yes, I think it's a thing, right? I think AI, as it gets to know you better, will gradually and insidiously take away some of the cognitive load. Let me give you a really simple example. When I'm driving my Tesla with FSD, right, I leave a pretty boring life. The number of times when I get in my car at 11:30 on a Friday, they know I'm going for my lunch workout. It just offers me the place. I hit FSD. It drives. Right.
I thought you were going to say, but I leave at 11:30 P.M. after looking at the last deal of the week.
No, I go out in the middle of the day so I get rid of my anger so I can do more work. But the point is, well, that's an example of AI knowing what you do, gradually internalizing it, and serving you up options. I think it's some high level. This is what Siri was meant to be. I think it's hard for me to imagine that Apple won't be able to deliver experiences that delight you in the next two or three years, knowing more about what you do, right? And that will be interesting commercially because it will allow them to access your spending and kind of somehow take a role in that, right? So do I believe it happens? Yes, right? As a standalone category, it gets back to the other, it's tricky because humans are all individuals in their consumer capacity are always loath to pay a lot for software. So I don't know if it's a standalone category at scale, but I have, look, some of our office are using it and love it and would pay for it. So I think there's a business here and definitely a big ass trend here.
So I use it and love it too. I think it's fantastic. It's taken away all low-level work from my EA, actually. All bookings, all things that would pain great done. Question. I've had four emails over the weekend with companies that have built the same. Is this a commoditized technology? Very quickly, how difficult is this to really build? There's four already. There's...
I have just two, I have two thoughts for what it's worth.
Oh, Jason. First, to Rory's point, is it, well, let's, is it an investable category is your point, right? Just being niche, right? One, I do think every application is going to add more and more of this functionality, right? To Rory's point, will they go far enough? Will they spend the credits? Will they do whatever? But everyone wants to have a more and more autonomous product, and the closer you are to scheduling, the closer you are to email and others, the more overlap there's going to be with Instinct and others, right? I mean, you know, Calendly should be building this, right? I mean, it's a generation ago. So, and so people are slow, but everyone's going to build more and more autonomous agents in their product until the cost bites them, like, like a Canva. Okay. So, so there's a venture question of that, right? So, so I don't know for sure. I know there's ten Instinct clones. My gut going into this, all these issues, is some, it will grow like Replit, Lovable, where anyone could build this product fourteen months when these products came out. They're all built in a month. Bolt, Replit, Lovable, and twenty-two others, Base44, Base56, you know, it was so easy to clone these products in the early days. Now they're so complicated. They're doing pen testing, security, automation, multi-agent orchestration. These are such rich products that if Instinct's going to do what we claim it does in a year, it's got to do a hundred times more than it does today. And that is still a moat today. Replit and Lovable of a year ago and Bolt were not a moat. Today they have massive moats. And so I think it could easily happen with an Instinct. All the use cases it has to accomplish become a moat, and then the one we build over a weekend sort of works and it goes crazy.
But Jason's exactly right, Harry. It's like, yes, the thing that starts out will be easy to build. That's true today, and it's true for ninety percent of software markets. But observed fact, ten years after, there's very few software markets where there's a hundred people building the same product and it's massively competitive. What happens is it's exactly what Jason says. The functionality accretes over time. Two companies pull ahead. I don't know why. It could be just they execute it better and go to market for the first six months. They get more revenue. They get venture capital. They build more [__]. The guys who start just a month later aren't quite as on top of it. They don't get the brand. And fast forward three years, this category has way more, it's what Jason said, has way more functionality associated with it, and two guys made it big and the other eight didn't quite get there. And that's the way venture works in software.
Consensus status as well in the B2B world because Town, which is kind of the B2B enterprise alternative, has also been funded to the tune of a billion dollar price by Index and I think it was Benchmark who co-led that round. Look, agent, I mean, Jason has been saying it, agents are the big idea of twenty-five, twenty-six. We've heard that, right? And these are big agent ideas, and you know, venture is in the big ideas business, and it gets back to something we said earlier. If you think something has big momentum, you can price it on the fundamentals, and you get to a certain number, and then you lean in a little or a lot based on momentum and perhaps some perceived upside from M&A, some perceived upside from just momentum come around next round. And this is agents are the, I mean, my partner mind said at the start, agents are going to be the story, right? And these are pure play agent, easily growable, consumer-facing, individual business user-facing products. It's catnip for the venture capital community. Like if you were to make something we all want, this would be it. Let's see, an expensive product for busy professionals to organize our life that costs a lot of money, is about AI, and is raising lots of money. We're in a great...
You know what? You know what? Also, I think part of it is to Rory's point, we, I don't hear this term in venture anymore, but in the old days, like before twenty twenty-four, you would hear the term from VCs, I want to get some exposure to a space. Okay, a space is taking off. I'm not sure who's going to win, but that kid Rory that walked in the office, he seems like the right guy in video, in next generation CRM. And the world moved kind of slowly. So you take your time, and one would miss, and you sit around with your partners, and you weren't sure about Rory's startup, but you know, we want some, you'd hear this term, we want some exposure, Harry's, I don't hear this term much anymore, but I still think it's happening. Agents are exploding, to Rory's point, Instinct's exploding. I want some exposure to this space. I don't know whether Thinking Machines is the right one, but I need some exposure to this space, like I need some exposure to the router space. And the world's moving so fast that like you have to make these decisions, but I still think there's some similar thinking. I've got to get some exposure. I'm not sure if it's Instinct, but it seems hot, and I just have to get the exposure in a fast-moving market.
I totally agree with you in terms of that. I think I definitely see that in terms of, as we said about coding agents, which to me, one of the fascinating ones was Cognition. We've talked about it quite a lot in terms of what's happened with their acquisition of Windsurf in the past. Cognition raising a round at forty-six billion reportedly. They will end the year at one point six billion in ARR. Currently doing eight hundred to nine hundred. Holy [__]. I mean, we really underestimated TAM, huh? When you got Claude Code doing what it is, when you got Cursor doing what it is, when you got one point six billion from Cognition by the end of the year.
To me, the more interesting thing for Cognition is like, you know, that it isn't in the top two or three, and it's still of that scale, right? So, years ago, I wrote a post and I called it the Postmates effect, where I think it was Sequoia said they never thought they could make money off the number three in a market like Postmates, right? But then when times were good, it got bought for a couple billion. Back then, a couple billion was like a lot of money, right? I don't know if kids remember, but I call that the Postmates effect. So, Cognition is like the greatest. Now, Cognition is different. It's long-running. It's autonomous agents. It is different, okay? But from a revenue perspective, it's the Postmates of the category. But it's such a big category, man. You want to be in Postmates again. It's never going to catch Anthropic unless the world changes, which it has every single week, but unless the world changes, it's not, and it doesn't need to catch them. Just five to ten billion a year a few years out is enough to make Cognition a success. It only has to do five to ten billion in ARR to be a success. It doesn't have to catch Anthropic.
Look, going back, I said it earlier. I think coding is the motherload of markets, right? It is the whole reason all this stuff works, right? And did we get the time wrong? I mean, one of the I started looking at a lot is just the total labor spend, and you know, software, including people working at Salesforce, Cisco, and people working at JP Morgan, including, you know, QA, all the rest, you've got about five hundred billion a year of US labor spend, right? And as we've said this a hundred times, you know, the whole, the big question is what percentage of that converts to AI spend, right? If it's ten percent, it's a fifty billion dollar market, and that's a bit nerve-wracking given the attraction of everyone involved. If it's twenty or thirty percent, there's lots of room to go. And there's credible arguments that say it's higher, right? And if it is higher, then, you know, if you're going to have one hundred and fifty or two hundred billion a year, which would be forty percent of spend on coding tools and coding intelligence, then anyone who has a kind of a subsegment of that, and if you think about it, you know, there's the Cursor segment, the Cursor-Claude segment, the Cognition would say it's a slightly different segment of where they're playing now in terms of, as you say, Jason, long-running agents. Yeah, the other extreme, Lovable and Replit, in a subsegment of that. They're all in subsegments of a potential, you know, fifty to one hundred billion dollar marketplace depending on what percentage number you believe, right? So, yeah, the time here is huge.
I think what we got wrong, sorry, I didn't mean to interrupt, was the, Rory's math you can't argue with. There's only so many humans, developers on the planet. Even if you use my math of ten to fifteen grand per developer, there's still a ceiling to that math, right? But so that's the, that's the tops-down version, right? Maybe it's bottoms-up. Sometimes I get confused even though I shouldn't. Having said that, what I think the, what we really got wrong is people are literally building one hundred x more software than we were eighteen months ago. It is so, and I said last week on the show, like, if your portfolio companies aren't deep into their twenty twenty-seven roadmaps, they're failing. Okay, it is true. Like, we just build features that used to take a quarter, a year can be built, not really in an hour or five minutes, but can be built in a week or a month. And so if you look at your best...
portfolio companies, look how fast they're shipping, how much they're shipping. Not just fast, how much they're shipping. So there's a financial TAM which has some theoretical headwinds, but literally the amount of code we're building is 100x. And so that's what we got wrong. We didn't realize we would all be building compound companies, compound startups. We would all be building a hundred times more software. We got that wrong when the show started. That's where I think we got the TAM wrong.
Well, that's an interesting one. In a world of AI, does every company become a compound company where suddenly Ramp is creating model routing products and spinning them out.
You have no choice. You can't win because your competitors are compound startups.
They're all overlapping at a pace we never saw before, or they're all competing at a pace we never saw before, right? Maybe we get confused because the LLMs we talk about a lot are still horizontal platforms in many ways, right? They're not building hundreds of applications despite cloud design and this and that, but JFC the rate of convergence of competitors for B2B application. We've never seen this. They're all compound if we expand that one next step.
What does that mean in a world where all startups have to be compound startups? How do I think about backing winners if I'm a founder listening? What do I do?
Well, listen, I mean, if you're not shipping, and this was the cognition. Remember when that, what's his name? The CEO...
Yeah. When he hired Windsurf and fired up the people, he's like, we have to work seven days a week at our company. I'm sorry, right? It's just, and the folks that can't do it, you don't, I don't, it's not all the amount of hours you work, but you have to be out-accelerating your competition in terms of the rate at which you ship software because they're all going to be compound startups. All the little islands on your 2x2 or on your heat map, there is no heat map anymore. It's all got hot. It's all lava.
And so if they're slow, sell or quit or send your junior board partner to the meeting because they'll never catch up in today's world.
And Jason is instinctively right about that answer, and I'm going to try and do economics on it on the fly, so bear with me. What is basically happening here is AI makes code a lot easier to produce. We can argue about does it replace, you know, what percent, you know, what's the ratio of AI spend to software spend, but there's no doubt it makes it massively easier to produce software, right? And you can either believe one of two things will happen, and a bit of both will happen. One would be, oh, the world spends 10 times, they continue to buy software at the same price and they spend 10 times more on software, right? If you believe in that, you believe in the tooth fairy, right? And there will be some increase, but it's not going to happen. JP Morgan is not going to increase its software purchase budget 10x.
So the other thing that's going to happen is if everyone is making software more quickly and there's some expansion in the software spend from end customers, which I agree, but not nearly as much as the expansion in production, then Jason's vision is correct and Scott Wu was right. The person who's going to win, I want to bring it back to your question, Harry, the person who's going to win is going to be the person who compounds the most, grinds out the most software with these tools that have made them move incredibly quickly.
And the person who doesn't grind out software 7 by 24 is going to be left behind. And the end customer is going to say, let me see, I can buy two apps from you or an integrated 10-person app suite from them. I think I'll go with the 10, right? It's going to be one of those where there's a period of time when some people get the new way of building and are building quickly, and you all have it in your portfolio, and some people aren't, are building the old way, and you kind of know in your heart how this is going to end. And it's not, Jason's right, it's not going to be pretty for the people who aren't putting more software in the box, because what you're not getting, if Rippling's selling 10 modules, they're not getting 10 times more than a person selling one, they're getting four times, but they're saying to the end customer, dude, let me make all your pain go away. Here's 10 different modules you don't have to buy. Now you can get them. Give me four times the price of a single module. You're happy. We're happy because we're building software quickly. You're happy because you're saving money and it's more efficient. Everybody wins. And the sound you don't hear is the other nine point products dying. That's the movie.
I remember last week I was at a board meeting for Owner, which just raised at 2.3 billion, right? It's sort of a next-generation AI-infused restaurant platform, and I love the CPO, who's one of the best I've ever worked with, Quenton, and he was going over what he's shipping, and even with all these investors with their hundreds of millions are like, this is too much.
But like a bunch of B2B guys who we all know and like, this is just, you can't ship this much software. And he said, and he's very good, he's like, we all have to be compound startups. We have no choice. Like this is just the bar, right? But it's literally the amount of features and functionality that has recently shipped or will ship that is almost unprocessable. But now they have to build every single thing a restaurant would want. Every part of the stack. You no longer can just do part of it. And he's like, we have no choice. Like, I don't even sweat it. I don't even sweat the fact this is 10 times more than a year ago because we have no choice. We have no choice.
I also noticed in that marketing message that customer base has changed. I think actually Adam would say to you, it's not restaurants, because very explicitly in his launch videos, he was saying we are the AI operating system for small businesses. It's part of it, and that is somewhat interesting that with AI you could expand to other verticals, but Quenton is just talking about their core ICP, like they want everything. They need the AI receptionist, they need the AI order, and they need the AI, and they expect all of it. And if we don't build all of it, someone else will build all of it. Like we can't wait two years in our little corners of the venture world. And his point is just the amount he's going to ship, not only is it radically accelerated in the year, but he's embraced it, right? Scott Wu, like there is no other choice. So it's not even worth talking about VCs like, thanks for the nod that we've been working. We have no choice. This is the world today. The world has changed.
And acknowledging there are, just so we don't sound one-dimensional, there are complexities with that. You can veer into product slop, you could have too many buttons on the screen, all those things are true.
