Anthropic's Fable Banned by US Government | Wix & Adobe Hit All-Time Lows | Mistral Raising at $20BN

1. 0:00 – 1:15

   Intro

   1. EREverett Randle0:00

      Anyone that has been blindly loyal to Elon, they've all got stupidly rich. At the face of it, this is a Rubicon moment in the history of the AI industry. It's the first time that the US has ostensibly regulated an AI model based on capabilities

   2. RORory O’Driscoll0:13

      Good intentions bite you in the ass more than evil deeds.

   3. HSHarry Stebbings0:15

      This week, number one on the agenda, SpaceX completes the largest IPO in history. Then we hit on Anthropic launching Claude Fable on Monday, and the government bans it by Thursday. What does this mean for sovereignty? Next, Salesforce acquires Fin for $3.6 billion, one of the nicest teams in tech. And then finally, Adobe beats and raises, but the stock falls 6% as the CFO exits. This and much more in an incredible week of news

   4. EREverett Randle0:42

      Outside of China, there are basically no good open source models. Where are the US open source models? There's a partner here at Benchmark that says any liquidity for pre-AI SaaS companies is top decile performance, any liquidity at all. Ready to go? [upbeat music]

   5. HSHarry Stebbings1:06

      Okay, boys, I am so excited for this. We have a very special guest. Ev, thank you so much for joining us, man.

   6. EREverett Randle1:13

      Yeah, Ev. Harry, Rory, thank

2. 1:15 – 3:57

   SpaceX's Record-Breaking IPO & Elon's $1 Trillion Day

   1. EREverett Randle1:15

      you for having me.

   2. HSHarry Stebbings1:15

      Now, where else are we gonna start? Uh, SpaceX, the largest IPO in history. It was a, it was a very successful IPO, and it's held its price really well. In the last 24 hours, it's hit 2.7 trillion. Elon Musk has actually earned in 24 hours what Warren Buffett took a lifetime to earn, uh, in terms of net worth, uh, and he's now a trillion dollars richer than the next person as a result of that increase. How did we evaluate the SpaceX IPO?

   3. RORory O’Driscoll1:45

      Well, it was, uh, with jealousy. It was an amazing outcome, and you said it's traded well. Uh, I mean, arguably, it's traded way more than well. I mean, it's up, like, you know... I, I haven't checked today 'cause I just got off the car, uh, off the plane, but, like, it's up 30, 40%.

   4. HSHarry Stebbings1:58

      Yeah.

   5. RORory O’Driscoll1:59

      I mean, it's been an astonishing performer. On the day, I mean, I was trying to decide, you know, Ev's old, um, Benchmark partner, Bill Gurley, like, you know, is there a number above which Elon gets into trouble with Bill for leaving money on the table? And I was watching it during the day. It went up to 30, I'm like, "Oh, Bill's gonna be mad." And then, uh, and then he dialed it perfectly. I think it ended at 19, which is the perfect pop, the designer pop. So he deli- So not only has he changed the world, built the biggest, most valuable company in the world, but he pretty much, without any price discovery, remember we talked about this last week, he didn't do any price discovery, he just told everyone the price he's gonna take, went out and got it, and then landed the plane at a 19% day one pop, which is kind of at the high end of perfect. So I was like, "This is perfection." And then in the last two or three days, it's just traded up nicely since then, so it's just amazing. I mean, I don't know what you say. L- it's really hard to say, "I made $1.2 trillion, but I might have left 50 billion on the table. How do I feel?"

   6. HSHarry Stebbings2:50

      [laughs]

   7. RORory O’Driscoll2:50

      I think he feels okay. I don't know.

   8. EREverett Randle2:52

      Yeah. Yeah, so I, I mean, the one piece about all these IPOs, a- and we, we can't help ourselves because the company starts trading, you can do the math on everyone's net worth immediately once it starts trading. But I do wish there was, like, a moratorium on even talking about or even if you own the shares looking at a stock price when, um, when, when the shares are still locked up. What I mean by that is, is, is, like, all of these... I mean, a traditional lockup means that any of the, like, venture insiders, the CEO, the management team, the employees, they typically can't sell for, like, six months. And what often happens when you have, um, so little of your shares trading, so, like, right now, of SpaceX's total share count, only about 4% of the shares are being traded. That's like a tiny, you know, they call it float. It's a tiny amount of float, and it allows for all sorts of weird things like gamma squeezes, um, which we can talk about, or any of these other things that make the share price just insanely volatile over the first few months of trading until the lockup releases, and so there are, like-

   9. HSHarry Stebbings3:52

      Ca- can we just stay on that? What is a gamma squeeze, and how does it impact SpaceX price?

3. 3:57 – 9:12

   What Is a Gamma Squeeze and Why It Matters for SpaceX Stock

   1. EREverett Randle3:57

      Yeah. So a gamma squeeze, so, so when, when people are allowed to trade options on a stock-

   2. RORory O’Driscoll4:01

      Which opened today. SpaceX options started today, so it's gamma squeeze day

   3. EREverett Randle4:05

      That's right.

   4. RORory O’Driscoll4:06

      You're exactly right, Ev

   5. EREverett Randle4:06

      Exactly. And so, so what a gamma squeeze is, if I buy a call option, I'm buying the right to buy a stock at a certain price, and so there's some market maker on the other side of that trade writing me the option. And what they need to do on their side in order to hedge their risk is buy some of the stock directly. And so what a gamma squeeze is, is when you have a ton of people buying a ton of call options on a stock, you then have a bunch of counterparties, a bunch of market makers that then have to go structurally buy the stock in order to hedge their risk. And then you create this cycle, this, like, self-reinforcing loop, where the more call options people are buying, the more forced buying there is from market makers, which then forces more, or counterparties, which forces, w- well, which then encourages more retail people to buy more call options 'cause the stock is going up. And so when there's only 4% of the, the float trading, only 4% of the, the shares trading, um, there's just, it's so thinly traded that something like a gamma squeeze where you have this forced buying reinforced loop can happen very quickly. So that's not to say that the stock's going to collapse or anything, but when you have this sort of situation when a, when a company starts trading, especially a really hot company, um, y- you, the, the price action, even though it's very fun to talk about because, you know, um, Elon can make, you know, Warren Buffett's net worth in a day or whatever, um, you know, nothing matters until the lockup's gone because that's the only time that anyone can actually... You know, it's almost like a private, it's like a private mark. Like, it looks good. You love to look at it. You put it in your little spreadsheet, but you can't take it to the bank. Um, and so it, it matters much less than, than, you know, what, what... Like, the, the stock price in six months from now, like, that is what we should be talking about, and that's going to be the barometer by which we can grade, you know, investors and insiders. And there's been plenty of IPOs before where, you know, people do the victory lap. You know, on day one of the IPO, the stock goes down 60%, and then it's a different situation once, once the lockup's up.

   6. HSHarry Stebbings5:56

      Ev, you'll regret joining this show very quickly. Six months time-

   7. RORory O’Driscoll6:00

      [laughs]

   8. HSHarry Stebbings6:00

      ... will the stock price be above or below where it is today? I'll give you the over or the under

   9. EREverett Randle6:04

      [chuckles] I, I would probably just because of the retail mania, um, around, around the stock in particular, I, I would personally probably take the under in six months, not because I don't think that the company's gonna be valued extremely well, and I have a couple funny stories on this. But I, I think that, you know, 4% float, um, call options coming online, there are just so many, like, ph- like, engineered things that are going to make the stock price go up over the next month, including some of this index inclusion-

   10. RORory O’Driscoll6:33

       Yeah

   11. EREverett Randle6:33

       ... where there's more forced buying. It's just there's no shares available. There's a lot of forced buying. There's a lot of retail activity. It's gonna be popular for being, for people to be buying call options on this. Um, so I think it's gonna still be worth a ton. Like, I still think it's gonna trade really well, but if I had to go over/under, I would go under from, from six months from now.

   12. RORory O’Driscoll6:50

       And as you know, that's where I come out when you ask me, but there's just such a lot in that. I mean, and yeah, and first of all, I just, in passing, had a laugh at that. You're exactly right. There's so much room between $2.6 trillion and even the last private round at $400 billion. There is $1.8 trillion of value between this and the last private round, so yeah, there's lots of room to adjust. But I do think Everett's right, is that, you know, the fi- two comments. One is, yeah, my gut would be lower rather than higher six to 12 months from now, and I think he's also right, the short-term comment of, you know, in the interim, there's just a lot of news that will be s- strong good news and strong technicals in the face of a low float, which means the short-term bet is much harder to call. Uh, you know, we can talk in a second about Cursor, where I believe Benchmark has a stake, so, um, I believe they just announced that's gonna close. Um, but I, I, I... And the funny thing is, when I do this show, I always try and avoid having opinions that I didn't have the courage to act on, so I started saying to myself, "Okay, Rory, today options became tradable for the first time," right? If I think it's gonna go down, you can make that bet, right? So you go out and you price it, and you realize how hard it is to make money. I mean, roughly six months, six-month-at-the-money puts, the idiot version, is you have the right to sell the stock at the same price, at the same price six months from now, right? It's gonna be depending on... They're estimating volatility so high, and volatility is what drives the pricing of options. And options, stepping back a million miles, options are the coward's way of shorting the stock. You know, if you short the stock naked, you can lose all your money. If you buy a put option, you just, you know, you, you're taking a bet. In the worst cases, you can, you can lose the amount of the consideration. So it's, it's the baby way of shorting, and I'm a baby, right? So I look at it and I go, "Oh my God." It's roughly 20%. If you wanted to shoot, buy an at-the-money put, it's gonna be based on the estimates yesterday, and I haven't seen the updates today, stocks just, options just started trading, 20% of the share price. So if the share price is at 200 bucks, it would be 40 bucks just to buy the right to sell the stock back at par, which means if it goes down to 160, you only break even, right? So you gotta get right down to 120 to make a 2X, and that's a 2X on a security where you can lose all your money, right? So I'm looking at it and going, "I might have these opinions about the SpaceX price going down. Do I have the nuts to put a million bucks on the line and say I believe it's going down and buy those options?" Decided I didn't have that just yet, right? [laughs] 'Cause the VLO-

   13. EREverett Randle9:11

       And I, and I will say,

4. 9:12 – 12:01

   Elon's Long-Dated Call Option Playbook: From Tesla to SpaceX

   1. EREverett Randle9:12

      I, I mean, I, I think the one thing that, um, people very much underestimate, um, I, I had a coming-of-age moment as, as a young value investor at the time. I think every value investor's gone through this. So I, I was in this value investing club in college where you basically, you know, you read Benjamin Graham and Howard Marks and think you're smarter than everyone else and buy things at, like, you know, five times PE. And, um, after college, I was still on that kick, and so I, I sh- I, with all my college savings, I shorted Tesla. And I feel like everyone has gone through this experience if they were a value investor at some point in their life, where it was like, you know, you just, "Oh my God, Tesla, maybe it's not a fraud, but it's so overvalued. It's an auto OEM," all these things, and I lost all my college savings shorting Tesla. That was such an unbelievably valuable lesson for me in my life, and I was able to rectify it years later, which many people aren't. But I was able to rectify it when I went to Kleiner Perkins. The very first investment I led in 2022 was actually SpaceX. And when we were going through and when people, you know, um, LPs or anything else, when there was questions around, like, "Well, how much, how much upside is there?" I think it was at, you know, $120 billion or something. "How much upside is there here?" And the way I talked about the investment was like, look, the numbers alone, the n- the numbers alone get you to a solid return. But it would be, it would be dumb to not incorporate what Elon can do, which is he sells the market on these incredible long-dated call options, like these tech call options. He's, he's made his whole career doing it. And so it's like at the time, whatever it was, 2019 or whatever, you know, Tesla wasn't worth, you know, the $800 billion of its or the $400 billion of its market cap, but he convinced everyone that, like, hey, I'm gonna figure out full self-driving, and that's gonna drive a trillion dollars of value, and it's gonna take, you know, eight to 10 years. And then you get to that point, and he solved it. Tesla on the numbers still probably isn't worth a trillion dollars, but now he has Optimus, um, that, that's another, like, 10-year project that's going to create trillions of d- of, of dollars of value if he figures it out. And guess what? He's probably gonna figure it out. And so with SpaceX, he's done the exact same thing, where at first it was, you know, rocket reusability, and then the second one was Starship's gonna work, and now he has orbital data centers, the Mars mission, the moon, you know, the moon, um, mass launcher, all these things. So, so he sets up these stories which he honestly usually accomplishes what he's gonna lay out, but they're basically these long-dated call options where he can go to the market, he can go to shareholders and say, "Look," like, on the numbers, obviously, SpaceX, you can't really look at the numbers of SpaceX. If you were just looking at the numbers and the P&L, P&L alone, it'd be very, very hard to even get, like, a $2 trillion valuation, um, as, as a fair market value. But he says, "I'm gonna figure this massive thing out," and that's where the next trillions of dollars of value are going to come from. And historically, he's had such a track record that the market's willing to believe him.

5. 12:01 – 20:27

   How Musk Solved His Compute Problem Twice

   1. RORory O’Driscoll12:01

      When we look at the assets that you mentioned, and that big news story from today is they're exercising the right to buy Cursor at $60 billion. Um, it's looking like an incredibly prescient deal. I think it's at four billion today. It's gonna be six billion by the end of the year.

   2. EREverett Randle12:17

      He's acquired one of the best teams with lockup, and so he's got retention of them baked in

   3. RORory O’Driscoll12:22

      Yeah. For a price equal to one-third of the variability of stock between yesterday and today. It's, it's, uh, exactly. It was a great deal when he did it. It's a better deal since then.

   4. EREverett Randle12:30

      Yeah.

   5. RORory O’Driscoll12:31

      Exactly right. Um, it, it-- And the odd thing is he solved his what am I gonna do with this compute problem not once, but twice. He solved it first of all with, "Oh, look, I have Cursor," right? I forward-bought Cursor to fill the gap in Colossus. And then obviously since then, uh, but before the IPO, he announced the contract with Anthropic for one point two five and a contract with Google for, I think, seven hundred billion. So roughly two billion a month of compute, again, filling the gap in the Colossus revenue stream and, you know, doubling his revenue. So I, I agree. I think that between, you know, twenty-four billion a year in, um, kinda CoreWeave-like revenue from Google and Anthropic and then a six billion dollar run way to six billion year-end potential Cursor, uh, yes. His, his AI business now as a, as a matter of fact, as a statement of reality, as of the minute those contracts kick in in September, the revenue run rate of the quote-unquote "AI business" across, um, Google, Anthropic and Cursor is larger than the revenue run rate across the entire SpaceX and Starlink business. So yeah, I mean, just great kind of corporate execution. I think it's the move fast comment. It's just everyone else has spent two years thinking, "Yeah, we should probably do something in AI. Maybe we should build a data center." He's like, "No, I've built two data centers. Didn't get the model working great, so I bought a company to fill it with, um, kinda coding, and then that wasn't enough, so I did two huge contracts." Moving right along, people. It's year two now. I mean, it's the, it's the execution speed that's just so impressive.

   6. EREverett Randle13:58

      Hundred percent.

   7. RORory O’Driscoll13:59

      And, you know, t-t-- I love, well, I just love your comment on the long-dated call, 'cause the really weird thing is, kinda contrasting it, the s- the volatility in the stock is super high, and that's, the, the, the option traders are saying that. But you contrast that, the implied on, the, the implied predictability of the long-dated call options is actually they're basically saying they're highly predictable. In other words, the stock is moving around like a crazy thing, but the valuation is effectively saying, "Hey, we, the equity holders, believe, as Ev believed, that this guy has done these things." They're worth e- whatever. I was gonna say X, but that would be stupid, because the thing itself is X. That would be self-referential. But they're worth four hundred billion dollars today. He has ideas to be worth a trillion more, and I'm just gonna give him a hundred percent credit for it. I, I, I agree. That is the secret to the fundraising. I mean, if it ever stops, if the time to deliver goes too long, there's a big gap between fundamental value and anything you got here. But right now, he gets the benefit of the doubt like no one else. And the weird thing I-- 'cause I've been thinking, obviously anyone in our business who's not thinking about it this week is a moron, is the competitive advantage that gives him is, 'cause his cost to capital is low.

   8. EREverett Randle15:07

      That's right.

   9. RORory O’Driscoll15:08

      Right? Because no one else could have built... I mean, if you'd gone into your local bank and said, "I think AI's gonna take off. I have a crackerjack engineering team. I really can build Colossus one and two in less than a year. Oh, I'd like twenty-four-- And I don't have any contracts, but I reckon shit'll turn up. I'd like twenty-four billion dollars," you'd have got laughed out of there. He gets the chance to roll the dice, and it comes up, right? 'Cause he's, I mean, no one could have done that unless they had the cost of capital, and the cost of capital allows, it's kind of a self-reinforcing cycle. It allows him to make these bets that no one else can. I mean, he's earned, he's earned the right to play.

   10. EREverett Randle15:39

       I remember both for the X deal and when xAI was fundraising, um, there were a lot of people that were the, you know, the, again, on, on, like, on, on just the pure dimensions of the companies, both of those investments were hard to look at and think that they were gonna be good deals. Um, so the, the X takeout at whatever, forty-five billion or whatever it was, and then just generally xAI just felt like it was too little too late, um, to, to keep up. And what people forgot in that moment is that anyone that has been blindly loyal to Elon in terms of any time he's asked for money, you could just give him money for whatever he wants you to invest in. They've all got stupidly rich. Every single person. There's like a small village of these people, by the way. Like, there's all these funds that you've never heard of-

   11. RORory O’Driscoll16:25

       To be clear, it's going to be a very-

   12. EREverett Randle16:26

       ... that ninety-five percent-

   13. RORory O’Driscoll16:27

       Of them are very-

   14. EREverett Randle16:27

       Like ninety-five percent of their cumulative invested capital is in SpaceX and/or other Elon companies. They all got st- they've all gotten stupidly rich. And so the, the, like, the amount of surplus... And by the way, you could say that about Tesla shareholders too. Anyone that's held Tesla for ten years, anyone that's had anything to do with Elon and just trusted him with their money has gotten unbelievable returns out of it. And so the amount of, like, surplus goodwill he has to burn down, like, I think he could, I think SpaceX, I don't think it will, but I think it could go nowhere operationally for five years, and people would still be a believer in him, because he's basically done it for shareholders every single time that they've gotten involved with him. And to your point, Rory, he has a much lower cost of capital because he has an army of people, like a small village of people, that will blindly give him money to do whatever he wants because he's such, he's been such an unbelievable steward of capital to anyone who's given him money.

   15. RORory O’Driscoll17:20

       Yeah. I, I, I wrestle, uh, and, and you're right, and I, I wrestle. To be honest, I don't know if I could get my head around that belief statement in an investment memo, even though I objectively recognize, Ev, it's true. And you're right. I say, you know, I know some of, as you say, the small village, and it may be a small number of people, but they're going to have very large houses in that village pretty soon. [laughs]

   16. EREverett Randle17:37

       [laughs]

   17. RORory O’Driscoll17:37

       So they're gonna be, and they're going to be in Incline Village for all the, which is just over the border in Nevada, for all the obvious tax reasons.

   18. EREverett Randle17:43

       That's great. That's great.

   19. RORory O’Driscoll17:43

       So yes. They're gonna do just fine. I would wrestle with that kind of approach. The rationalist in me says that's not the way capitalism should work. Even the best person should have all these different deals. But you're right. It's worked. And it does mean it's single point of failure in the sense of if it goes wrong, it will go wrong for everything, 'cause it'll all be... 'Cause he's gonna, I mean, they are gonna roll in Tesla, and it's all gonna be one big happy family. It's gonna be awesome, right? But it does mean he's kinda taking on this ever larger burden of making more and more return for more and more people, and the whole mystique of it is tied up in that sentence. Anyone who's piled in has never lost money. And there's a little part of me that goes, "Wow, that's a tough way to live your life," versus the we do deals, some work, some don't. But- It's worked for him, and it's probably gonna... You, you-- When people-- Something works that well for someone for 20 years, they don't change it.

   20. HSHarry Stebbings18:33

       Will Tesla and X be together in three years' time?

   21. RORory O’Driscoll18:39

       Look, I don't... It seems to me that it's the thing that's happening. But, you know, I, I mean, I told, um, the CEO Gwen Shadwell just said, "It's probably easier for Elon if they're together." And I, I reckon if I was worth a trillion dollars, I would probably solve for things on the basis of, "Oh, it'd be so much easier if I only had one board, not two. I had one set of shareholders." It seems to me a thing that could happen very comfortably. And again, back to what I've said is that, and there ain't gonna be a ton of votes saying no, especially after you've made them two, $2.2 trillion.

   22. HSHarry Stebbings19:07

       I think if you look at the Polymarket Kalshi odds on a Tesla-SpaceX merger, again, you, you know, not, not that those markets are so big that they're, um, that indicative, but I think the probability that they have on there is something like seventy percent in the next-

   23. RORory O’Driscoll19:19

       Yeah

   24. HSHarry Stebbings19:20

       ... two years or something like that. And so the, the-- some free market seems to believe that it's, that it's more likely than not, but I have no idea.

   25. RORory O’Driscoll19:28

       On the other hand, some poor fool bet-- felt that it was ninety-five percent likely that Spain would beat Cape Verde last night and put a million dollars [laughs]

   26. HSHarry Stebbings19:34

       That's right. I saw this for an eighty-five thousand dollar gain. Yeah, yeah, yeah.

   27. RORory O’Driscoll19:37

       So yeah. So, so yeah. But El-Elon is more predictable than a Spanish soccer team, so keep rolling. I'm sorry I cut you off. The coffee's kicking in.

   28. HSHarry Stebbings19:46

       [laughs] He's not, he's not-

   29. RORory O’Driscoll19:48

       Oh no, I was done. I was done

   30. HSHarry Stebbings19:49

       ... he's not, he's not used to such caffeine. It's okay.
6. 20:27 – 29:11

   Anthropic's Claude Fable Banned by the Government

   1. RORory O’Driscoll20:27

      you know, this is the front-end model to the Mythos model, which was the mo-- I mean, if you start by Dario, when they announced that, they said, "This is so dangerous for cybersecurity that we're not gonna make it available to everyone." So they kind of prefigured that this was dangerous, right? Then obviously, they announced th-this model, which is effectively a front end to that model, and quote, unquote, "shouldn't be able to do this kind of cybersecurity stuff," right? So they pref-- uh, prefigured that this was problematic. And then what happened was, I guess maybe sticking with the facts, first of all. It looks, now looks like that, uh, uh, Amazon discovered a case whereby you could in fact interrogate the model and have it give, you know, some cybersecurity input, right? So they pick up the phone, they ring the government, and you know, that's to be fa-- Stepping back, I'm sorry, I'm a bit incoherent here, but you know, the two parties in question don't really like each other, right? So no one's assuming good faith, right? And I look and I read the Dario statement, I read David Sacks' statement, I read all the other stuff, and you can literally see everyone according to their rights is correct, right? But there's just zero communication. That's the big u- There's just zero communication, zero trust. So something that I, I, you know, I listen, I go, in terms of the actual facts on the ground, I see Dario's point in terms of what he was saying, th-- why that quote, unquote, "jailbreak" of the model didn't represent a meaningful threat. And it was coherent with why he thought Mythos in general was threatening. So it was a very logical, coherent, as you'd expect from a world-class scientist who is a very, very smart man. At the same time, politically, I get why the other guy said, "Fuck this. After ninety minutes, I'm pulling the plug," right? And it's gonna be really problematic. So that's kind of the big picture comment, and maybe kind of drilling specifically on the cybersecurity comment. The thing is, it-- So what they've said consistently about Mythos is really, it's not that any one cybersecurity bug it finds is so terrifying that, oh my God, no one else could have done this, and now you can. It's that the scale at which they can find bugs and problems is what's terrifying, which is what AI, AI doesn't get tired. It does, you know, can find a thousand, and you can't defend yourself against a thousand. So when this one thing happened, their comment was some version of, "Hey, look, yes, I get that this thing gave cyber advice, and the way the model was released, it wasn't meant to do that," right? Um, and, and Fable wasn't meant to do that. So technically, it was a foot fault, right? So, but they were saying to Amazon, and they were ultimately saying to the government, "But the real danger of Mythos is, hey, it's that they can spin up a thousand or ten thousand different instances, all of them finding cybersecurity things, and we get overwhelmed, and that's not what's happening here, and this fault wouldn't have allowed that to happen." So they were kinda technically correct. However, if you run around telling everyone that you've got the scariest thing in the last 20 years and it shouldn't do any subscri-cyber, and we're not gonna let it do any cyber, and then it does cyber, and they don't like you anyway, they're gonna pull the pin on you.

   2. HSHarry Stebbings23:09

      That's right.

   3. RORory O’Driscoll23:09

      So that's, that's what happened. They just said after 90 minutes, they're like, "Fuck this. Um, you're not taking us seriously." And you can see it. Dario being very logical. "I'm a scientist. I'm a computer scientist. Let me explain why this doesn't matter." And you got a bunch of people, "I'm the chief of staff for the US..." You know, they had a assistant chief of staff on the call, so reporting to Susan Rice. They're like, "Dude, if you think I'm gonna brief the president of the United States on, you know, multiplicative instance of cyber instance, and he's gonna get that, you're dreaming," right? "We're g- we're pulling the pin here because you either stop it..." They were like, "You stop it now, or we're gonna, um, declare this export restriction thing," which means you just can't bring it. And it's very restrictive, and the terms of this are pretty restrictive. So you can see how it all happened, right? And both sides to their lights, in ki- in the context of where they're coming from, it all made sense to each of them individually. But it's just a comms mismatch. And, and it's problematic for Anthropic because one of the big zahaz is this Export Restriction Act, which is the thing upon which the ban was put in. Unlike the thing that happened with Hegset, there's no judicial review here. This is clearly within the powers of the administration to make this declaration, right? So they're offside, right? They don't have a ton of leverage in court, right? So i-i-it's a d- it's a tricky situation.

   4. EREverett Randle24:19

      You think that the hardest, the hardest part of this story is that it's just that there, there's a lack of reliable narrators.

   5. RORory O’Driscoll24:25

      Yes.

   6. EREverett Randle24:25

      You know, like, like you mentioned, Rory, there's like, there's two parties that clearly don't like each other. There's been a ton of conflicting reports on this.

   7. RORory O’Driscoll24:32

      Yes.

   8. EREverett Randle24:32

      Um, that there hasn't been- I mean, not that journalists get things, um, 100% right, uh, most of the time anyway, but that there's been, like, actually contradictory reporting on, on the way things went down and, and how, how it all went through. I actually think what matters most, though, is what happens next. Like, I think, um, we're gonna know a lot more about the significance of all of this, um, in a few months' time, in like six to 12 months' time than, than now because at, at the face of it, this is a Rubicon moment in the history of the AI industry because it's the first time that the US has ostensibly regulated an AI model based on capabilities. The first thing, the first scuffle with Anthropic was due to a, um-

   9. RORory O’Driscoll25:14

      Terms of service, yeah

   10. EREverett Randle25:14

       ... disagreement over the basically, like, the contract that they were signing and what the government could use the models for. They were not regulating it or restricting it based to, to, y- you know, to non-US citizens based on capability. They are now saying, at least on face, that they are regulating an AI model based on the capabilities and what those capabilities could do in, um, foreign adversary hands. Um, and I think that's a huge deal because... A- and what we're gonna see is if they're, they actually mean that, um, or if this is, again, just like another battle between Anthropic and the government. But when OpenAI or when Google, like, th- th- they're going to get to Mythos Fable quality models-

   11. RORory O’Driscoll25:53

       Capabilities. That's a great point

   12. EREverett Randle25:54

       ... very soon, within the next three to six months, and then we're gonna see. We're going to see if the government actually wants to regulate models based on these capabilities, and if we've crossed this Rubicon where now the government is going to gate access to intelligence. And I think the most interesting thing is if we extend the analogy even further, and there's been some people on, on, on, um, on X that have talked about this as a, as a possible future. Imagine if we get ASI. Imagine if we actually get things like recursive self-improvement kicking in, and we actually have artificial super intelligence. We have this vision of geniuses in a data center. What happens when the government is gating access to super intelligence? Who gets access to that? Do NATO allies, do, uh, you know, just a, a select few allies of ours get, get access, and their citizens get access? If we think about how powerful a lot of people think AI is going to be and the power of the government to either give or restrict access on a country-by-country basis or a citizen-by-citizen basis, it gets pretty freaky from a, from a, uh, like a macro and geopolitical standpoint because, um, imagine if you have, like, the economic implications of the United States and China having u- uh, having access to super intelligence and, you know, Greece not having it. Um, it's, it's, you know, that's just like a random, random country-

   13. RORory O’Driscoll27:09

       Whoa

   14. EREverett Randle27:09

       ... but, like, imagine maybe-

   15. RORory O’Driscoll27:10

       What did the Greeks do to you? [laughs] I mean, yeah. [laughs]

   16. EREverett Randle27:13

       Or, or Argentina or Brazil or, or India-

   17. RORory O’Driscoll27:15

       Or on holiday in Athens. [laughs]

   18. EREverett Randle27:17

       Yeah. [laughs]

   19. RORory O’Driscoll27:17

       The fucking Greeks. [laughs]

   20. EREverett Randle27:19

       Yeah, yeah, yeah. But, but any of these, you know, like, if you only have two countries that have super intelligence and everyone else is, is, is metered or gated, it makes a lot more sense why people are starting to take the idea of sovereign AI much more seriously. The only issue is that n- you know, none of these sovereign AI plays have, have amounted to anything yet. And so it's, it's, um, it's, it's sort of a, uh... Yeah, so, so I think in, in general, um, it's, it's, it's a Rubicon moment, and I think we're gonna know a lot more in six to 12 months than we do now about how important this moment in time was.

   21. RORory O’Driscoll27:47

       Agreed, and yeah, I'm sure if I am the Mistral shareholders, I am lighting a candle in church at this one and going, "This is the best thing that happened." But I wanna go back to... It's, it's so- The funny thing about this thing is it's so illogical. What I love about some things in politics and crisis is often people end up on surprisingly the wrong sides. And if you're looking at Dario, the guy who's been saying, "This is dangerous. We all need to be careful," he's been funding the pact for regulation, right? Meanwhile, on the oth- And, you know, some VCs have been on that side. Meanwhile, on the other side, you've had, um, you had the administra- David Sacks, in my view, doing, basically hard to say, but excellent work trying to keep, you know, the government out of the hairs of AI, right? You've had a bunch of VC investors giving money to the Leave AI Alone pact and kind of feeling like they're with the administration, and suddenly everything's mixed up. Now, Dario's in the very tough position of saying, "When I said it was really dangerous and we should regulate it, what I meant was it's really dangerous and we should regulate it, but not regulate me." Ooh, it's a tricky one. And then on, on the other side of the table, all the, "Oh, leave it alone. We should deregulate this thing," suddenly this administration, as Ev said, is taking one of the strongest actions we've ever seen against the technology, right? So it's, it's kind of everyone's like, uh, as they piece through what's happening, everyone should be looking, going, "Hang on. I thought I was on this side of the table, and you were on that, and we've just switched," right? Which, again, makes it hard.

   22. EREverett Randle29:06

       Forgive my naivety. If you were internal within Anthropic's research teams, what do you do now?

7. 29:11 – 33:12

   Does the Export Restriction Act Even Make Sense?

   1. RORory O’Driscoll29:12

      Well, first of all, if you're, uh, internal to Anthropic's we t- uh, team and you're not a naturalized US citizen, so you step away from the keyboard. Because otherwise, you and your organization is gonna be in breach of the Export Restriction Act, and I'm willing to bet there was an attorney at Anthropic who knows exactly the penalties for that right up and down the line. So first of all, and let's get frank, a huge percentage of these teams are overseas, highly intelligent-

   2. EREverett Randle29:33

      I think Andrej Karpathy is actually one of them.

   3. RORory O’Driscoll29:35

      Oh, wow. It's, uh-

   4. EREverett Randle29:35

      Yeah

   5. RORory O’Driscoll29:35

      ... I mean, there's a reason my wife, on the fifth year, the, the day we were eligible, applied for our green cards, 'cause at some point they'll turn on you. Um, so yeah. So first of all, those guys step away. What do you do if you're an American and you're there, hmm, you're like... I mean, the weird thing is, I, I really like what Ev said 'cause, uh, uh, you, you, you triggered a thought, me, Ev, is that, but if they don't treat the other labs the same, then interestingly Anthropic may have some kind of due process claim. Because, uh, y- uh, I, I, I'm guessing here, but yeah, if the Export Restriction Act... I mean, there's probably some op- uh, it's pretty untrammeled CEO administrative power because it's a national security thing, so it's, like, delegate. But I do think if you don't treat everyone the same, it'll be problematic. So sometime within the next... [laughs] I'm just trying, I'd love... Again, it'll just mix things up. Does, does OpenAI go with the, "We're as good as them"? But then they don't want to get regular. Or do they go in, "We're not quite as dangerous as Anthropic, so you kick those guys in the nuts. We're the AI that's not quite clever enough to fuck up your cybersecurity, so we can sell to everyone." It's just really zany. But if they are comparable and they don't ban all of them, then they have a problem. You're right, because they're being illogical. But if they start doing it, it's a huge moment. You're right. It's-

   6. HSHarry Stebbings30:44

      I- I'm just struggling here though, guys.

   7. RORory O’Driscoll30:45

      Yeah.

   8. HSHarry Stebbings30:45

      'Cause you said that-

   9. RORory O’Driscoll30:46

      Have we done that?

   10. HSHarry Stebbings30:47

       'Cause you said about 6 to 12 months. 6 to 12 months today, it, I mean, it's, it's years in prior cycles.

   11. RORory O’Driscoll30:54

       That's right.

   12. HSHarry Stebbings30:55

       Uh, we were planning a, an Anthropic IPO.

   13. RORory O’Driscoll30:58

       Yeah, this is definitely a downer for them.

   14. HSHarry Stebbings31:00

       Yeah [laughs] .

   15. RORory O’Driscoll31:00

       I mean, you know, I can tell you that there's someone updating the prospectus.

   16. HSHarry Stebbings31:04

       Say, say, you know, Titanic on go goes 7 out of 10 holiday. Yeah, it's a downer [laughs] .

   17. RORory O’Driscoll31:08

       Well-

   18. HSHarry Stebbings31:09

       Yeah

   19. EREverett Randle31:09

       So, so here's... So, so, um, th- this is, this is a bit of a tangent, but I think, I think the, um, it, it's related in that some of the... Like, if you talk to some smart people in AI, they will tell you, actually, with the right harness and the right amount of test time compute, you can use open models to find-

   20. RORory O’Driscoll31:26

       Yes

   21. EREverett Randle31:26

       ... all of the vulnerabilities-

   22. RORory O’Driscoll31:27

       Correct

   23. EREverett Randle31:27

       ... that Fable can find. And so actually this is sort of a nothing burger because you can actually replicate, replicate this, um, with, with non, non-Mythos level models. I think again, like the, the legibility of AI to the administration, I mean, hell, the, the legibility of AI to a lot of people in Silicon Valley, it's just such a fast-moving field that if you're not inside one of the labs, you know, sometimes-

   24. RORory O’Driscoll31:48

       You don't know. Yeah

   25. EREverett Randle31:48

       ... it's hard to just keep up with everything that's happening all the time. Now, you know, now think about DC and the average age of someone in the administration or a lawmaker in, in Congress or, or the Senate. The, the, the legibility of these things is very hard. So I do think that by saying like, "Hey, we're..." You know, by, by pounding their chest and saying, "We're, you know, we're, we've created this God model that's very dangerous," um, Anthropic has painted a, a target on its back. I'd say the s- the, the, like the flip, the inverse of this whole thing where you could say, "Oh, well, that's like a negative for Anthropic in the labs and, and IPO prospects and everything like that," the massive positive from this is this idea that with the right amount of test time compute, basically the right amount of inference thrown at a given query into a model, um, or, or a harness with an open source model, you can replicate these results. And Noam Brown- [laughs]

   26. RORory O’Driscoll32:39

       Sorry. The balloons are blowing up here, guys. [laughs]

   27. HSHarry Stebbings32:43

       Oh my God. That was, that, that just inside of the test time compute.

   28. RORory O’Driscoll32:46

       The Rory celebration balloons.

   29. HSHarry Stebbings32:47

       Yeah.

   30. RORory O’Driscoll32:47

       Yeah.
8. 33:12 – 38:09

   Why Benchmarks Are the Wrong Way to Think About Models

   1. EREverett Randle33:12

      And so the really, really good thing for I think all the frontier labs and just the inference market in general, companies like ours, like Fireworks, that run inference platforms for, for their enterprise customers on open models, all of these companies that have to do-- that are in the token path, is that it's very clear that if you just throw more test time compute at any frontier level model, you continue to get results and it's, it's... Like you, you almost like... They, they haven't found where the wall is where you stop getting better results the more test time compute you throw at it. And so w- we've, we've... You know, there's been all these like step function increases in how much compute AI models use. You know, we went from like very simple autoregressive, you know, next token completion to these agentic models that, you know, do chain of thought and like agents use, you know, an order of magnitude more inference. And now you have the, this, this whole thing around, well, if you wanna, you know, have Mythos-like performance, just spend a lot more compute it. You know, do a lot more inference, spend a lot more tokens. And so I think the, the, the other side of this is that even though we've had such an insane, um, growth in the amount of tokens processed for the industry over the last three years, we actually might see another kink in the curve, um, as more test time compute creates more and more of these unbelievable results for, um, for capabilities and, and things that you can do with models, whether or not they're, they're, you know, Fable or Mythos or whatever OpenAI comes out with, but open source models as well.

   2. RORory O’Driscoll34:33

      But what you're saying on that is, and yes, and I've seen the same statements, which is that, yes, the, uh, the Mythos can get there quickly, right? Um, but, and it actually is the defense that Dario's offering is that a, a, a, a, a open source model just given enough time will find many of the same, um, issues, right? 'Cause what it means logically then is, you know, if you can't have, if we can't have Mythos at all for national security reasons, then we can't have open source models provide if they have more than X compute. And, and you're right. What it points to is this is going to be an unsustainable medium-term position, right? It's hard to imagine, like unless the US Government really wants to start regulating AI up and down the board, this is gonna be... It's going to sound like it was an easy decision one Friday afternoon. It's gonna be a very hard decision to implement it with any degree of coherence-

   3. HSHarry Stebbings35:21

      Yeah

   4. RORory O’Driscoll35:21

      ... across 12, 24 months. And for what it's worth, I think that's because the original premise, which is that this is dangerous, is in and of itself incoherent.

   5. HSHarry Stebbings35:28

      To-

   6. RORory O’Driscoll35:28

      There are risks, but the idea, the... I mean, I think this is where the kinda the cry wolfism, overwrought cry wolfism is biting everyone in the ass, and I think we're just gonna have to figure out how to deal with this.

   7. HSHarry Stebbings35:39

      Well, that was gonna be my question, which is like just to what extent is it a model capability question versus a miscommunication question from your marketing?

   8. RORory O’Driscoll35:47

      Well, it is a model capability question. They can do the thing that they've said, and it's also true that, yeah, Devin's point, while the open source stuff can do it, um, a, a, a non-frontier model can get to the same place. You know, I love this. I, I often use this expression, you know, the Russian army quantity has a quality all its own, right? And the ability to find 1,000 bugs in 10, you know, 2 hours-

   9. HSHarry Stebbings36:09

      Yeah

   10. RORory O’Driscoll36:09

       ... is more terrifying than the ability to find 1,000 bugs in 1,000 hours, right? So it was. There wa- there was an issue that had to be addressed here. It wasn't just marketing. And oddly enough, in a weird kind of way, they tried to address it, but, um, a- a- Anthropic credit, right? One thing in politics I often notice is that good intentions bite you in the ass more than evil deeds. Machiavelli explained it years ago, right? They were trying to do the right thing, and then they got caught in this buzzsaw of, "We've warned this is dangerous." But now that you have an instance of it, people are coming at me. And the, the real truth is they're trying to say, "It's dangerous, but we're the good guys, so trust us." And I think the stepping back comment is this: Private citizens don't get to run around and say, "This is really dangerous. This could be awful. This could cause world damage. Oh, but by the way, we're the arbiters of the decision-making." You've made yourself part of the political process, because you claim to have invented the most dangerous thing since the atomic bomb. You are ipso facto political, and you better get good at politics really fucking quick.

   11. EREverett Randle37:05

       Zooming out, I think the m- the most important thing in all this is that we step back from, from Anthropic and realize that the models are at a place now, whether they're from Anthropic or, or elsewhere, where they can autonomously find and chain multiple vulnerabilities together and orchestrate an attack autonomously. And if it's good enough, if they do get good enough where they can take down parts of digital infrastructure that run the US economy or Western economies, then there is some concern about a nation state adversary like a North Korea or China or Russia having those capabilities. So, so it, it's a real conversation that we're gonna have to have as, as, you know, Western democracy very, very soon, and we probably should have already had it, and Anthropic right now is, is the poster boy for it because of all, you know, because of the relationship that they have with the administration and the things that they've said. But it, it just... Like this is, this is true for AI now. This is an AI discussion. This isn't an Anthropic discussion anymore.

   12. HSHarry Stebbings38:02

       Yeah. Yes or no before we move on. Do you think they'll go public this year?

9. 38:09 – 56:00

   Will Anthropic IPO This Year?

   1. RORory O’Driscoll38:09

      I mean, statistically more than 50%, yes. Because I- i- in a world where this doesn't get solved for four or five months, there's so many other problems that our heads are gonna hurt, right? So you have to say if a problem... In, in a world where it does get solved, then yes, they should go public. So still statistic- But, uh, obviously if you're... Again, I'm just a simple Bayesian. If your prior before was 90%-plus, which it should be, because we've just seen a stellar reception to a company that even though it was called SpaceX, turns out that the vast majority of the market is in AI, at least how they s- frame it now. You've just seen that have a stellar re- reception. You'd be pretty much of an idiot to see SpaceX trading 2.4 trillion and to say as the board of Anthropic, "Let's hold on for a better market." Of course... So your prior should be they're going public with a 90% certainty. So this thing is a, is a significant wrinkle, and they're gonna have to work through it, so it lowers the probability, but still way more than 50%, because they'd be crazy not to.

   2. EREverett Randle39:03

      I, I think they will. I, I, I think they will. I think, um, one, one last final anecdote and then we can move on is, um... I mean, obviously they're already on this incredible trajectory that's been reported on publicly. We had a founder in our portfolio when Fable was, was active, um, that after a day of using it said, "With Fable, they're not gonna get, they're not gonna hit 100 billion of ARR this year. They're gonna hit 150 to 200." Um, and that was just an anecdote from their personal use, because the, they thought that the model was so unbelievably powerful and they were gonna spend that much more money on it. Um, but so I think with or without Fable, we'll see. But I, I think i- it's about as good of a market as any to, for them to go out, and I think the results and the numbers are just gonna be eye-watering.

   3. HSHarry Stebbings39:44

      I'm, I'm sorry, Ev. While you were explaining how America has changed, you know, model capabilities, um, Rory was struggling with European inventions of bottle caps on-

   4. EREverett Randle39:53

      No, I-

   5. HSHarry Stebbings39:53

      ... water bottles. [laughs]

   6. RORory O’Driscoll39:55

      [laughs] I'm, I, I'm, I'm aware of the European nanny state. By the way, it's really sweet that as an English person you call it Europe given that you're not in Europe anymore. But I know what you meant. I know what you meant.

   7. HSHarry Stebbings40:03

      Well, speaking of Europe and to, to the point that you said there, sovereignty, and Mistral investors rubbing their hands with glee at the potential. Um, Mistral in talks to raise three billion at $20 billion. They've actually been incredible in terms of their kind of FD model. They've scaled to over half a billion dollars, some of the biggest enterprises in Europe. Will we have many more sovereign models like Mistral? How do you read this?

   8. RORory O’Driscoll40:28

      I think you'll have a push for it, right? And look, even if... I mean, this is gonna sound awful, but even if the European alternative isn't as good as the US alternative, there will be cases where, you know, it's not quite good enough, but you don't have the sovereignty risk. So yes, uh, I gu- I told you-

   9. EREverett Randle40:44

      That's just the whole European economy, isn't it?

   10. RORory O’Driscoll40:46

       I, I, Ev, I was just-

   11. HSHarry Stebbings40:47

       [laughs]

   12. RORory O’Driscoll40:47

       I was, I was, I was getting up for doing the swing. I was gonna build up to it, but you just took the words right-

   13. HSHarry Stebbings40:52

       Ev, come on dude. [laughs]

   14. RORory O’Driscoll40:53

       You took the words right out of my mouth. Absolutely. No, I mean, you know, I, that's... Yeah, I mean, the, the, there are areas where they're actually genuinely better, but there are a lot of ar- Like, you know, you guys are still try- Europe is still trying to do some kind of GPS alternative. The more we exert our sovereignty in the United States and do this kind of thing, the more important it will be to decouple from it, right? And, you know, would you prefer to have the second-best model that you have access to or the best model that gets cut off once every six months on a random basis? So I mean, I think you're seeing it in defense procurement.

   15. HSHarry Stebbings41:27

       Mm.

   16. RORory O’Driscoll41:27

       You know, there's still stuff that you can only get from the United States, but more and more if you can get it from either place, you'll do it. And I think Europe, to the extent it perceives, you know, this import at the margin... I mean, it'll be the, it'll be the classic European thing. They'll care enough to fund it, but we won't be able to care enough to find the $100 billion to compete with it.

   17. HSHarry Stebbings41:46

       Mm.

   18. RORory O’Driscoll41:46

       Right? So, uh, good, good for Mistral. They're in a good, they're in a good place. I find it plausible that that would continue even if it's not the best model.

   19. HSHarry Stebbings41:56

       Ev-

   20. EREverett Randle41:56

       Do you think-

   21. HSHarry Stebbings41:57

       You think Gr- You think Greece is gonna come out with a model soon?

   22. EREverett Randle42:00

       They should. Well, I, I, I don't know, I don't know if this is real or not, but I saw on the timeline that Rio, like the city of Rio, uh, in Brazil apparently, um, you know, like post-trained or fine-tuned, um, a, a Chinese open source model and created like their own city, city model, and it was like Near Frontier or something. I... That m- That might have been fake, but that's, that's what I saw. Um, I, I do think like... I, I think, again, I think everyone, um... It, it's sort of like what we've seen in defense where as, you know, especially in, in Western Europe, you've seen so many, um, so, so many startups pop up to kind of, um... And, and, and I'm sure incumbents as well to, um, to, to respond to the demand From Western European countries to have their own defense supply chain and manufacturing and all these things. I think we'll see the exact same thing where sovereignty will become much more important. The thing that's really struck me is just how hard it is to actually build these models and have them actually be good. If you think about it, like, outside of China, there are basically no good open source models. Um, like Nvidia has one that's pretty solid now with Nemotron, but like, the like, where, where are the US open source models? You know? Where, where... Like, you know, we don't even have any in the US now. And so I, I do think just the talent, the capital, and the focus over a long period of time to actually be best in class at pre-training, at mid-training, at post-training, doing everything that you've done, or, or doing everything that you need to do in order to create a really, really good model is an extremely scarce skill set, and I don't think it's this fungible thing where everyone's going to, um, going to, to be able to do it. And so I don't know if that leaves people where, you know, may- maybe the, the answer for a lot of countries is to just take an open weights model from China or elsewhere and, and, you know, post-train or fine-tune their own version and have that be, be what they start from. Um, but I, I also don't think that it's just this trivial thing that every country or every region can have a, um, can, can have their own sovereign player because most of the sovereign players, they've fallen far behind. Like Mistral itself has fallen very far behind on, on the actual model side, and they've done a very good job becoming this inference platform and the FDE model and all these things, but they're, they're, they're ha- outside of China, there's, there's been no great, you know, model, model producers besides people at the frontier.

   23. RORory O’Driscoll44:13

       Yeah, and I think, you know, uh, 'cause there, there's, there's the question of, yeah, people beyond, uh, op- um, Open Labs and Anthropic. That, that's one dimension, and then the other dimension is open source, closed source. And it's worth pointing out many of the closed source vendors are... sorry, the open source vendors are morphing more towards a closed source model. I mean, obviously Facebook, uh, Meta did that. To some extent, I think some of the Chinese vendors are starting to do that. So, A, it's, as Ev says, it's extraordinarily expensive and extraordinarily difficult to do, and if you can't command a return at the end of it, you know, what's it all for? It can be done, but it's a brutal struggle, and I don't know if the critical mass of knowledge for a compute task like that exists in Europe in a way that obviously aerospace was a core competency in Europe. So I agree. I think you'll have a whole bunch of people flailing around and trying. It'll make, um, Nvidia ever so happy 'cause everyone will feel the need to have their own chips. But fast-forward five years, you know, unless the US is ludicrously obnoxious in its sovereignty acts, it should be a small oligopoly like cloud of two to three players in the US with some kind of affiliations in Europe, and that sh- is the likely structure of the industry, just given the costs required to play.

   24. HSHarry Stebbings45:27

       Ev, uh, direct question, and, and the latest Benchmark fund is gonna be the, I think, one of the best Benchmark funds in history. Like, like truly astonishing. But I, I had a GP from USV on the show earlier, and I said, "How do you evaluate not being in one of the model providers as USV?" And, and can I ask you the same, which is, like, Benchmark, one of the best firms, do you sit and think, "It's not our game. They're too large"? How do you guys sit around the table and reflect on not being early in a model provider?

   25. EREverett Randle46:00

       No, it fucking sucks. It's terrible.

   26. HSHarry Stebbings46:02

       [laughs]

   27. EREverett Randle46:02

       It's a, it's a, it's a complete and utter failure, um, on, on our part, and I think we, we'd all say that. And I, we, you know, we had, we had dinner with, with, um, some of the leadership of another, you know, one, one of the other best funds in the Valley that, that also, y- they, they have a large position now, but they weren't early in, in, um, in the, in the model providers either. Um, a- and you, you, you can't, you can't be in a situation where you have a chance to make, you know, a 30X on scaled capital as... You know, if, if you think, if, if you want, if you want to claim that you're one of the best firms in the Valley and you have a chance to make a 30X on scaled capital in four or five years and you don't do that, that's always a failure. So that's always, it's always a failure, um, no matter the way you cut it. So it sucks. So it's, it's great that we did a bunch of other amazing investments. These were all obviously before my time, but you know, the Sierras and Fireworks and Legoras and Mercors and Langchains and Hagens of the world in that fund, um, and, and many others as well. And so the, the funds look, the funds obviously, um, look, look awesome, but, um, but no, it, it stings, especially when all your friends y- y- you know, are, are, uh, are sending you their implied look-through ownership of, of Anthropic and OpenAI and SpaceX. You, uh, you know, those are eye-watering numbers, and so it's, it's a, it's, um, it, it definitely stings.

   28. HSHarry Stebbings47:20

       Ev, my friend, I've told you before, attribution is everything in venture. Those great deals were all you, okay?

   29. EREverett Randle47:26

       [laughs]

   30. HSHarry Stebbings47:26

       The amount, the amount of Excel GP-
10. 56:00 – 58:24

    Wix Slashes Guidance & Cuts 1,000 Jobs

    1. RORory O’Driscoll56:00

       Speaking of turnarounds, that was a turnaround that's been very successful. Speaking of potentially in need of something changing, you mentioned the word Wix. Wix slashes 2026 guidance, cutting 20% of staff, 1,000 employees, and cuts outlook by 50 million and revenue by 25 million. What, what the fuck do we do here? I really like the Wix team. They're really good people. What do we do here? It's hard 'cause, you know, you look at... L- let me just make the, the, the pro case. By the way, for some reason now this is one of the two or three corps... Uh, not corps. That's not fair. This is one of the two or three businesses we seem to love to dissect, especially when Jason's here 'cause he's really good at the Replit lovable versus Wix discussion. So I feel we give Wix more air to cover than... Uh, uh, more grief than perhaps they deserve. But I think the two things here... I mean, so but there are some things relevant here. One is they did do the right thing. They made an acquisition, right? The acquisition is growing nicely. It's kind of that next generational website. It was at 100 million, but of course Lovable and Replit are 4 or 500 million, and Ev just decided to walk away. He's not interested in Wix. Um, but the thing-

    2. EREverett Randle57:02

       Jumping in the window

    3. RORory O’Driscoll57:03

       ... I'm giving you shit, man. It's a good idea 'cause it's pretty toasty in here. Um, but, um, the interesting thing is the stock is now... I mean, they did a buyback. Yeah, they did... I think the acquisition was a great idea. The buyback was obviously a bad idea because the stock is now I think well below half that point and trading at one times revenues. To Ev's point, if you don't have a compelling AI story, it's just really hard, right? And they're just... I, i- in my view, they're just going through the same journey Intercom went. I don't think it's done, and I think at one times revenue, I actually made a mental note to think, "What would I have to do to buy at that price?" Right? You know, the, the, the bear case is that the website building mark... Yeah, that what they did is gonna become a subset of the Lovable, Replit, just build me a website, and they just don't have any relevance there because they're just not far enough along. If they can create any kind of leverage, then at one times revenue [laughs] that's actually a very cheap stock. But obviously, you know, it's hard to say things are great when you guide down, you bought your stock back, and now it's at half the price. So it's been a tough period. I'm not... I wouldn't be giving up, but I'd say this is, this is one where, to Ev's point earlier, there's clearly a relevant AI story, and not only is there a relevant AI story, there probably isn't a story without AI. They are definitely in the burn the boats because you're stuck in Persia, um, because you gotta figure this out. Um, so, uh, there should be a doable answer, but, you know.

11. 58:24 – 1:08:21

    SaaS Winners vs Losers: The Good/Bad AI Framework

    1. EREverett Randle58:24

       Yeah, I think, um, the, the SaaS market in general obviously took a huge bath, um, sort of over the last year, and, um, until recently, y- you know, every, you know, every SaaS stock was getting, you know, cut in half, cut by 60%, cut by 40%. It seemed very indiscriminate, and I think the, the nice thing that's happened for the SaaS market in, in the public markets for software over the last, like, three to four months is that there's now at least a filtering mechanism, and it seems like there are, like, clear scales. And so you... I mean, you, you have, I mean, Palo Alto, CrowdStrike, Cloudflare, Datadog, Palantir. All of these companies trade above 15 times NTM revenue again. And so the, the premium stuff is, is still being priced at a premium, um, which is nice because for a while it just seemed like, wow, like, I guess this, this category's just dead. And I think now there's, like, very clear, um, kind of scales at, at play, and we, we are actually going through with, with the founder and, and trying to, like, actually collate, well, like, what do, what do public markets want to see? Like, what are the good attributes and bad attributes of any of these companies that impact their multiple? And so, um, I, I'll, I'll just go through a few of them. So, um, on the good side, I think, you know, all these companies that are trading well, they have a usage-based, uh, component that scales in relation to tokens or correlated AI. So, like, Datadog and, and Snowflake are good examples of this. Um, two is that there's a clear AI tailwind for the use case, so cybersecurity is, is, is a big one there. And then, um, a third good one is that you can actually leverage AI to accelerate share gains. So maybe you're not the market leader or you're a faster-growing smaller company, and if you are a newer company that has better AI than an incumbent, you can actually accelerate your share gains, um, share gains from that. On the bad side of, of the, of the T chart you have, um, the first one obviously is perceived business model exposure. So, like, you're a per-seat model, um, in an economy going down this AI outcome-oriented or token-oriented business model path. Two, you have an easily replicable product or easily replicable value with coding agents or something else like it, so this is obviously where, where Wix gets really dinged. Obviously Intuit as well. I think a ton of Intuit's, um, profit comes from TurboTax, and everyone's scared now that TurboTax, um, you know, will become, um, extremely easily replicable in the future. Um, a third bad thing is, like, if you're already the market leader and you only have share to lose, um, if you're already the incumbent, that's a really tough place to be because there's not really share to gain. There's only share to lose, and AI is a great way to, um, for, for startups to take share from you. And then the fourth is, like, if your product's just lame- Um, ultimately, like enterprises have an IT budget. There's like a, there's a pie that equates to 100%. And right now, let's say on average, you know, that pie is 10% AI spend and 90% IT s- uh, other like software. People probably wanna make it more weighted to AI, like maybe even 50/50. So that 40% from 90% down to 50%, it's gotta come from anywhere so... Or from somewhere. So if you just have like a lame product that was already on the fring- fringes anyway, um, you're gonna get dinged and your, your retention's gonna go way down. So if, if we take like Wix through this, and, and I think like, you know, everyone has some mix of these good and bad, and what the market is doing is weighing, do I believe the bad is worse than the good or the good is better than the, than the bad? And right now for Wix, it's like massive replicability problem. Um, they were an incumbent that perceived, uh, that is perceived to, to have a lot of share to lose to AI, um, players that are, that are, um, coming up in, in the market. Um, it- some of their, um, product does have business model exposure. Uh, and like the product is, y- you know, at least relative to the product experiences of Lovable and others is sort of lame. On the good side, you know, they, they did the base acquisition. They have a usage-based component. There's an AI tailwind for what they're doing. But the net of the scale is people are seeing way, way, like there's way more credence to the bad than there is the good. And I think, y- you know, it's like Figma also has Figma Make, like they're also in this market. They have this usage-based component now, and people aren't giving them credit for that because they think the bad outweighs the good right now.

    2. HSHarry Stebbings1:02:21

       You're essentially getting the core business for free though. When you look at Base44 at 150 million of ARR and you look at them-

    3. EREverett Randle1:02:26

       Well-

    4. HSHarry Stebbings1:02:27

       ... Rapid being priced at 10 billion, you're pricing Base44-

    5. RORory O’Driscoll1:02:29

       Yeah. But if you... A- a- agreed. But if you step back on that, I would... If you sp- and, and the two, the two of you, what Ev outlined was your pros and cons of a business, right? And he didn't mention the stock price once. He basically said, "These are good things to have. These are bad things to have," right? And then what happens is the stock market is basically taking all that and saying, "Okay, you've got six good things and only two bad things. I'm gonna give you 10 times." And Wix, you've got two good things and six bad things, they're gonna give you one times, right? And what it means is at one times versus 10 times, at a shitty enough price you can be a value investor. We'll talk about Adobe in a second. And, you know, maybe at 1X, Wix is now priced to the point where the ex- um, logically the expected return at that point should be equivalent to the 10X, right? But what it, what you're saying is exactly. You, you- if you're on the bad side of the T accounts, the only forcing function left is price. And, you know, price has its wicked way, right?

    6. HSHarry Stebbings1:03:21

       I'm so- if you're a founder, price has its wicked way.

    7. RORory O’Driscoll1:03:24

       Yeah.

    8. HSHarry Stebbings1:03:25

       I'd be lining up the debt providers and the financers to take this private.

    9. RORory O’Driscoll1:03:29

       I'm telling you, if I was-

    10. HSHarry Stebbings1:03:30

        This is I'm going private on this one

    11. RORory O’Driscoll1:03:31

        ... could not agree more. If I was running Wix, right? Instead of having wasted that buyba- like actually maybe the buyback... Yeah. Because you're right. If you're gonna break your pick for the next five years, and they're willing to give you the company at one times revenue now, well, screw it.

    12. EREverett Randle1:03:44

        Old Michael Dell. Find your Silver Lake, take the company private.

    13. RORory O’Driscoll1:03:47

        That's, that's exactly... That was the sweetheart deal of all sweetheart deals, which is why he's, you know, top 10 on the billionaire list. Exactly. If I'm gonna grind through, if the market doesn't like me, well, I like me, right?

    14. EREverett Randle1:03:56

        Yeah.

    15. RORory O’Driscoll1:03:56

        And I fancy my chances. So you're right, actually. Maybe we need to go do a PE roll-up of Wix with the boys and just call them and say, you know-

    16. EREverett Randle1:04:03

        [laughs]

    17. RORory O’Driscoll1:04:03

        'Cause at one times, you know, right?

    18. EREverett Randle1:04:05

        Don't give Harry ideas. I see him writing down a note. You're gonna-

    19. RORory O’Driscoll1:04:07

        Ah, entirely. Ah, he, he'll-

    20. EREverett Randle1:04:08

        He'll put him on his train here. [laughs]

    21. HSHarry Stebbings1:04:10

        Anyway, I'm excited to, I'm excited to have Avishai on the show in the next few weeks.

    22. RORory O’Driscoll1:04:13

        Yeah, good. And look, I mean, you're... Again, you know-

    23. HSHarry Stebbings1:04:15

        I actually am.

    24. EREverett Randle1:04:15

        It's true. [laughs]

    25. RORory O’Driscoll1:04:16

        And you certainly look. I mean, genuine, you know, I have a ton of empathy a- and respect for you. You know, you build this thing. I mean, just capitalism is built. You build this thing, you're doing a couple of billion dollars in revenue. The architectural crank turns, and suddenly you're, "Oh my God, I gotta do it again?" Right? And I have a ton of respect for anyone who says, "Damn it, I'll strap in and do it again."

    26. HSHarry Stebbings1:04:36

        Could be Wa- this could be Webflow. I mean, that's-

    27. RORory O’Driscoll1:04:40

        Well, I think the, the, the genuine comment on that is if you're going to be in this situation, I, I actually th- it's interesting, is that being private, late stage with a lot of venture and a high valuation, I'm not commenting on Webflow in particular, but you're right. That's even tougher, 'cause at least these guys have capital and are profitable. But yeah, it's, it's a haul if you're-

    28. HSHarry Stebbings1:04:57

        Or, or like Squ- oh, sorry, I'm not picking on Squarespace.

    29. RORory O’Driscoll1:05:00

        Yeah.

    30. HSHarry Stebbings1:05:00

        You haven't got the Base44 acquisition and just have the legacy business. I mean-
12. 1:08:21 – 1:11:20

    Why Hedge Funds Are Dumping SaaS for Semiconductors

    1. EREverett Randle1:08:21

       or I can go buy Nvidia for 16 times earnings." So, like, who, who's, like, the poster child of every single tailwind that we're talking about in AI that has, that has, you know s- you know, 80% share themselves of the most important piece of the Compute tech stack. And so the, the... What, what I've heard from a lot of, um, public managers is that SaaS, you know, maybe, maybe it's oversold, maybe it's too cheap, but why does it matter? Like, why- it's just too hard. Like, I, like, you can buy these memory stocks for very cheap multiples. You can buy... And, like, if you look... And, like, again, all these, all these guys, all these guys and gals in, in the hedge fund industry, they're graded on how they do versus the index. If you look at the SOXX index this year, I think it's up, like, 45 to 50% this year, whereas, like, the SaaS index is probably down 20 to 25%. So, like, even if you're an expert in SaaS and you got the good ones, you've underperformed the SEMI's index. And so their job isn't to be smart, you know, on SaaS. Their job is to make money, and they can just invest in SEMIs right now and play that trade.

    2. RORory O’Driscoll1:09:20

       Yeah. Let me give you that number for five years. I, 'cause I, I, I don't have it with me 'cause I'm on the road today, but when I do this show, I have my ETF list. And the five-year return from World Cloud, which is the definitive ETF for SaaS, is down 30%, and the five-year return from SEMI is, is 2.7X up. Right? It's just been a great trade every time. And there was one month where the S- there was one month just recently where the SaaS thing bounced off the bottom and outperformed for a month. And then even the last two weeks, it's do- it's kind of relative underperformer. And what the trend, these guys, the trend is your friend, is the first thing that momentum traders learn, right? And-

    3. EREverett Randle1:09:56

       So I th- I just pulled it up really quick, Rory-

    4. RORory O’Driscoll1:09:57

       Yeah

    5. EREverett Randle1:09:57

       ... just for, for-

    6. RORory O’Driscoll1:09:58

       Yeah

    7. EREverett Randle1:09:59

       ... since you mentioned it. So the, the Bessemer NASDAQ Emerging Cloud Index over the last five years is down 44%, and the, um, the SOXX, the iShares Semiconductor ETF is up 325%.

    8. RORory O’Driscoll1:10:11

       Yep. Yep.

    9. EREverett Randle1:10:12

       And so it's just like-

    10. RORory O’Driscoll1:10:13

        20X

    11. EREverett Randle1:10:14

        ... you just, like, y- y- you know, go short S- SaaS and go long SEMIs, and you've made, you know, better money than, than any other hedge fund manager in the world, you know, um, besides Leopold. [chuckles]

    12. RORory O’Driscoll1:10:24

        There's a point, though, where the trade makes sense.

    13. EREverett Randle1:10:26

        Yeah, there is. But I think that it has to be a c- I mean, the, it's one of t- So each of them, so you ask yourself, and I- I- I- i- in each case, you have to s- ask yourself, "What breaks that trend?" And it's a different thing for each of them. What breaks the trend on the semi-

    14. RORory O’Driscoll1:10:39

        What's the catalyst?

    15. EREverett Randle1:10:40

        Yeah. The catalyst on the semiconductor CapEx trend is a flattening out in CapEx. If that happens, then all bets are off and those things are gonna go down so fast and hard. So that's a call you can make. That's one call you can make. The, the trade on the S- the, the, the thing that breaks the catal- the catalyst on the software trend, I don't think there is a single thing. It's what I've said. There's a bunch of sorting going on 'cause this is the winnowing of the weak, right? It's never pretty, right? And the guys who are... Individuals get through the gates to be winners, right? And, you know, the Datadogs, you know, are, are kind of on the positive side and, you know, are starting to outperform. And for the others, for something like Adobe, I actually think there has to be an institutional catalyst before there can be a pricing catalyst,

13. 1:11:20 – 1:16:47

    Adobe Beats, Raises, and Still Crashes as AI Fears Intensify

    1. EREverett Randle1:11:20

       right? In other words-

    2. RORory O’Driscoll1:11:21

       Wh- what is that?

    3. EREverett Randle1:11:22

       I think you need someone like Owen running it. 'Cause I'll tell you, and now I'm gonna vent. The one thing I, I, and I don't like to pick on, you know, amazing companies. I mean, I was around when, you know, one of my partners years ago was early at Adobe, and she recounted, you know, if, to your point, they had 90% market share and were stuck just below a billion dollars for four years, right? Turns out when you sold everyone PDF, you know, this, you, you get stuck, right? My vent on that company is they have milked their users for so long that it just feels a piece of financial engineering. Interacting with the product, it's always the login are crap. The... I don't even understand the licensing model. I don't even know what I can use and not use. It's just constant... So my sense is they've ext- it's like being a PE-owned without ever being PE-owned. They've extracted every piece of value, right? And someone's gonna have to go in there and kinda rethink through what it takes to make their users love them, right? And that's a bit... And until they do that, I'm gonna have maybe a 10 times or 8 times it's cheap. In shock horror it might go to 10 times, so you might get a one-off 20% pop. But to Ev's point, it's just a lot easier to own Nvidia and have it go up 30%, right? 'Cause there's no second tier. Rory, are you, are you calling your shot as, as Owen as the next Salesforce CEO? Is that where, is that where you're going? [chuckles]

    4. RORory O’Driscoll1:12:32

       You know, it's been my experience, having sold companies to Adobe and Salesforce and many things like that, I've sold companies in the past where I kinda go, "That person could easily run the acquired company." Most times it doesn't happen. It's just too hard. You've been your own boss. It's too long. I mean, you know, Brett T- you guys know so much better than me. Brett Taylor could comfortably have run Salesforce, but it turns out there's someone running Salesforce who appears to like running Salesforce and has done it well enough to keep running Salesforce. So no, for that reason alone, no. Right.

    5. EREverett Randle1:13:00

       I just, I just know Benioff. There's no fucking way he's leaving.

    6. RORory O’Driscoll1:13:03

       Yeah.

    7. EREverett Randle1:13:04

       Unless he's 80 and 90-

    8. RORory O’Driscoll1:13:05

       No

    9. EREverett Randle1:13:05

       ... and in a coffin.

    10. RORory O’Driscoll1:13:06

        That's-

    11. EREverett Randle1:13:06

        Like, this guy is loving life-

    12. RORory O’Driscoll1:13:07

        Yeah

    13. EREverett Randle1:13:08

        ... more than ever in the AI world. True. Very true

    14. RORory O’Driscoll1:13:10

        But yeah, but as, as a comment, I don't know if you see this, Ev, but we're saying it, kind of taking that founder... One of the smarter things we've seen some of our good companies do is some small acquisitions, building in founder teams.

    15. EREverett Randle1:13:21

        Mm.

    16. RORory O’Driscoll1:13:21

        And one of the best ways to Ev's point, to do a little bit of a culture change can be picking up some of these founder-led early AI companies, and you should be doing that. And so I, I think that is one way to get really good talent, and I'm thinking of one of my companies. A really well-run company. It's an AI company, but kind of pre-gen AI. They've done a magnificent job of hiring, doing two or three small acquisitions, and you fast-forward a year and each of those guys is running a $20 million BU, and you're like, "Wow, that works." Going back to the Adobe comment, I don't know how they do it, but they need to f- it's not more of the same. If they hire more financial engineering, you need to just shoot it in the head, right?

    17. EREverett Randle1:13:55

        Yeah.

    18. RORory O’Driscoll1:13:55

        They need to hire someone who says, "I know where this thing needs to go."

    19. EREverett Randle1:13:58

        I think, I think this is where, th-this is, you know, this will be my la-last comment on this, but, like, th-this is where it gets really... where this cycle becomes really insidious to these incumbents-

    20. RORory O’Driscoll1:14:06

        Yes

    21. EREverett Randle1:14:06

        ... because I think in 2021, y-you know, when, when your currency as, as stock is worth so much, you can do a lot of these really ambitious product acquisition things. So, like, you can have the Square/Cash App deal, um, and it's fine because even though the stock goes down later, it's like, well, at least you paid for expensive stock with ex-expensive stock.

    22. RORory O’Driscoll1:14:26

        Good stock.

    23. EREverett Randle1:14:26

        The really tough thing is when the, like, pair trade where, like-

    24. RORory O’Driscoll1:14:30

        Is shifted

    25. EREverett Randle1:14:30

        ... all of the AI valuations or, like, are going, like, thermonuclear up and your valuation's getting cut by 60%, you're cur- you can't buy anything. Like, you, you wanna know what would make Adobe's stock price go down another 30%? If they paid, like, $15 billion, you know, if they paid a quarter of their equity for some AI company that, like, public shareholders might not even wanna buy. And so all of the, the really amazing AI t- Like, well, they should have done this two years ago. They should have robbed the cradle of, of all these AI companies before and, and overpaid when they were still, like, Seed or A companies and, like, gotten an AI product suite. They failed to do any of that. And, you know, of course, they sort of attempted with, with Figma, even though that, that was kind of before-

    26. RORory O’Driscoll1:15:08

        Pre-AI, yeah

    27. EREverett Randle1:15:09

        ... the whole AI wave. It was, it was pre-ChatGPT. But, like, they failed to do any of that, and now every single good AI company that they could acquire is too big for them to acquire, and they can't do it. And so then it's like, well, we don't have the talent internally. We can't build a good AI product portfolio because we don't have the talent. The talent's very scarce. And now we can't acquire the talent because all those teams are too expensive. I'm not... You know, you can't pay $10 billion for a company that's at 200 of ARR because it'll just nuke your stock price more.

    28. RORory O’Driscoll1:15:33

        Which is why, going back to the first principles, the, the people running the company have to be good enough and close enough to the metal to themselves know what to do, at least well enough to, you know, have a product vision, hire people, as I say, maybe smaller acquisitions. If you're trying to do it from a McKinsey management perspective, you're doomed because you don't... I, I, like you said, you've no buttons left to press. You had buttons to press in '21. Now you don't have buttons because the kind of things that you can do... Like, obviously if Adobe bought Runway or Higgs Field or pick your beloved thing, great. But I think you're right. They'd have a shit fit 'cause the, the only people who own this damn stock now own it because it's trading at eight times cashflow. And if you tell them, "I've taken you eight times cashflow and now we're trading at 47 times cashflow 'cause we've just spent all the cashflow and bought this loss-making thing," they're going to have a conniption. So you just... You're right. You can't do that. It's, it's a tough place to be. So you've gotta fix... It's why, again, I want the Wix guys to make it, 'cause they strike me as knowing what they have to do. They did the small acquisition. Just gotta grind it through.

    29. HSHarry Stebbings1:16:30

        They also have a phenomenal acquisition machine.

    30. RORory O’Driscoll1:16:32

        Yes.
14. 1:16:47 – 1:27:59

    Standard Bots' $200M Raise

    1. RORory O’Driscoll1:16:47

       I, I'll pick on just one random one. I think I saw Standard Bots raise-

    2. HSHarry Stebbings1:16:50

       Yeah

    3. RORory O’Driscoll1:16:50

       ... which is a company who raised $200 million. I do a lot in robotics, and I just like those guys. I mean, I spent a lot of time talking about it. I had some- talking with them. I had something else going on. And they wrote a really good piece that Packy McCormick published, right? And it was just very... And what I liked about it, it was very... It was calling a shot against humanoids. It was basically saying for a lot of these use cases, um, the humanoid is a mistake because a humanoid is putting a whole bunch of money into legs, which maybe you don't need for most industrial manufacturing. And what these guys do is they built a next-generation robotic... The bet they're making is... 'Cause then on the other... So you have the humanoids, which is a lot, and then robotic foundation models. You're putting a lot of money into a lot of enabling technology that maybe you don't need for the task at hand. In fact, you're o- you're over-scoping it. And then on the other extreme, you have the old-school robotic arm manufacturers who are doing all of the current, you know, robotic work, and most of them are... most software vendors are using one of those old arms. And they're basically saying there's somewhere in the middle you can build this next generation kinda... The, the analogy they use, like, you know, you have an integrated hardware-software stack like Apple. They're gonna do the same thing for a robotic arm that can be integrated, and they think they can take a lot of revenue that way. And it's, it's... What I liked about it, it's a pragmatic play, right? I think so. I think the, the, the, the case they articulated against humanoids in the short term I think is more correct than not, which is it's overkill for many industrial practices. And what you really want is a thing that can see, that can pick things out, can flexibly be trained very quickly, and then can pick things up and kind of do discrete tasks pretty efficiently. So I think it's a good play. It's a US-based arm manufacturer, and we don't have many of those at scale. So I, I like, I like that deal. I like that team, and I wish them luck.

    4. HSHarry Stebbings1:18:33

       Ev, do Benchmark have a robotics investment?

    5. EREverett Randle1:18:36

       We do. So, uh, my partner Eric Vishria led the Series A of Sunday Robotics, which is, um, I would say pseudo-humanoid. It doesn't have legs. Um, you, you, um, Sunday AI I think is the URL, but it has sort of like a platform that can help it go up and down, and then it ha- it almost looks like Ness from, uh, Super Smash Bros. It has, like, a little hat and these long arms. So it's, it's, like, pseudo, pseudo-humanoid for, um, [lip smack] for the home. But, but I mean, but Rory, Rory's totally correct that I think the two opposing views on, on robotics writ large, at least on, on the vector of humanoid or non-humanoid, the, the pro-humanoid argument is, like, the world is human-shaped. Like everything, our homes, our factories, you know, our workplaces, they're shaped for humans. Like, we've built them so that humans can use them, and so therefore, humanoid robots are gonna be able to most naturally, um, interact with the environment because the interact- the environment is designed to be interacted with by humans. Um, the pro-non-humanoid argument is a lot of what Rory articulated, which is like, look, like, you know, robots are expensive. Like, these parts are expensive. Like actuators, which are like the engines that, that, you know, provide torque for a robot, they're expensive. And so, like, for a lot of these use cases, um, depending on what the use case is, you know, it might be gratuitous or vain for there to be, like, these expensive legs that are, like, running around when you can just wheel the thing up to, you know, a, a warehouse where you're doing pick and pack for logistics, which is like, you know, you take the packages and you sort them. Um, like, you might as well just have the arm that's doing that. And so I, I think there's like-- I think in general, we as a team, I mean, we talk about robotics a lot. A, a thought experiment I love asking people is like, "In the year 2060," so take like a really far out view to, to allow the supply chain to catch up, "how many-- um, what, what will the ratio be of digital agents to humans? And then what will the ratio be of physical agents like robots to humans?" And I think there's, like, some argument that by 2060 in, in some places like the US, I mean, obviously, like, you know that there is no ceiling to the digital agents one. Maybe it's like ten thousand or maybe even more, depending on, um, how the, how the agent landscape evolves. But by 2060, you might have like a one-to-one ratio of, of like useful robots to humans, or it could be like much greater, um, depending on, on how, how fast we go. So obviously it, it takes longer to scale robotics AI than it does ChatGPT because, you know, a, a robot is not accessed through a website or an API. Um, you have to buy it and put it together and put it to use. Um, but I think that we, we, we think it's i-in terms of the things that are still in the first inning that are gonna become, um, trillion-dollar, you know, where, where trillion-dollar companies are gonna be produced, we think it's one of the prime candidates for sure.

    6. RORory O’Driscoll1:21:09

       It's interesting 'cause look, I, I'm, I'm on the board of Locus Robotics. We have fifteen thousand robots in the field. We do kind of a hundred and eighty million dollars a year, right? And but it's stunning how long it all takes. I mean, you know, I remind people the, the total number of robots in the world today, like in total, about three million, right? And there's, you know, there's a billion people doing real work, right? So, you know, a-and those robots have been around f- yeah, arms have been around for twenty or thirty years. So in twenty or thirty years, we've replaced less than one percent of the humans doing manual work. It's a long journey, right? And what you see is when you get to the front line where these companies are actually com- kind of trying to do the work, you just see frankly how good humans are, right? How flexible they are, how for anything other than very repetitive tasks with a large kind of thing, you know, the human, the, the ec- And, and the buyers... The funny thing is, I think i-i-in software industries, the buyer is typically a buyer of a industry that themselves has high gross margins, and they're kind of a bit loosey-goosey on the ROI. Yeah, you're, you're a high-value knowledge worker. We paying you three hundred grand. Fuck it. We can give you a twenty grand piece of software. Let me tell you, when you're running a warehouse with four hundred employees, each of whom is getting paid minimum wage for pick and place, right? You know to the penny how much labor costs, and if the robot doesn't cost half that, you're not gonna switch, right? So it just takes longer to adopt. I do agree. I love the long-term. That's why we did four or five, four or five different robot deals. I love the long-term trend, but I've been sobered. I, I've internalized that it's a long journey, right? Not a-- I don't think there will be... I could be wrong on this, and so I'm interested in Standard Bots. I mean, those guys made a very compelling, "Oh, this could be the sweet spot." I'm, I'm watching it like a hawk, right? But where you go from what we have now, which is steady adoption, to some kind of takeoff. I mean, Optimus, Tesla's at bat. If Optimus takes off, for example, that could be it. But you know, what you see when you get out on the factory floor is finickity little stuff that you wouldn't think takes time and kind of g- you go, "Ooh, that was the issue. I didn't have that in my investment memo." I have one robot company now where the biggest impediment to them getting a very large order is something to do with when the poly bags aren't flat, the label reader can't read the barcode, so the whole thing goes pear-shaped 'cause you have to have people smoothing them out, and if you need people to smooth them out, you don't need the robot. I'm like, "Wow, I did not have that in my memo."

    7. EREverett Randle1:23:31

       [laughs]

    8. RORory O’Driscoll1:23:31

       The whole poly, the poly bag problem. Hmm. Who knew? Right.

    9. EREverett Randle1:23:34

       The poly bag problem.

    10. RORory O’Driscoll1:23:35

        But you know, but it's, yeah. So lo-love the trend and love the deals in it. But yeah, I thought Standard Bots is super interesting. I, I hope those guys make it. I, I really liked it. It was a well-designed US ma- I mean, the arms today, going back to sovereignty, uh, the big manufacturers is German, the Japanese, and the Chinese. I think it's a good thing if we can make a hardware arm in the US that's cost-effective and probably will be less, not as cheap as the Chinese robot, but probably have way better software.

    11. EREverett Randle1:24:03

        Hmm.

    12. RORory O’Driscoll1:24:03

        And that's, that's a credible bet. 'Cause I mean, I know I'm riffing on this now, but someone did a translation. The Unitree, there's a humanoid company, Unitree-

    13. EREverett Randle1:24:11

        Hmm

    14. RORory O’Driscoll1:24:11

        ... that's going public in China, and it's doing five hundred million. It's a real company, and it's profitable. Most of the humanoids are still being used for demonstration purposes, but there is a trend there where you go, you gotta keep an eye on that because that could be where I'm like, you see that takeoff and I'm wrong. So I'm just, you know, watching that space.

    15. EREverett Randle1:24:27

        There's a blog post I really love, um, or it's like a little mini essay called Reality Has a Surprising Amount of Detail. Um, and you know, it's, it's like ostensibly about like, you know, making a set of stairs, but the whole point is like in the real world, in the physical world, stuff is just really complex. [laughs] And like you think it's like, "Oh, I'm just gonna like nail the stairs up there," but then you kind of decompose all the steps, all the complexity, all the edge cases, um, and, and reality just has a surprising amount of detail. It's, it's a very complex place. So I, I think that, that, that is what makes us so excited about this moment in AI for robotics though because for the first time you can actually have an edge LLM on a robot, even though the LLMs have to be much, much, um, smaller than we'd have for, um, you know, you know, like a frontier, um, digital LLM because they have to be on the robot itself. Um, there's these really, really, y-you know, like you can-- like if you think about talking with-

    16. RORory O’Driscoll1:25:18

        You can do things

    17. EREverett Randle1:25:19

        You can do things and, and like, and, and they can be very versatile. Like, they can learn from very few examples that like, "Oh, I actually need to smooth out this label before it goes next..." Like it, it's, it's easy to, to begin teaching robots these things to be dynamic and to be as versatile, um, maybe not as a human for, for a little while, but, but probably a- approaching that asymptote, um, which makes these things generally useful versus... Historically they've been very useful for like an extremely brittle narrow scope of responsibilities.

    18. RORory O’Driscoll1:25:47

        Totally. No, I would... But we also put some money into Journalist recently, and I, I was just wowed by the demo. It's like wow, you can... 'Cause again, I, I have bigger robotics companies in the field and I know a- and Everett's exactly right. The good news is they can do stuff, you know, um, and they can do it at speed and scale. The bad news is it's very brittle. In other words, if the process drifts even slightly, there's a lot more programming. Well, the next generation of LL- uh, of, of AI- of software, which is LLM based, the bots are way more flexible. Instead of telling it what to do at the individual, you know, move your hand here, it's like put this thing in that thing, and it can figure it out. And wh- and the way you see that manifest itself in a demo is, you know, you, you do three of them in a row and then the next one you push it or you move it to a different place, and the machine stops and it thinks, like the raptors in Jurassic Park. And then it goes, "I see it," and it reaches over there and pick. And then you're like, "Ooh, that's a..." You know, that's when you see that is the brain working, and that is the future. So robot companies have great demos.

    19. EREverett Randle1:26:41

        Yeah, we took a, we took a field trip down to Sunday, um-

    20. RORory O’Driscoll1:26:44

        Yeah

    21. EREverett Randle1:26:44

        ... and the, the, you know, in the basement all the robots, they, they were all folding laundry on beds. You know, it was like, you know, it was like a ton of these things all like fold- And, and like, uh, and, and the Sunday employees were purposely messing... You know, they'd be in the middle-

    22. RORory O’Driscoll1:26:56

        Yeah, yeah

    23. EREverett Randle1:26:56

        ... folding jeans, and they'd rip the jeans out and like crumple them and throw them. And the robot would sit there and be like, "Uh," you know. It was... And I almost started feeling bad for the robots. I'm like, "Let them, let them finish the laundry," you know? Like [laughs] I started to feel... 'Cause they have these cute faces. I was like, "Gosh, let them, let them finish folding the laundry." But they're very patient. That's the other beautiful thing about robots. They don't get mad. They don't skip work. They don't do it... You know? They just, they sit there and they just keep folding and, and that's gonna be-

    24. HSHarry Stebbings1:27:18

        In 90 minutes of hanging out with us we've turned him into Jason Lampkin. [laughs]

    25. RORory O’Driscoll1:27:22

        Yeah, yeah, yeah, yeah.

    26. HSHarry Stebbings1:27:23

        The agency-

    27. RORory O’Driscoll1:27:23

        They just, they don't complain.

    28. HSHarry Stebbings1:27:24

        They don't complain. [laughs]

    29. RORory O’Driscoll1:27:26

        They don't... J- Jason just wants normal people.

    30. HSHarry Stebbings1:27:27

        They're just easy. They just-

Episode duration: 1:28:09