The Exchanges, every show

Q If I'm the founder of a company, an early-stage company, do I just accept that I'm gonna have B-tier or C-tier AI talent? And I don't mean that denigratively or rudely or horribly, but they're an anthropic and open AI. I mean, Google can't freaking keep.

A I think it's the wrong, um, framing, because if you think about it, I mean, look, when I was, you know, when you were building a software company in the age of the PC, You had four-tier chip talent because you weren't building a chip, right? The point is, if you're an AI company and you feel the need to build a frontier model, then yes, you've put yourself in direct competition with someone, and if you don't have the good people, you're toast. So, what you got to do is make the model a compliment and have A-tier talent at UI, A-tier talent at, you know, AI implementation, A-tier talent at the things that you have your competitive advantage in.

AI assessment note: “I think it's the wrong framing... make the model a compliment”

Q Rory, should they be valued like traditional financial services businesses, or should they be valued like a new technology-first provider?

A I always think that's kind of a bullshit question, because in the end, everything should be valued on a basis of risk-adjusted free cash flows. So let's start with that. But what you're really saying is what's the best, in the absence of free cash flows, what you're really saying is what's the best rule of thumb to value those things? And I think the truth is they have the margin profile and core dynamics of a financial services company, but they have the growth rate of a software company. So you, you have to adjust and come somewhere in the middle with the expectation that this is the key sentence. Once the growth rate slows, they will be valued. Just like if they're growing the same as Amex, they will be valued the same as Amex. The growth is what's saving them.

AI assessment note: “come somewhere in the middle with the expectation that this is the key sentence”

Q is if you look at companies like Pinecone, They went up and they went down and fucked very, very quickly with changes in technology cycles. Squarespace, Wix, any of the website builders, I'm not picking on them or being horrible, but your lovables and your bolts are absolutely killing them on the consumer and, you know, freemium modes. Are we going to see the acceleration in that terminal decay rate?

A Short answer, of course. I mean, that's almost a given. I mean, look, one, all technology companies have obsolescence written into them from day one. You know, it's, it's sad, but it's, you know, true. I mean, anything before Microsoft and Apple at this stage isn't here and Coca-Cola has been cranking for a hundred years, right? So obsolescence is inevitable in general. On top of that, periods of acute technical disruption are going to increase the amount of, you know, technical obsolescence that companies face and then overlay that with, as we've been discussing longer hauling periods. I mean, I think I said this last week, but my number one fear is That the technology life cycle to obsolescence is now shorter than the holding period of privately held software companies, which means every company at least one time before it gets to go public will have an existential reinvent itself second product crisis. And if they fail that.

AI assessment note: “Short answer, of course. I mean, that's almost a given.”

Q You guys done any deals in the last seven days?

A Not in the last seven days, no sir. No. I do, I do want to come back to the one other thing that really struck me as interesting. You put them in there separately, right? But I've been thinking about this a lot. You had the TSMC's announcement, ASML announcement, and I was thinking, oddly enough, about different kinds of trusted supply chains. And I'm just going to contrast two because it's quite funny, right? You have the NVIDIA relationship with TSMC, which famously, they don't even have a written contract. They've dealt with each other for 30 years. NVIDIA is now TSMC's largest customer, and, you know, it's, you know, there's tensions because they're pushing TSMC to invest more, but, you know, they're managing that. And then the same kind of relation, TSMC and ASML. ASML makes the machine that enables TSMC, and TSMC makes the wafers that makes NVIDIA. And no one in that entire supply chain has ruthlessly gouged each other. ASML has raised prices gently. TSMC has raised prices gently. They're pushing people for forward commits, and it's a real, hey, we know we're going to be dealing with each other for, you know, 10, 20 more years. Trusted relationships. How do we cooperate for the last? And there's tensions, but it's not all that crazy. And then you just compare and contrast that to the adjacent market for DRAM. There's three suppliers in there, right? You've obviously got t…

AI assessment note: “Not in the last seven days, no sir. No.”

Q Roy, would you add it to your public book at 300?

A Probably not. I go back to the base rate. You know, the base rate return on IPOs, not from the day of the IP, not from the day of pricing, but from the first day's closing price, which is typically way above it, the base rate return on that is negative. One, six months, 12 months, one year. Yeah, two years, right? In other words, it cost a thousand or so of them. If you buy on the pop, you know, you tend to be on a happy camper, right? Is it a company at the right price you believe can be a long-term, enduring company? Absolutely yes. It's got technological differentiation like no one else, right? So just blindly buying the day every other retail idiot on the planet is buying is probably not the best way to make money, just statistically. And base rates matter.

AI assessment note: “Probably not. I go back to the base rate.”

Q I mean, and I'm getting cash or I'm getting space?

A Well, you're almost certainly getting stock. I, again, I'm, I'm going to opine on that without looking because the IPO is 75 bill, allegedly. Again, we haven't seen, again, we haven't seen a printed S-one yet that's been publicly unveiled, but the IP is, it's highly unlikely that you're going to write, raise seventy-five billion, write a check for 60 bill. So almost certainly you're getting stock. So then the second order questions are, is it registered? If you're getting that stock in the, uh, restricted period, is that going to be restricted too? Because the whole, the dynamics of the SpaceX lockup, given the size of the round, are going to be one of the most interesting parts of the whole transaction. There's talk of, as you say, employees getting out early. There's also, to be fair, talk about some shareholders being held in for much longer because the combination of a two, I mean, it's going to make a, Difficult problem around float management, slightly more complex, because a seventy five billion raise on a two trillion dollar valuation is a teeny tiny float, right? And, you know, going from there to the other 97% of the company being freely tradable in six months just doesn't work. So I'm sure there's going to be a convoluted float management thing, and then this stock will probably go into that. So I wouldn't be surprised to discover that as investors in Um, cursor, con…

AI assessment note: “Well, you're almost certainly getting stock.”

Q growth has been amazing. Um, Pharmaceuticals ad spend in the US on media is twenty two billion a year. If you think about the transition of that to their business model and assuming a, a reasonable take, you can see them being a four to five billion dollar revenue business in that alone. Doesn't feel crazy, but then in other aspects it does. How did you guys read this one?

A I think it's, start, it's a great company. It's a perfect use case for AI, and it's one of the use cases where, you know, the general models are good, but the combination of Um, specific, um, relationships with Journal of New England Medicine and all that, plus, um, restricting access only to medical professions, plus HIPAA compliance means you've got this really nice product to allow doctors to do decision support, which is, you know, go and check online, watch the recommended treatments for some obscure disease I haven't seen, right? And then the obvious thing you do with that is you sell them ads, right? And the obvious people to advertise to those doctors Are the drug companies, because they want to sell to the doctors, right? So it's a perfect business, and they've escalated to, I believe, a hundred and fifty million in revenue, right? So I was actually impressed that you led with the market size, because the only question, so the things that are clear here is they're the winner in the space, right? Doximity is the old pre-Gen AI competitor, but in terms of thinking about doctor media mindshare for doctor-like things, Doximity helps you a lot with, you know, Thinking about salary, thinking about job, but I have a medical question to which I want a highly technical medical answer. They appear to have commanding market share, so you've won that business. So the only question…

AI assessment note: “the only question is how big is the market? And you're right.”

Q I like it as an addition. I can totally see that. Vlad is incredible. Um, okay, we're going to go to number two. Fund of the year. Which fund deserves the title as fund of the year?

A You know, again, with the continual caveat that I despise this process, but I'm going to answer specifically on the performance for the year. I think it's index. Competent execution across multiple different exits, because exits are the coin of the realm. The only fair way to measure things is, did you end up owning a lot, have a big ownership in big exits? And then you measure that for the year in which it happens. You don't try and account for other years, and then it's at least objective. And they had Wiz, which for the record hasn't closed this year, but was announced in March of this year, where they were an early investor. They had Figma, where they were the seed investor, and it's not an exit, but Revolut, where they just raised at seventy-five billion, and they have an early position in. And, you know, also done by a number of different partners. You look at all those things. I just love, that's kind of, there's a lot of people talking about new venture models, and we'll talk about that in a second. Index just took that same old model, which I like, because I'm a boring kind of guy, and just executed to perfection. So I, I, I give him the nod for 20, 25.

AI assessment note: “I think it's index. Competent execution across multiple different exits”

Q Alrighty, back to the company side. What is the breakout company of twenty-twenty-five? Whose year was made in twenty-twenty-five?

A There's lots, first of all. I mean, and I'm interpreting, just to be clear, because you may, we're basically saying, at the start of 25, no one had barely heard of these things, and oh my god, in 25, they kind of, so it's not the open AI, the entropic, it's the, yeah, so I think there's a ton. It's really interesting. You know, just to pick a few, I think open evidence, In the kind of AI medical space really nailed it. I mean, what they do is one of the things we saw stepping back when we looked at a lot of these academic research, um, academic research search engines is if you look at people looking up papers, about 60% of that traffic was doctors. Doctors are the biggest single demographic of people who want to look up a recent academic paper and find out a result. And I remember we were looking at some of the broad horizontal science search engines and thinking, hmm, that's interesting. OpenA, open evidence didn't just think it was interesting. Very, very wisely focused on that space and said, look, doctors, even ordinary GPs, sometimes they have a patient with special, an obscure question. They want to know what the laser's research is on XYZ. This perfectly lends itself well to an LLM-based application. They built that product. Totally suited for the target audience, and they went from nothing to 500,000 doctors out of about a million in the U.S. in the space of one year. …

AI assessment note: “I think open evidence, In the kind of AI medical space really nailed it.”

Q I found this next one a hard one. What was the biggest surprise of 2025? What did you look at the news and go, oh my gosh, I can't believe that.

A That's easy. I didn't. The talent wars. The fact that everyone would tear up everything to get shit. In retrospect, it's obvious, and I'll talk about that in a second, but wow, Meta's willing to just give a dude a hundred million bucks to show up. Met is willing to buy a company for fourteen billion dollars and say, give me the people, you know, you guys keep the company, see if I care, and you guys can give that fourteen billion back to yourselves and then let the company be a husk. See if I care again, right? Um, you know, Nat and Daniel have a venture fund. No, um, we should just buy that too, right? You know, and it's, they were the biggest, but they weren't the only person doing this. I mean, you had the whole Google windsurf saga, right? That, The entire convention on why people buy companies, how other employees get treated, how much a human can be paid to do a job got thrown out of the window in the space of six months. Blew me away, right? And yeah, as I say, in retrospect, I mean, the logic that someone has unveiled is pretty compelling, which is if you're spending seventy three billion dollars on CapEx, spending five billion dollars to make sure that the people using the CapEx know what they're doing probably makes sense. So you look back and you go, it shouldn't have been surprising, but at the time you're like, wow. I mean, remember it was incredulity about the mon…

AI assessment note: “The talent wars. The fact that everyone would tear up everything to get shit.”

Q baby, and I wanna start with Oracle and OpenAI. Oracle touching a trillion dollars. Um, Rory, you're always quite good at setting the scene, and I actually got given some good feedback the other day, which is we need to set the scene for the stories. People love the analysis, but they'd like to know actually what happened. Can you just help set the scene of Oracle, OpenAI, what happened?

A I think it is good feedback. I often, we often do forget we dive right in. So step back. What happened is Oracle announced, I want to say that whatever was Q two results, they have an off year, they have an off cycle year, so it might've been Q one. And for what it's worth, they were actually a little light on the quarter's results, but they announced a future RPO revenue performance obligation of north of three hundred billion dollars. In other words, they said in the future, we've got orders as of now that we have to deliver in the future for well north of three hundred billion dollars. Most, and they didn't say this, but you figured out pretty, most of which is a big open AI order for around three hundred billion of future cloud compute for their AI platform. And the stock exploded. The stock went up 36, 38%. I think briefly making Larry Ellison the richest man in the world. Yay, Larry. And, um, And Oracle touched a trillion dollars. So a total, I mean, it's unparalleled for a company, you know, a top 10 company to jump by 38% in one quarter. So a huge jump. Since then, some skeptical commentary, but that's what happened.

AI assessment note: “What happened is Oracle announced... a future RPO revenue performance obligation of north of three hundred billion”

Q fields, the replets, the thing that could also kill them is actually the fact that Wix has base 44, which is actually doing incredibly well. The fact that Adobe or Canva could do what Higgs field does, and with the existing distribution they have, do it. Going to the distribution and going to the incumbent versus startup. How are we thinking about incumbent versus startup? And the core crux there.

A The market would say that the Wix acquisition has worked very well and, you know, it's probably going to be fifty million in ARR by the end of the year. So that's a win. The market would also say, you're just based on the thing that Adobe and, you know, let's be on Salesforce aren't seeing that kind of, have tried to announce AI products and have gotten some traction, but aren't able to access that explosive growth. And I think market reaction, the Adobe stock price kind of reinforces that. So it's not a kind of simple binary answer. It's some folks have pulled it off and probably in fact, as I think of it in real time, because it's easier to make it, it's easier for a hundred million to make a difference at Wix. Then it is to make a difference that Adobe or Salesforce, twenty-three billion in the case of Adobe, and forty-something in the case of Salesforce, where it's, it's hard for those incumbents to move the needle significantly, which is why the stock prices have been down, because you're not getting the AI explosion. Yeah.

AI assessment note: “it's easier for a hundred million to make a difference at Wix. Then it is... at Adobe”

Q All right. We're gonna, we're gonna, we're gonna, we're gonna do a quick fire. Um, okay. So what price will Figma be at in 365 days? It's at 52 today, which is a twenty five billion dollar market cap. Give me some numbers, Tim.

A I'm gonna give you a number that's gonna say it's about mid forties, and I'll tell you why I give it. It'll prove all these silly people wrong. I'm going to give credit to the bankers. They priced it at 35. They get a roughly 10, 15% pop and one year's compounding. The price that it should be, if the bankers were roughly correct, would be around the mid-forties. So, I'm just going to assume that they're more correct than all the idiots who priced at a 110 and moved around and talked about it. So, I hope it ends up at that price and it'll allow the bankers to say, you told you we got it right. It just looked wrong for a while. I mean, do I have an actual opinion? Hell, I don't know. But that would be a pleasing outcome.

AI assessment note: “I'm gonna give you a number that's gonna say it's about mid forties”

Q Going from employee payouts to one we didn't expect, author payouts. Anthropic had paid out 1.5 billion dollars to authors. Is this a one-off prayer for forgiveness? Is this a continuation or a new business model? How did we analyze this?

A Yeah, easy and super clear. Doesn't, doesn't, if you read the judgment, it's really interesting. I actually, these judgments are interesting because the base, the judge said the following, if you bought the damn book once and you used it to train your model and provided you pay the 15 bucks per book, that's totally legal. If, however, you downloaded this corpus of books, didn't pay anything, and use it to train a model, I'm going to fine you 3000 dollars per book, which is how the fine was wise. It was 500,000 books at 3000 dollars a book. So it's actually a fair amount of clarity here. What it says is if you want to train on 400,000 books to, to build your LLM, what you actually have to do is buy the book, slice it off, OCR, the whole damn book, and you can legally use that. Right. But if you don't do that, and you just don't pay the 15 bucks per book, you get fined three grand. So I thought it was actually a fairly coherent legal opinion that said, this is where, this is the cutoff between fair use and non-fair use. And I think Antropic just made the mistake way back when of not doing that and got caught for it, but cheap at the cost. Probably like, yeah, we should have done it. It's not a like, it's not a crime. It's like, We should have done this. We're going to pay our three grand per buck. Wish we paid 15 bucks a buck. Life goes on.

AI assessment note: “Antropic just made the mistake way back when of not doing that and got caught”

Q Do you think it will work out? When you look at some of them, we've mentioned like the Databricks of the world, and how much that will return. It's astonishing.

A The truth is this, probably, if it does or it doesn't work out, it won't be because of their seed program. And that's the big aha. Like, the seed program could get lost in the noise. It will work out If by virtue of the CDNA program, they get the small number Of absolute outliers. And they stated this right back in 2009. So give them credit for wild, wild consistency. As long as they get those few number of companies that are absolutely outrageous, upside performers, and they stuff a billion dollars into them. Like they did a data bricks and they do it at the right price. It'll work out fine. Everything else is a loss leader. I mean, the seed program, it's basically like cheap milk in the supermarket. It brings in the crowds. Right, you know, it's the last leader.

AI assessment note: “It'll work out fine. Everything else is a loss leader.”

Q come to the table. The other thing that was kind of a little bit less typical about it was how much shares they indicated they were going to be selling on the sell side, both from the founder and from the venture capitalists. Dylan cashing out 60 to a hundred million. Fine. But it's like double the normal allocation that's sold. Is that relevant? How do you think about that?

A I don't think it matters that much. I mean, I think, look, it's one thing when we used to have these IPOs of, you know, a company doing seventy million dollars. It's been around six or seven years, barely probably. These guys have done their time. They nearly got twenty billion bucks. I'm sure they all made mental models on twenty billion dollars. Now you're coming to the IPO, you know, two or three years later. It's a relatively small IPO of primary shares. I think they're only raising around six percent primary share dilution. Some part of what might have happened here is they don't need a ton. I mean, look, one of the problems on the IPOs, you have to come up with a use of proceeds. They're profitable. They have a lot of cash. There's not a lot of obvious things to do with the money. So my guess is they were being fairly restrictive on primary shares. And then the bankers whine and say, you need a bigger float. And part of what happened here is everyone said, we'll do some secondary.

AI assessment note: “I don't think it matters that much.”

Q So you think it's pent up demand that's just latching onto available supply, not necessarily the quality of companies coming out?

A No, I think they're great. No, I did not say that at all. No, I think, look, how do windows open? Traditionally, good companies go out at attractive prices. That's how, that's how you, if you've had the window shut, You know, the only way you open it is with good quality assets and at prices where it's attractive. And the way it manifests is unfortunately you get these pops where effectively The company going public has effectively bribed the buying public to like IPOs again because investors were all Philly Pavlovian. If you were a IPO buyer in 2022, your searing memory is I bought them all in 21. I lost my, I lost 30%, 40% on average, 90% in some cases. I'm never doing that again. Right? So you take a long time to come back. Now in the last month, your narrative has changed. It's Oh my God, I piled into the last five IPOs. I'm up 70% on average. I'm up 250% in the case of Chime, in the case of Circle. This has been a huge boost to my fund performance. I got to do me some more. So the pop helps the win. It's not ideal. There should be a better way, but there's clearly positive momentum now from good companies and there's going to be more of an appetite.

AI assessment note: “No, I think they're great. No, I did not say that at all.”

Q You know what I love? I love the American banking sector. Years of trying and you guys create Fuck all enterprise value. We have a Russian in London who creates a hundred billion dollar behemoth that makes chime and everyone else look like kids play.

A I want to go back on that. I didn't realize you're going off on that tangent. That partly, I'll tell you very directly, that's partly because you have a mediocre incumbent banking sector there, and thus there's a huge amount of surplus value to be extracted. It's just like, what's the guys in Brazil? I'm sorry, Nubank, right? The crappier the existing banks, the bigger the opportunity for fintech. And broadly speaking, you know, in the United States, with a few exceptions, Some of these existing guys are pretty efficient. So you're right. There wasn't the same idiocy to kind of attack there. So the market cap, so the opportunity for fintech, it was a tougher business to get to scale. Now you had the countervailing fact that you have kind of the visa and interchange revenues, which are pretty exciting in the US. So that's been an advantage to the US over Europe where they're more capped. But yeah, I don't think attributing success In fintech in London to the greater entrepreneurial qualities of Russians living in the UK is perhaps the most logical analysis, Harry.

AI assessment note: “The crappier the existing banks, the bigger the opportunity for fintech.”

Q On the flip side though, they had Hinge, five extra money, returned four hundred million. Great win. It is in a 6.2 billion dollar fund. My take on that was like, is that just the nature of the game they're playing? They're never going to have fun returners by nature of it. Or is that just actually a pretty immaterial exit, as awful as it sounds?

A I think it's the former. I think they're playing it. And, you know, I think that we have to let go of this vision that venture has to involve fun returners. I think as the fund size gets larger and The kind of bets that investors make are you have more bets and a more later stage. I think the probability of any one deal return, quote unquote, return in the fund goes way down. It's not impossible at any level, but at seed, it's almost a necessity. And I defer to you guys on that. You have a better feel for it. But, you know, for example, our model is we figure we do 20 deals. You know, 30% are bad, 50% are solid and return one to five X, and 20%, four companies return more than five X with an average of 10, which by definition means about a .5 X the fund. We've had fund returns, but our mental model is we have to get four of them right, each of them good enough to return half the fund, and therefore you get two X from your big winners, another one and a half X from your base hits, and there you are, right? That's a model at our stage. Now you go to a 6,000,000,008 billion dollar fund, you know, 10 X our size, you're probably looking at some kind of model like that. If you're doing roughly equal size bets, it just gets harder and harder to assume that a single deal is going to transform the whole fund. So you're left with this dynamic of having to, as you say, make five hundred m…

AI assessment note: “I think it's the former. I think they're playing it.”

Q In terms of then the kind of LP pressure on these valuations, is there a, is there a kind of pressure or a desire from their behalf in terms of investments from you?

A No, I mean, I think most LPs understand that pressuring their GPs to make investments when they don't want to is a very good way to lose a lot of money. So we really don't get that kind of In fact, you usually get the alternative pressure, which is an understandable anxiety about valuation. You know, if you think about it from the LP perspective, it's a ten-year blind pool that they're doing. They're giving us this money for 10 years and very little say over what happens after they invest. So there's a fair amount of anxiety, which I totally respect. So, you know, there's always, you get the eternal question on valuation. You know, and I think, you know, our philosophy on that is, you know, valuation matters. It matters a lot. But it's not as dispositive. For us, as it is for the LBO guys or the public markets. And that's because there's a simple engineering principle, which is you're only as accurate as your least accurate variable. And it's a very important point. If you're trying to value Coca-Cola, it's going to grow two or three percent a year. You know, there's not a lot of variation up at the gross margins aren't going to change. The SG&A isn't going to change. The stock is what it is. So if you get your valuation along by more than 10 or 15%, You know, you can be just way out of whack and not make money for a couple of years, because everything else is pretty bounded. T…

AI assessment note: “No, I mean, I think most LPs understand that pressuring their GPs to make investments”

Q In terms of then the kind of LP pressure on these valuations, is there a, is there a kind of pressure or a desire from their behalf in terms of investments from you?

A No, I mean, I think most LPs understand that pressuring their GPs to make investments when they don't want to is a very good way to lose a lot of money. So we really don't get that kind of In fact, you usually get the alternative pressure, which is an understandable anxiety about valuation. You know, if you think about it from the LP perspective, it's a ten-year blind pool that they're doing. They're giving us this money for 10 years and very little say over what happens after they invest. So there's a fair amount of anxiety, which I totally respect. So, you know, there's always, you get the eternal question on valuation. You know, and I think, you know, our philosophy on that is, you know, valuation matters. It matters a lot. But it's not as dispositive. For us, as it is for the LBO guys or the public markets. And that's because there's a simple engineering principle, which is you're only as accurate as your least accurate variable. And it's a very important point. If you're trying to value Coca-Cola, it's going to grow two or three percent a year. You know, there's not a lot of variation up at the gross margins aren't going to change. The SG&A isn't going to change. The stock is what it is. So if you get your valuation along by more than 10 or 15%, You know, you can be just way out of whack and not make money for a couple of years, because everything else is pretty bounded. T…

AI assessment note: “No, I mean, I think most LPs understand that pressuring their GPs to make investments”

Q Do you see this, Jason? Okay, great. I thought one that was really interesting is Aura's IPO in two different respects. One is this Robin Hood's first role as an underwriter. They're listed 18th and last. But as a precursor to what could be an underwriter of the future, is this foreshadowing of Robin Hood's next mega line of business, and Robin Hood becoming so much more?

A Absolutely. I mean, look, IPOs is about distribution, and one, you know, it's not the, no, retail is not the primary source of distribution. You know, there's typically this mental rule, you only want a certain percentage to go to retail, but that percentage has expanded, and I think SpaceX had a high retail allocation of 30%, right? It's to some extent, it's, it's, it's not enormous money, but it's free money if you're Robin Hood for just, you know, you're not, it's not like you're riding the S-one, you sign on the bottom, you distribute your shares, you can allocate them to, You know, clients, and especially in a market where you get an IPO pop, it's, you know, it's gravy all round. You make money from the underwriting fees, and you make your best clients happy with an IPO pop, right? So it's a good business to be in, and probably from the lead underwriter's perspective, especially for these high-end tech offerings, the Robinhood clientele is probably one that has a high propensity to want to buy these stocks. So yes, it just totally makes sense. Stuff like Schwab, In, in frankly, not as successful a way has ended up being an IPO distributor too, but not at scale.

AI assessment note: “Absolutely. I mean, look, IPOs is about distribution”

Q thought this would be as big as it has got as quickly as hit, as it has done. It's crossing out two, two and a half billion ARR. Um, they're doing a new round led by General Catalyst at twenty billion. And then there's rumors, suggestions, reports that NVIDIA is joining that round in a significant way. How do we think about this next layer of the start for NVIDIA?

A I think one of the things we saw with Intel back in the, the closer you are to having a hundred percent market share, the, the more you spend your time trying to move the whole ecosystem along, and clearly that's where NVIDIA is now. It's, they're using their capital to, you know, fund the NeoClouds, fund OpenAI, fund, as we just discussed, poolside, and for, Whatever reason, also fund Merkur. I will admit, when I think about all the things that Nvidia should be doing with its money, it wasn't obvious to me that funding Merkur made as much sense as some of the other bets, because the other bets are all about time expansion. If I fund the NeoCloud, they can buy more chips. I'm happy. If I fund Poolside, I can sell more open source. I'm happy. I don't get why if I fund Merkur, they can do more training, but my, probably my, unless there's some kind of strategic deal around Needing that training information from a purely financial perspective, it doesn't directly lead to more chips being sold. So it wasn't as obvious to me as the others, and maybe it is as simple as we think it's a good business at twenty billion and stop thinking about it, Rory.

AI assessment note: “the more you spend your time trying to move the whole ecosystem along”

Q We, we mentioned margin pressure on foundation models. Anthropic turns its first profit on 11 and a half billion dollars of Q two revenue. The business is getting better for Dario. This is also in a week where Gavin Baker said about Dario saying he believes that they will be the final private company. Did you see this?

A We did, and Again, let's, let's separate the hyperbole and the future from the facts in the present, right? It's not surprising they're making money, right? If you just go back to last year, right? They did four and a half billion last year, and I think their operating margins, not operating margins, gross margins went from negative blah the year before to like positive 30 or something like that, right? On track, I think, end of the year, roughly 40, right? When you have decent operating margins, like 40%, And you go from four and a half billion in a year to ten billion in a quarter, right? That means you have four billion of gross margin, right? You can't add, and that's literally in two quarters. You can't add expenses below the line fast enough to stop yourself making money, right? So it's inevitable. I mean, yeah, they, 12 X'd growth, right? Which means they probably 14 X'd gross margin if it continued to increase even slightly And the trajectory has been increasing. You're not got a X headcount or below the line training costs in six months. So yes, I'm totally not surprised that they are operating in composite. We had run numbers at the start of the year that kind of came to that conclusion. I mean, the interesting thing will be as they continue to grow, as they buy that expensive compute from Elon, if you remember, that has a big price increase two quarters, two months i…

AI assessment note: “We did, and Again, let's, let's separate the hyperbole and the future”

Q announced an incentive package to the CEO, or it was leaked, um, whereby it basically ratcheted up with different, um, prices of the company. Um, he'd get another, I think five to seven percent at two hundred billion, and then he Ultimately at five hundred billion would have circa 39 to 40%. Is this the new norm? And should every CEO be asking for Rated incentive packages alongside valuation bumps.

A It's not going to be the new norm, and if it is, logically, stock prices should go down by 10 or 15%, because, I mean, what you're basically saying, I mean, I, I read the, I saw the, I read what's available about the package, and, you know, the first question is multi-year packages, and, you know, with incentives around market cap, and, you know, in other words, significantly beyond the normal, um, CEO comp. That's been a thing. It obviously worked for Elon in the first, the 2018 Tesla package, and obviously after a lot of two-ing and four-ing, he got another package just recently, finally approved after they moved to Texas. So they're not quote the norm, but they do have, they are put in place for reasons we'll come back to, for a small number, typically of founders who fully vested in all their shares and who feel, who want to be incented again by boards who feel that they have to be incented again, right? By definition, that's not quote the norm.

AI assessment note: “It's not going to be the new norm, and if it is, logically”

Q project, I think I read. Um, it's, as we said, in terms of jobs, extraordinary in terms of how many jobs will be delivered, different numbers, but between two to 3000. Um, really it's him saying that he wants to sidestep TSMC's queue and obviously build out his own fab capabilities. How did we think about this news, both in terms of the strategic decision and the scale of it?

A This is someone with boundless ambition plus success at delivering on this boundless ambition plus access to capital at an unprecedented rate. So he's probably going to try and do all these things. It feels wildly ambitious to You know, you have to build the gas turbines to build the fab, to build the robots, to build, you know, it's like vertically integrated on every level, but he has a piece of vertical integration that's been superb, which is satellite launching integrated into Starlink has been superb vertically integrated, right? So you can join the dots in the past and say it all makes sense. I continue to think the scale of ambition, if there's any slowdown In the AI spend, then the all in bet is the one that slows down the most, the fastest, and this is the all in bet. So, you know, watch his space, but right now he's got the capital and he wants to do it.

AI assessment note: “I continue to think the scale of ambition, if there's any slowdown In the AI spend”

Q Roy, would you have broken your rules for your LPs to put money into this?

A No, I don't think I would have. Um, I've done a lot in robotics over a decade and a half. You know, we've, as I said, my first robotics deal was in 2016. My first drone deal was in 2015 or 16. There's a lot of feeling now that they're going to happen quick. I think they're real and significant, and everyone uses the, oh, the GDP of the real world is bigger than the software world. Well, no shit. It turns out two percent of the world is software and the other 98% is real. I just think it takes a lot longer than you realize to roll out robotics in the real world. It's not clear to me putting a bunch of different companies together in the same place makes it any better. I mean, it is doable because he can raise money at a great price, but by definition, a great price for the fundraiser might not necessarily mean a great price for the investor. So even though it feels like heresy to say it, and I could be totally wrong, and if I am, that's great. I mean, I, based on what I know from a distance, it's not obvious to me.

AI assessment note: “No, I don't think I would have.”

Q No, no. Uh, and then Emil Michael obviously latched onto it, and I'm trying to remember, is it Dean Ball, his name?

A It's Dean Ball, and he is current, I think it's at a policy or communications director for OpenAI. He just started there two weeks ago. Before that, he's at the Trump and part of the administration kind of on AI policy, and before that, a bunch of Hoover Institute type stuff. And he, you know, he set off a firestorm with the tweet, and then he did a little bit, oh, I can't really post because I'm now on OpenAI. Everyone was mean to me. Because I posted a bunch of stuff, and I think it was frankly a little naive comment, because there are two comments about the tweet. One is you're in a senior role at OpenAI. One, there was a hysterical tone to it, right? He used the word, um, AI communism, and yeah, it was very, kind of, over-exaggerated. And then secondly, you know, when you start even hinting about, I mean, we saw this when Sarah fired it, when you start hinting about significant regulatory Hinting at regular changes that will massively benefit you. You got to expect that everyone's going to say, dude, of course you're going to say that you're, you know, that's your side. And if you start, if you make the expensive closed source product that sells for, you know, 10, 20 bucks and the Chinese are shipping something for two bucks and you say, well, totally independently, just speaking as a common citizen, I think they should ban this shit. You got to expect that a whole bunch of…

AI assessment note: “It's Dean Ball, and he is current, I think it's at a policy”

Q Why, why do they do it then? These are most dumb people.

A Yeah. And because they believe rightly or wrongly that the impact of this technology is so important that all these things need to be on the table. And to be fair, that belief is what gave them the self motivation and the confidence To raise billions of dollars and tell that, that narrative is what it took. Because if you walked in and said, hey, I need ten billion dollars to build some stuff and it's going to have a minor impact on, you know, some parts of compute. I don't think you got your ten billion. You needed to tell a story like every great CEO. This is the world's greatest fundraising CEOs, right? Telling the biggest story. And they told the biggest story and that's what allowed them to get the, you know, Now, one hundred sixty billion dollars, right? But once you've told that story, and genuinely, once you believe it, and in the case of Entropic in particular, once all your employees believe it, if you believe this thing is dangerous from a cyber perspective, from a jobs perspective, you just suddenly end up down this road, right? All these things become next level logical if, in fact, the basic premise is correct. And if, on the other hand, you believe as for what, you know, little old me does, and I'm not them, I didn't invent this shit, but it's like, it's a really important technology, but it's not going to put 50, it's not going to put 50% of the U.S. labor marke…

AI assessment note: “because they believe rightly or wrongly that the impact of this technology is so important”

Q think they're about 0.5% to one percent of US GDP transactions or whatever it is that Eric used as the statement to validate themselves. Uh, Ramp data suggested that Anthropic now Captures 73% of all spending among companies buying AI tools. 10 weeks ago, it was fifty-fifty with OpenAI. Early December, it was sixty-forty in OpenAI's favour. Are we seeing Anthropic run away with the enterprise lunch, so to speak?

A Just to start with the facts, they actually said 73% of new spending. Right. And what it is, and the same graph shows OpenAI is still actually ahead of Anthropic in terms of total spend. But the marginal buyer in the last six weeks, eight, 10 weeks has massively shifted, which is obviously the most leading indicator. You know, people in the market today for a new AI went 70% on tropics. So just in the interest of being precise, yes, it's a claim from Rampt. I thought the open AI response of it's not, you know, the snarky comment about extrapolating from a lemonade stand was just a bad look. Right. First of all, it kind of doesn't really understand statistics. I would argue Ramp is probably a pretty accurate statistical reflection of, um, especially digital company spend in the US. I think they have got a pretty diversified customer base and they probably have decent data and they've got good data scientists. So it was, so OpenAI trying to be snark, I think was a mistake. I think the real, I think it does represent the facts, which is in the last Three months, there's been a shift in the zeitgeist, and I do believe that the marginal user, the marginal person opting for AI today, or even people switching today, the switchers are moving towards Claude, and they're moving away from OpenAI. Doesn't mean it's the end of the world, but sometimes the first thing you got to do in dealin…

AI assessment note: “I think the data was good, and the conclusion is real.”