I see a large IPO on the horizon. You're surrounded by journalists. What's your position?
Overnight success. The place 3000, right?
There's misinformation. Clearing order inbound. Let's just roll.
We are surrounded by general. Hold your position. Get up. Trust the experts here.
Five to founder Mel 5. I see multiple journalists on the horizon. Stand by.
UAV online. Double blaze. Triple blaze. Double kill. Finally, talk is please team deathmatch.
We are experts. Triple blades. That's just wrong. Right. Hardy clearing order inbound.
We are surrounded by journalists. Hold your position. Strike two. Activate golden retriever mode. Retreat.
Is win team deathmatch. We are experts. Triple blades. That's just wrong. Right.
Clearing order inbound. Get up. You're surrounded by journalists. Hold your position. Strike one. Strike two.
Activate golden retriever mode. Retreat. Trust. Market clearing order inbound. 5.
I see more and more journalists on the horizon. Stand by. Founder.
You're watching TVPN. Today is Thursday, July 30th, 2026. We are live from the TVN Ultra on the Temple of Technology, the fortress of finance, the capital of capital. Let me tell you about ramp.com. Time is money. Save both. Easy to use corporate cards, bill pay, accounting, and a whole lot more all in one place.
Absolute chaos on the timeline last night, this morning around Leopold Ashen Brener's hedge fund situational awareness. They have been forced to unwind their public stock portfolio after steep losses on AI infrastructure bets. CNBC reported this on Thursday today. Prime brokers reportedly rushed to raise cash to meet margin requirements while Ken Griffin's Citadel, my former employer, reached a deal to purchase the fund's publicly traded assets.
Did Ken ever do anything to you?
I was an intern. I would have loved to be getting him.
Did he ever try to maybe intentionally send the markets into turmoil just to test you?
No, but I mean honestly the story of Citadel is crazy. I mean after the housing crisis, the fund was down 50%. And it was a very very dark time. There were
Good thing for Ken though at that time.
Ken Griffin, what he would become, didn't exist and otherwise he probably would have eaten the young Ken Griffin alive.
Much like it seems that Griffin has done to Leopold.
The other frustrating thing is that they went down 50%. I think the next year they went up 50%. And then they were, you know, this is the classic, you know, explaining fund math to people. Oh, you're back up 50%. Great. You're back to where you were. Nope. You need to go back up 100% if you're down 50%. Of course. Anyway, so the news, the fund had built concentrated positions in AI infrastructure companies including Nebius, SanDisk, Micron, and Coreweave while also betting against software companies such as Adobe. Those trades have unraveled as AI infrastructure stocks plunged in recent weeks before rebounding sharply today.
Now, how much of the plunge is around shaken faith in AI's ability to deliver value, open source or just oil, inflation, the Fed's actions? We'll get into all of this because there's a lot of moving different, moving pieces that led us to where we are today. Let me continue giving the news, then we'll go through the timeline and discuss a little bit more in depth. So, also relevant here, this is from the TVPN newsletter. You can go sign up at tvpn.com. And what a bunch of people are pointing out on X is the fact that on Tuesday it was reported that Citadel expected a surprise rate hike from the Fed meeting that took place yesterday, which did coincide with more selloff in the market. So, the market has been selling off based on what might happen at the Fed. We reported on the Fed news. There were three Fed governors that said we should raise rates, but the rate held steady. But mortgage rates are high, are at over a one-year high at 6.66% today. Very odd number, but yesterday the Fed left the rate unchanged. And today, many of the stocks in Leopold's portfolio are up double digits and we'll sort of go through them. They're up today based on the news that Ken Griffin is buying the portfolio, but they are still down over the last month, for example, in many cases.
Ashen Brener, a former OpenAI researcher, rose to prominence after publishing his 2024 essay series situational awareness, the billion-dollar PDF, as Will Manitis put it, I believe, which argued that rapid AI progress would require an enormous buildout of chips, memory, power, and compute infrastructure. That thesis became the foundation of his investment strategy after launching the fund. He's also engaged to Anthropic CEO Dario Amade's chief of staff. The news is coming in hot and fast on the story. Here's a timeline of the most important headlines so far. So, Bloomberg 9:25 p.m. yesterday. I remember I think you texted me this as I was going to sleep. And we were like, whoa, this is big deal. I wonder how crazy this will get over the next few days. It got very crazy very quickly. So, Bloomberg reported Leopold Ashen Brener situational awareness seeks to raise capital after AI route. There was an article in the Financial Times as well last night just saying that, hey, there's some rumors that are leaking out from LPS that they got a letter saying like, hey, the market's down. Now's a good buying opportunity. The thesis is as strong as ever. If there was ever a time to put more money into this fund, now is the time. That can be good. You want to be buying when things buy low, sell high, right? But at the same time, if it's to cover margin calls, if it's because the fund's getting beat up, it's a little bit rougher of a pitch. Then at 6:05 a.m. CNBC announces that AI investor Leopold Ashenbrer has been forced to unwind all public stock positions after steep losses according to CNBC sources. And then the Wall Street Journal reports at 8:39 a.m. that Citadel has stepped in to buy situational awareness's stock portfolio after big losses in AI. And so
after living through FTX and SVB and now this on the timeline,
No, the key takeaway is like when leverage is involved, things just move so so so fast, right? You remember with FTX, there's kind of some rumblings, a couple posts from SPF saying like we're fine, it's all good, and then it was over. And then the same thing with SVB, like couple rumblings, maybe like a couple weeks, a week beforehand, few posts here and there, and then it just moves so so fast, right? And um yeah, quite a bit different than traditional venture world where when a company's dying, it dies over two, three years often, sometimes more. Right.
Yeah. Yeah. Yeah. Um, I'm thinking of like I mean we've had some of these companies on from the private markets where they've gone through big booms and busts like Bird and then they've built back and they've turned around but there's so much more
No talkingime lime turned around but it took like an extra five years for bird to actually wind down and it's because there's no leverage in the system. There's just a bunch of dollars that sit there as equity and those get burned down. But every month if the business is deteriorating, you're cutting costs, shrinking the business, tightening things up, making that 12 to 18 months last 24 months, and then you wind up, you know, 24 months in, you're like, "Oh, we're not going to be able to raise again. Let's stretch this again and rightsize the business again." And all of a sudden is so it takes like years and years for these things to unwind. Although they are correlated in the venture world, they can be decorrelated in the unwinding process and then there can be other things that are outweighing the portfolio. So every VC that had bird on their books probably also had some SpaceX on their books or something. And so there's this balancing effect and it takes years for these things to balance out and they can be unwound at different periods in the market as opposed to everything needing to happen all at once.
So speaking of public markets, let me tell you about public.com. Investing for those that take it seriously. They got stocks, options, bonds, crypto, treasuries, and more with great customer service. I like this post from Richard Crab, one of my favorite investors. He runs the quant hedge fund, Numero. And he says, I think it's cool that funds like situational awareness can exist in America and that there's a market for them, but the outcome was never about being right or wrong on AI at 150% V. variance drag alone is 113% a year and risk of ruin is roughly a coin flip over the fund's life. A child can do the math on a napkin. Claude did it for him. He says AI says ruin wasn't unlikely. It was roughly even money. So there's a 50% chance that the fund sees so many losses that they have to do this liquidation process. And that's basically what happened. And it must be so frustrating because this really does not feel like oh Leopold was wrong about AI and the AI buildout. It's like well there's oil and a war and interest rates and all these other things going on that are creating some jitters. And then also once the AI trade and the infrastructure trade got so big, you wind up with like this retail froth on top that makes things even crazier. And then all was benefiting from
benefiting from. But also, it's much harder to do sort of a first principles analysis on what the psychology of a frothy market will do as opposed to just retreating to okay, well, there's this the model progress is progressing like this and token pricing is you know, counting themselves will this particular stock become a meme stock, right?
Pull up this picture. There was a lot of this going on this morning.
The memes are flying. This is truly like
this was the first meme that popped into my head. If we can pull this up
the chat is asking for a non-sponsored sleep score to see how we slept
in the midst of a financial crisis. Um I've been doing pretty well. I got an 87 last night, 92.
Let's scroll up a little bit. There's a lot of this on the timeline this morning.
Yeah. people saying, "Oh, I don't know why the guy's head's cut off." You guys okay?
No, this was just a lot of people on the timeline being like, "I knew he would blow up."
the armchair experts are out in full force today. And in many ways we are among them. But
so I like to think about it like at least some of the more high-profile LPs.
Yeah. that in situational awareness. Yeah. A lot of them are like, you know, great founders. You know, these, you know, maybe they have big positions in the labs and all these different things. And it's quite possible that situational awareness at least when they invested was like 5% of their portfolio and they're just thinking like
go giga long go Yeah. Yeah.
For some of these people it might be like
one less than 1%, right? Whatever it is. And so it's actually somewhat pragmatic for them to just be like, "Yeah, go crazy. Do whatever you want."
Yeah. That's the product.
That's the product. That's what I want to buy. The rest of my portfolio is fine. You're going to have a lot of AI exposure whether you like it or not.
But yeah, it is actually crazy that it didn't even take a three-month draw down, right? Yeah.
Uh it was, you know, what was it? June 1st they were at 45 billion of AUM something like that was the NAB
end of June so beginning of July and then how quickly things can change and
uh you know poor Leopold already went through this with FTX
I believe he and the rest of the future FTX future fund team I believe resigned like right when
what's interesting is people are framing this as like they got pennies on the dollar or Ken Griffin bought the portfolio for pennies on the dollar. And when I think pennies on the dollar, I think like five pennies per dollar. So like 5% recovery, but it might be closer to like 50% of book value. Might be 80% of book value. I don't know. Um fortunately, we have Martin Skrey joining in just a minute. He's here in the waiting room.
crazy because when Martin was the first account that I saw
like this, he posted it before any mainstream media had picked up on it at all.
Um, and there was a bunch of comments on his post saying, "I have good sources that say this isn't true."
And of course, Skrey was right. There was trouble. And we'll bring him in now.
Let's bring in Martin Skrey to break it down for us. I believe he's here. How you doing, Martin? Good to see you again.
Hey guys, I'm doing great. How are you?
Last 24 hours been like for you?
It's been interesting. I do invest myself, so it's been probably one of the craziest months in Wall Street history. Um I was talking with some friends last night about long-term capital management, Amaranth, other famous liquidity driven blowups, and this is up there. And um yeah, it's just a really crazy thing. We had heard rumors sort of mid last week and then they really started crystallizing last night and this morning obviously sort of a fatal compon plea and I actually think they did a wonderful job of keeping it relatively quiet. I think some players were already positioning say early in the week Monday Tuesday looking to do what my old boss Kramer used to call you know shooting against a fund. So if you know somebody has to liquidate the best thing for you to do unfortunately sadly Darwinian is to go sell all the positions you have in common and go start shorting everything they have. Yeah. and it accelerates the sort of downfall as quickly as you can and this is a very common practice when these things you know happened. Um certainly not something I had overlap positions with them so certainly not something I would do but know a wide number of funds that were shorting all these stocks hoping to cause a panic and a crash.
How do you trace back the start of this correction? Is it the war? Is it oil? Is it jitters around open source or just hyperscaler capbacks? There's so many different narratives around why the AI infrastructure trade, the bottleneck trade might be weakening. At the same time, it feels like there's some really solid progress and the models are progressing along like pretty like as expected.
Yeah. Yeah. You have the labs having some of the best months in business history of any companies ever. Yeah. But then all the infrastructure correcting.
Yeah. None of that stuff matters. You know, the only thing that matters is the propensity of the buyer and seller to buy or sell. And what you had happened was the smart guys get in early, start buying.
See the prices go up, buy some more.
And then less smart guys take note and say, I want to do that. I want to be up 400% this year, too.
Guys like me started buying right near the top. I was just like, "Hey, this is great. I love memory. I love bottlenecks."
And uh and then but by the weakest hands are buying at the top. So they're also the first to sell.
And it just creates this like, you know, every bubble's sort of the same. You have this euphoria, this peak, and then, you know, everyone sort of panics at once. And you know, the fundamentals basically don't make a difference. you know, I think they, you know, they sort of drive the marginal buyer and seller, but you know, the 80 or 90% of the assets shareholders don't change hands. It's that 5% of the margin that's deciding the price. And if that 5% is in the state where they're levered up 3x or 4x, as we heard, SALP was a 4x levered fund, which is that's a lot of leverage. You know, a 25% draw down takes you out of business.
Yeah. Um, interestingly, we heard that three firms were bidding on the assets. So, Jane Street, Millennium, and Citadel were sort of brought in in a closed circle sort of late Friday to bid on the remains of the firm and uh, we got offered a look at 100 million of anthropic stock, which we were puzzled
puzzled by. you know, sometimes you see these SPVs sort of interest comes across uh you know, here and there and we thought that was interesting.
Um I sort of raised my eyebrow. I was like, is that Leopold? Because you know sometimes when you want to sell a $4 billion of something, you don't come out and say you want to sell $4 billion. You come out and you say I want you want to sell $100 million of it and usually a guy who wants to buy 100 is enough to buy 500 or more and you sort of fill them out and say here's a hundred. Okay, do you want five by any chance? And then, you know, your eyebrow starts to raise a little bit that, you know, maybe he's got even more. Um, now, of course, this is a really odd situation. So, we heard Millennium did put in a bid. Citadel's bid was better. You know, I think Ken wants to be the guy that everyone goes to when they're in trouble. And that's the, you know, Buffett is getting older. This is not the kind of stuff Buffett wants to do anyway. But, you know, Citadel did this in the Amar deal. You know, when Ammeth blew up natural gas futures, I think Citadel took that portfolio and virtually every blow up in finance.
Enron Enron where they just raided all the talent.
Yeah, they wanted to do an Enron as well. I think Yeah, they just sort of Ken is a very smart guy. sort of shows up and says, you know, how can I, you know, how can I, you know, be a partner to the Goldman's and the Bank of Americas when they need to get out, you know, of a really risky position, they basically take over the book, right? So, if you, I'll give you sort of an example, you're asking the question. So, let's say, you know, you're at 45 billion, you know, sort of try to trace this back and you're 10 billion of that is in anthropic from what we understood. So you have $30 billion of cash in your bank account. And running forex lever means you have 120 billion gross market value.
So if your GMV drops, I don't know, 25%. That doesn't sound so bad at 100 billion, maybe that's you know, I don't know, 30 billion. So you're down to 90 billion. But that's not your equity. So your equity drops from 35 billion to 5 billion. Yeah. And no prime broker is going to let you keep 90 billion of gross market value because once you dip your equity below zero, it's their loss, not yours. And they're not going to lose a penny after archaos and after these other kind of blowups. That's not their job. And they kind of have the right to take over your portfolio, which is sort of you know, something I hope nobody ever has to experience. But they basically call you in and say, "Listen, you know, these are our assets now, and you know, we're going to decide what their disposition is going to be." And the rumor is over the weekend he contacted about 10 parties to place Anthropic in an effort to shore up liquidity, selling the Anthropic stake for allegedly that the offer was at 1.1 trillion equivalent market cap. Um, which you know is I think roughly where it's trading. And um you know it's unclear whether that was sold or half of it was sold is what we reported that half of it was sold. It's still a little unclear who bought that what's happening exactly but that's the best we've got. And then, you know, when it came to the public book, it does sound like the buyer of that book basically got a from what we were told a three to four billion dollar insta markup.
So, they basically now have to work out of three to four billion.
Uh more than three to four billion.
Quite a lot more. But in essence if they work out of these positions without disrupting the market they'll have printed 3 to four billion on the trade which you know is unusual and interesting trade but you know really exciting. One of the parties reached out to me last night, one of these three parties. Interestingly, after my reporting, and they said that in essence, at some substance, yes, Leopold flew a little too close to the sun and your numbers are a little off. And I asked what direction, and they wouldn't confirm or deny. I received a lot of push back on the reporting to your point privately and publicly that it's not so bad and that, you know, he's only down 30%. 30% you can kind of live with. But also if anthropic hasn't changed its mark that means you were down 60 in the public book and if your forex levered you know that means you're sort of down 15 on the public book. Um which sounds too good to be true. If you're trading these stocks they were down like 15% a day.
So we've also heard the other AI funds are hurting. Maybe not as much as in trouble but certainly hurting as well. Where does the fund go?
He gives some good cover to all the funds that were effectively copy trading him. Oh, sure. Maybe even being more riskone and later to these positions because they're naturally just late. If you're trying to copy trade someone and you're
trying to catch up, you know.
Yeah. You're trying to catch up. Yeah. Yeah. More leverage. You're coming into these trades way later. Um, do you recall like how did you process Ryan Jacob in around the year 2000? Because you were at Kramer's firm. I believe you joined maybe right before the Ryan Internet Fund started collapsing.
Yeah, there's also the Amarind fund. There was a fund in the 60s called the Manhattan Fund that Warren Buffett criticized for being the go-go kind of like fund was run by a guy named Gerald Sai. And so like every generation you've seen the memes about Kathy, you know, every generation has it, you know, the guy that believes in that cycle and goes balls to the walls on that cycle. And look, I have a lot of respect for somebody who's willing to do that. I used to tell a friend who kind of did the same thing. He followed this trade, but he was very early. So he had sort of Leopold like numbers and he sort of did hedge at what sounds like close to the top. So sort of a miracle trader, best trader I know. And I joked with him, I said, you know, if Leopold sells at the top and turns short, like I will absolutely agillate him as the greatest of all time. And it's just that, you know, usually when you're so spellbound by that narrative of whatever happening, in this case, AGI, you know, there are people out there that say, "Look, AGI is here slash coming. When it comes, the entirety of finance is not relevant anymore." Yeah. You know, we might as well just run it up and kind of see the end of days this way. And of course, to some guy sitting on a trading desk at Goldman Sachs, you're like, "These people are nuts." You know, it's just the stock market deal, you know.
did you given that Leopold had been at FTX right up until the fall, did you think that maybe as risk on as he was like maybe he was like, you know what, I just I can't go through that again? uh he wasn't necessarily tied to any of the sort of nefarious activity at FTX, but he did have to viscerally experience it and I believe resign the day of the collapse and uh I just would have expected to not like yeah to run it back like so quickly. you would expect even like you know go and do have a normal you know great career for a decade or whatever then maybe come back to leverage and be like I'm ready to dance again but
there's a lot of questions like one question is what's his carry you know when a lot of firms in the hedge fund industry believe it or not they have clawback provisions for carry like high watermark provisions right so you have to clear something s everyone has a high watermark but what's increasingly happened is a carry provision where you have to return the two and 20 you earned if you have a severe draw down which you know could actually end up being a tough situation. Now as you guys know the fellow is getting married this weekend as well which is you know a little bit of tragedy with a little bit of triumph mixed in. Um but um obviously you know when this
but does every how common are those clawback clauses cuz you have to imagine in this fund raise you had like massive massive leverage you know like demand was very high
demand was very high that feels like a term
the numbers were so good yeah
it's a more institutional thing and you know speaking of which you know obviously the guy basically had no experience um and again you know in times like this nobody wants to grave dance and I'm doing that. But I had some institutional friends, one of the biggest fund of funds in New York, for example, who passed on Leopold basically laughed at him and said, you know, there's no way I could invest in this. And of course, you know, he goes on this tear, you know, makes like 20x or whatever it was since inception and does fantastic and he feels sort of sheepish, but ultimately, you know, somewhat vindicated after all of this. So you did have a manager that had no experience um kind of a long only or extremely long biased starts to do private which for many hedge funds is kind of the death nail. Um you know when hedge funds put on their VC cap and try to do what those guys do it often doesn't end well. And um that goes back like you know 50 years basically of hedge fund history. And um very few people have been able to do both. And the other thing I'd point out is we're gonna see July numbers very soon here from quite a lot of hedge funds that I think were in the same trade. Sure. And so this is not just Liupold's 100 billion gross. It's like that times maybe five or 10. And while the market's liquid, that's a lot of downward pressure in a few weeks. And you know, it's amazing to see this all compressed in a month whereas like the dot bubble took three or four years to like patiently go up and patiently go down. Um you know, seeing that compress instantly is interesting. what's going to happen next is really going to be fascinating. There's some theory out there that, you know, that we see all-time highs again now that all this liquidity is out and there's other theories there that we actually were just having this nice big downtrend and that this liquidity pop will fade and we'll be back down further and further. Um, you know, nobody knows what'll happen, but it certainly while you're right that you know, the anthropics and open AIS are having record business results. So is Microsoft and Google and Meta for that matter.
There's still I think some more discerning questions about is this capex investment worth it.
Um you know they rewarded Meta Microsoft for being prudent. They punished Meta and Google for not being prudent. So one wonders what the future will bring there. But
yeah about as crazy as things have gotten on Wall Street in many years. uh probably at least since FTX and certainly crazier than the sort of Tigersoft bank venture boom of 21 and then you know really since then uh the '08 insanity so it's quite a spectacle and I think you know no matter how much people want to learn the lesson of leverage over and over and over again we all seem to repeat it and uh you know it is what it is but I think the Jane Citadel Millennium kind of like entire hedge fund complex sort of becoming this like shadow bank is quite interesting you know in that like these guys are sort of there to normally the banks would sort of take this on the chin but now that there's other folks who are like you know Jane was an LP for example in reportedly was not interested in bidding uh which is fascinating may have taken the anthropic however um really unclear we're going to learn more obviously as some days go on here but it's an unprecedented time and you know really insane story that may just get more insane as we learn more.
Is there a world where the fund continues? Because I'm just hearing the numbers and it's like, you know, up at 45 billion. The actual money into the fund was maybe five billion or something if you sell the positions. There's a world where you wind up with like 10 billion in a bank account and the LPS are like, "Well, we gave you five. Keep going. Get back in the game." You know,
I hope that's the case for the LPs who are awesome, for the fund manager who obviously got quite a lot of whiplash, but you know, at the end of the day, you know, there's this concept on the street as you guys know, like once there's blood in the water, like these positions will go to zero. Like we'll send Micron to $5, you know, just to eliminate this guy at three, right? Like that's, you know, the craziest thing is like that's the nature of Wall Street when this happens. and there's a guy that has to sell a hundred billion, you'll have a trillion dollars in front of him just like, you know, let's see this guy cry uncle. And it's the saddest kind of most machavellian thing, but like he sort of had to blow up, you know, there was no other ending sadly.
Um because of the leverage level. It's just like one slight, you know, I remember my old boss was a Tiger portfolio manager reminded me of the 2000 era where there's this very slight change in tone from one optical component supplier and that's like him and his partner from Soros just decided to go like as short as they could this because they knew ultimately these vulnerable hands were sort of sitting there after the easy part of the bubble was over. You had this like okay what's next? things have to get a lot crazier. You saw Doresh's tweet. Things like that would have to sort of happen for there to be enough second derivative for somebody to be surprised. You know, everyone knows AI is in this boom. Everyone knows chips are in this boom. What could possibly shock you to the upside? Not much. So, if you hear any little like, h, you know, we're not going to spend as much, the whole hits the fan and every it's just too heavy. So, I actually wonder if we're not in for a longer, more protracted decline. Things feel great today. You know, you have this huge boom, this relief rally. Um, a lot of the froth is out of the system. But, you know, what next? You know, I don't know that, you know, a patient and calm market is going to emerge because you had the hyperscalers and the big companies, they fomoed, too. They fomo just as hard as Leopold did, right? If not harder. So, this isn't just him. It's the whole world collectively saying, "Fuck, I got to go all in on AI." And it's and who had the guts, you know, other than one man Tim Cook in the back saying, "Not me. Do nothing."
Yeah. the funny thing, you know, we had been joking uh we were joking in like Q4 when you know there prior to like coding agents really starting to rip, you know, OpenAI revenue growth had like slowed a little bit and like there was some jitters and a lot of this stuff wasn't, you know, public at the time, but you could tell some of the kind of crossover types were like getting a little nervous, right? They kind of expected
MU DAU numbers, you know.
Yeah. it really plateaued and then we and then there was a correction like there was like briefly you know for a period it was probably like eight weeks it was like okay like uh and then it started ripping again and we were taking like sort of a bit of a joking like victory lap being like cool like AI corrected you know bubble pop we're able to build back sustainably we're good from here on out it's smooth sailing
no I completely agree I think the most unexpected thing is it would be if we saw brand new all-time highs for the entire thing. I think almost everyone on Wall Street is skeptical this will happen, which means it has a chance of bullish.
So, you're saying there's a chance. I love it. Can you give me a little bit more insider baseball on what it takes to unwind a big position as a shareholder? Because a lot of people who are not inside the hedge fund world are sort of maybe confused around, okay, yeah, you own $50 million of a $1 billion chip stock. Can't you just dump that on retail? Can't you just like sell markets sell that on Erade or Robin Hood? And in fact, it's much more complicated when you're at this level even though it's public markets. There's not just a big button. Can you walk us through what it actually takes to like sell a big position when you're at that level?
Yeah, there's a lot that goes into it interestingly. So the first is you have this advertisement system. So if you sell into the market, you can try that. And those that's called selling into the screens. The screens are the numbers on your screen. Anybody could buy and sell Robin Hood, whatever. So you don't normally do that if you can help it. Uh selling on the screens is at least somewhat quiet. You can just sort of trickle out. There's always this conspiracy that as I'm selling on the screens, there's some guy who's can see my screen and he's like, "This guy's got a BW market order to sell 10 million shares. That's not, you know, I'm going to tell somebody." And that knowledge would be very very powerful. And there's even some even crazier conspiracies out there that quants could actually use different all kinds of insane you know ideas around what they can do to sort of sniff out that this is happening. So there's people that are scared of that. Then you can pick up the phone and this is the way you'd normally do it and you call Goldman and you say, "Listen, I need to sell five million shares of Microsoft or something like that and they say, hm, you know, should we take it or do we find a guy that wants to take it?" And they'll sort of try to decide. Now Microsoft is easy. If you're trying to sell Sharon AI, a neocloud in Australia that nobody wants, that's a tough one. And you own like 10 days of volume. So, if you try to hit the screens, you have 10 days of volume. You'd have to be the entire volume for 10 days before you'd be out. You'd probably take the stock down 50% or more, and you don't want to do that. So, you try to do this advertisement process. Um, you know, and you basically can post in the stock market that you're a seller of a stock and you can post that your four-digit what's called market maker ID and so Goldman's is GSCO. So GSCO would be a seller of say, you know, Nebius, which was one of his positions. And so you'd call up, you'd say, "Okay, Goldman, I'm a client too of Goldman. Uh, you know, what do you got on Nebius?" And the guy would say, "Listen, we got a pretty big seller here." You know, and say, "How big?" You know, half a million shares. And you say, "A lot bigger." You know, and so you'd say, "Hm, okay." Because they have to advertise that, you know, they're working your order. So they have to sort of tell people that there's a seller. uh they kind of are trying to be koi about how big, but they're not going to waste somebody's time either. So, the guy who's heard that there's a big seller, well, he might turn around. He's not supposed to do this. He sort of might turn around and say, you know, there's a huge seller of Nebius out there and I'm just a little baby fish. Maybe I could short 50,000 shares and get in front of this guy. If you're an actual interested buyer, you might also still be nervous because you'd say, "Well, if he's really got a ton of size, I might have to be judicious about how I step in." And so, if you combine that with the sort of like pressure in the market and you add it all up and then usually what you do is you'd have say, "Oh, I know a guy that works there and let's see if he's returning calls." And you know, when you hit up the guy and he's not on Bloomberg, he's hard to reach. It's kind of like, well, it sounds like it could be them selling. Uh, so it's not too many people that own that many shares of that security. So, you look at the holders list and you're sort of like, who could it be selling 10 million shares? So, you call Fidelity and they say, no, we're not selling. You call the next guy, no, we're not selling. Next guy's an ETF. Next guy's a index fund. You know, it's got to be him, you know, and so if it's them and they're and then you start noticing all of their positions are down, it gets really hard. So ultimately the bank decides because you might say, you know, I don't want to sell. The bank says, I don't care what you want. We're selling regardless. And Goldman Sachs is not in the business of holding AI stocks. You know, we're going to sell at any price we can because our board would rather know for sure that we're down a billion and just take the rip the band-aid off than to wonder if we could lose 50. And so it's Goldman's position that we're just going to cut the arm off right now before it metastasizes. And so they'll do a fire sale. And of course Goldman's smart. They're going to reach out to a guy like Citadel or somebody else to place it carefully. But selling the whole portfolio in one shot was a very smart move. Now again we've heard the discount could have been as big as you know 20 to 50% which is you know mouthwatering discount to buy you know some quality companies at but to end it and have finality what was really to answer the question finally what you really needed to do is the buyer of these stocks has to have the liquidity to hold them for 5 years and do nothing because the market guys like me and to a very small extent and guys to much bigger will sit there and say I don't think you can hold this and they'll start shorting it and shorting it and shorting it and trying to make you cry. Uncle Kyosha in Japan, one of Leopold's holdings, also one of mine is trading at three times earnings. You know, they basically force you, you're forcing the guy to really, you know, to sell. And if you're going to hold this stock, you have to make sure that you can hold it until it's two times earnings or one times earnings. And the only player big enough and more powerful enough to sort of hold a hundred billion dollars and not blink is somebody like a citadel. And even still, some people there rumors out there they're the people who are going to try to pressure Citadel, which I wouldn't advise uh you know, but something like that where you know, maybe they'll now have to suffer the same contagion. So, it's a very crazy time in the markets and I don't think we've seen everything yet because I do think there are some large tech funds that have had the same trade on. I do think liquidation is over thankfully, but I do think that there are some funds that are about to be found out to be down 30% or down 40% or something.
Take me through the mind of Ken Griffin like a couple weeks ago. there's this rumor that he was sort of like pushing or signaling that there might be a rate hike. But what I'm interested in is if you suspect that there's going to be a fire sale on X, Y, and Z companies, is there a world where you build the hedges before you acquire those assets or is that two 4D chess? because that I if they wind up acquiring these for 50% 20% off but they already have offsets then they sort of come in market neutral. Is that possible?
I don't think so. So I'm familiar with the Citadel's performance for this month which is surprisingly up. So I think they're probably one of the only hedge funds in the world that's up this month. It's up very small.
They were actually hedged is what you're saying.
Yes. They have a diverse platform of different businesses. A guy trading weather, a guy trading rates, a guy trading stocks, you know, about a thousand guys trading stocks. And they have a computer fund, you know, called Cidel Securities that is a market maker that trades a good chunk of the volume of every instrument in the world. And ultimately, I think that the prime brokers, the Goldman's and Bank of Americas, they do so much business with Citadel and they've done this before where they know who to go to just the same way the US government went to Warren Buffett when they wanted to shore up Goldman.
They know that the right person to call is Ken and he is really going out of his way to make himself the guy to call. And I think that is a great brand because you may not need to be that guy more than once every decade but look once a decade to make a free 5 billion or 10 billion is a great great guy to be. And you know it's sort of like he becomes a dependable trusted partner to these banks and if he wants something from the banks he's helped them because without him they might have had to sell that at a negative number. In fact, some people think I don't think this is what happened, but some people actually think the equity in Liupold's prime brokerage accounts went negative.
Which I think is you know something that again gold the Goldman and Bank of America's try to stop you before you get there.
But you know they also don't want to sell like I said a share in AI which is an illquid tough to sell security.
you know, they'll sell out your micron very fast or you'll sell it out before then. But if you're left holding this bag of like a liquid crap that you have 60 days of volume to get out of, it's pretty tough to sit there and tell your prime broker, don't worry. Uh which is why again I think he needed cash. Probably somebody on Monday or Tuesday tapped them on the shoulder and said, "Your margin's looking a little thin. You know, can you add, you know, a couple billion here or more?" And things happened so quickly that there was just no time. And um yeah, it's I think Citadel learned about this at the 11th hour as every as you're supposed to. You know, the firm didn't leak out that they were hurting. Um they didn't have, to my knowledge, daily performance. In fact, from what I'm told, um situational
Awareness as a young hedge fund was not so great with communication. Not surprising, especially with monthly and quarterly letters. Could have been more timely on some of those. So it's a small group of a couple of guys, so I don't think that this was the same.
You rewind — what was it, only a month ago that the... or 13F was late? 13F was like late and everyone was questioning, like, did he work out some kind of deal to keep it confidential? But it sounded like he just didn't get around to it. They had other priorities, maybe.
Do you think you can rebuild a career as a venture investor? Because in venture you're just gigal long always — it's one of the few forms of investing where it's just so hard to get out of position.
That's the thing. I mean, why become a hedge fund manager? I have a friend who wants to start a hedge fund. I told him this is the most painful, horrible business in the world. Why do this? If you start a newsletter business that makes $100 million a year — even $50 million a year of revenue — you've done better than almost every hedge fund on the planet. You do not want to do this job. And the reason people do it — and I did it too, and I would never do it again — is it's the sexiest thing in the world. You think you're the glorious universe... it's incredible. You're the master of the universe. And I had friends wanting to quit really high-profile jobs to be a hedge fund manager. I was just like, you're out of your mind. You don't know what this job is. It's waking up at 3:00 a.m. checking Korean stock prices, waking back up at 6:00, wondering what's happening in the world. And there's absolutely no productive thing you're doing. You're providing capital, but other than that, you're really playing this high-stakes crazy poker game. It's certainly fun and interesting, but when it's painful and raw... you know, I hope he'll do something. He's a brilliant person. Brilliant people like that — look, Peter Thiel had a hedge fund that didn't quite have this level of liquidation or anything like that, but it had a rough last few years. And Thiel was able to not only continue his venture investing efforts, creating one of the biggest funds of all time, one of the most successful funds of all time, investing personally, doing amazing, also getting back into macro trading with Thiel Macro, which supposedly has done well. So I do think there is this period of a few years where he can reset and take the learnings, take whatever talent, skill and certainly genius that nobody denies he has, and rebuild. I don't think it's the end at all. And I hope he's keeping that even temperament about this, because I think a lot of people respect him quite a lot. No matter how this turned out, he'll be back and successful. But it is a little bit of a humiliation thing that I think most people on Twitter and other places are sort of saying, well, the market tends to humble you. And this is an extremely humbling moment — from being just two months separated from the biggest hedge fund on planet earth and most successful, to being forced to sort of liquidate. That is quite a rapid reversal.
Also just imagining what the fund looks like in two or three years if you just survive, right? There was a clip circulating yesterday from his appearance on Dark Cesh where he's like, 'Oh, there's obviously like a 100x left before AGI, right?' So he was up 20x or whatever, thinking he had so much room to run, but just couldn't stay in the game. Extrapolating is always a risk.
I got to say, it felt like a huge moment for you and your business, just because everyone — the whole finance world — was learning about the situation from your post. I'm sure a lot of people were glued to your terminal, and it felt like a changing of the guard, because again, you were getting pushback. You were getting some pushback, but then two hours later it was like Financial Times and Bloomberg and Wall Street Journal — they all clearly needed a couple hours to run it down. But you got to it first, and yeah, I was quite impressed.
Thank you. Yeah, I mean, I think that we've talked about this in the past. There is a changing of the guard there. You guys help change the guard in your space. And I think the folks at the Journal, the folks at Bloomberg, the folks at these other companies — they're fantastic reporters, but they're not active or former players. And we will always hear things before them, especially on the street, because that's just...
And the craziest thing is you actually waited until it was effectively over to share, right? Like you had been hearing about this.
There's a lot that we sit on that we don't want to — we've been in that position hundreds of times where it's not appropriate to share anything. And sometimes you're sitting there being like, I'm really surprised that legacy media hasn't picked up on this story. It feels like it's just common knowledge, and there's definitely a time and place to just not say anything and let something work its way through the system. Yeah. I mean, to give the devil their due, the information is also quite good at this type of thing. And they are particularly good at scoops on OpenAI, which I still haven't unraveled how, but they're obviously very good reporters. But a reporter at a place like that in Tri Media — they generally don't care about burning bridges and resources or contacts. So they want that news out yesterday. You know, I do care. And it also is a conflict of interest because I don't want to hurt somebody that's given me good information and betray their confidence, because I have to keep the confidence of these folks if I want to keep talking to them. But I also, in the case of this situation, as the carnage is unfolding, there's sort of the balancing — the need for everyone to know with the need for protecting friendships and relationships. You have to make that judgment call each time. I mean, I hope that our customers understand that there will be things that we know before others and we can't disclose because we want to protect folks and protect our friends. Bloomberg, Wall Street Journal — they'll never do that. They're always going to serve their customer, who is the reader. We can't necessarily do that. You'll probably know things about a litany — like you said, hundreds of times — different fundraises going on, different things like that. And we have to all keep our lips closed because that will be the last time we hear about a fundraise. And I think that this was a situation where it sort of merited discussion. It was going to happen momentarily anyway. In fact, to your point, the thing that got me to publish was my friend saying everyone is hearing this now. Once that happened, I said, all right, well, it's time to let the cat out of the bag. It's about to be let out anyway. So...
Yeah. I have two more quick questions if you have a minute. One is just about how leverage works at a hedge fund. I think, again, from the retail perspective, from the much smaller player, you might know that you can go to a brokerage and get a little bit of leverage. But what does the process look like as you're scaling into the tens of billions of leverage? At a certain point you have to go to all of the banks, certain banks. Who's actually — what does that process to get leverage at that scale actually look like?
And also let's appreciate for a moment that I feel like just a month ago the West Coast broadly was taking this insane victory lap, being like, the West Coast is eating Wall Street — like the best and biggest hedge fund is no longer on the East Coast, we just have everything now, finance and technology — and then just deeply humbled within the span of 30 days. And it turns out you guys over there, you know a thing or two, and here we are asking you. So how would one go about getting...
So one of the things that I think is not well understood is the prime broker — they make their business to make a spread on financing. So if you go to a prime broker and say, I'm never going to use leverage and I'm never going to really trade a lot with your firm, they're just going to sit there and say, we'll still take the assets because we can rehypothecate them and lend them to the guys that are going to take leverage. But in general that's not a great customer. So if they're making a 1% spread, which actually is a relatively huge amount, and you're borrowing 4x, you're actually giving them 400 basis points of free money, which is sort of fantastic. In fact, their borrowing costs are probably less than SOFR. So they may be getting as much as 600 or 800 bips of free money on huge amounts of capital. So leverage is the best friend to a prime broker. Now, the risk guy is sitting there saying, well, wait a second — I love lending, but I don't like lending to concentrated portfolios. I don't like lending to short sellers. Short sellers can get big leaps in their — like GameStop, for example. So the most a long can lose is 100%. But if a 4x lever, the most a long could lose is 25%. So there's sort of this mix of things you have to think about. I think getting into the privates is usually, for me, a really bad sign for almost every fund, because as tantalizing as private companies are, there is a whole group of people on the West Coast who are much better at that than the guys on the East Coast. And of course there are funds now like Altimeter and CO2 and others that are doing both. And what made it so tempting obviously for Leopold is just how close he is — he couldn't be closer to Anthropic, and it's a company that over the last six months has had 100x the demand relative to the allocation. So it just felt like — and I don't know what the structure on those investments looks like — but it's like, if you're going to break your rule and do privates, then that's the company to do it with. But then you still get into a situation where you're like, wow, I really wish this was more liquid.
Yeah, I can't press the sell button. Give us an update before you leave on Korea broadly, because a lot of people are commenting on just how similar Leopold's approach is to Korean retail. I don't know how true that is, but I can imagine there's blood in the water over there and the whole country is probably in shambles.
Yeah, I think so. I made a Kelly criterion calculator and a little portfolio simulator tool. Basically, and Paul Tudor Jones said this a while back, and I had a problem with this — every single trader out there seems to make the same mistake over and over again, which is their position size is probably two to ten times more than it should be. And if you actually...
So it sounds nuts, right?
Yeah. But if you actually run the simulator — and ours does Kelly — so Kelly was a guy at Bell Labs. He was a member of the technical staff. His original OG MOTs. And so Kelly came up with the proof called famously the Kelly criterion, which gamblers use mostly — it was a gambler thing before a finance thing — and it proves the optimal bet size. And the optimal bet size is your edge subtracted by the reciprocal of it. So if you have a 55% edge, your optimal bet size is 10%. That's still quite volatile for folks, and so people do half Kelly or quarter Kelly. Most folks don't actually have an edge when they trade, but if they did have an edge, they're trading as if they had a 4x or 5x Kelly edge. Which is interestingly — you might sound, okay, well that just sounds squishy and like the guy takes a lot of risk. No, if you run the simulator, you will go to zero each time. And the simulator is a really cool tool that shows you even with a 60/40 edge on every trade you make, you'll go bust if you overbet. And it's an eye-opener. We might say, who has a 60/40 edge in the stock market? Nobody has a 60/40 edge. But you will absolutely go bust if you don't size correctly. And it's something that I've had to learn very painfully over the years — that I'm almost always overbetting. And I think every fund is sort of the same. And certainly every retail is the same. And it's just sort of this weird variance math game that very few people actually map out and say, can I simulate a portfolio and just see what is the right thing to do in most cases. And in fact, after I left Tiger, I worked briefly in the office of a guy who worked at SAC Capital, now called Point72, for years. And he was one of the best managers — a quiet guy nobody's ever heard of, kind of retired — but I got to watch him before I set up my own hedge fund and did the exact opposite, way over on everything. I got to sit with this guy for a few months and I was astounded. What I found is that he was managing, I don't know, three or four hundred million of his own. Basically, he almost never used the capital. 80, 90% of the capital was just cash, and he would just make these tiny trades. And the guy had almost never had a down — I think his record is he never had a down quarter in twenty-something years of trading, and he had like 20, 30% returns, which is great. And the guy just kind of did these little nibbles, and he never lost money, and it was this incredible thing. And then of course the second I get the chance to get some capital, I'm 8x, you know, and it's just like, you know, it's the dumbest thing in the world, you know, and you live and you learn.
Psychology. Psychology. Well, thanks so much for coming on the show and breaking it down.
This is always a great time.
Yeah. Looking forward to seeing where we go from here. Have a great week. Have a great weekend. We'll talk to you soon. Cheers. Bye.
Let me tell you about Codex. Codex is a powerful workspace for getting work done with AI agents. Whether you're writing code, analyzing data, creating content, or automating business workflows, Codex helps you move projects forward from start to finish. We have a couple of guests coming in soon. But first, I want to talk about Ferrari. I don't know — do you want to jump straight into a next guest? I don't. Oh, let's talk about...
I want to debate this because there is news in Ferrari world that the Ferrari EV, the Luce designed by Jony Ive, has already hit the 2026 sales target. The haters are in shambles. Everyone doubted that this would sell, and the Italian car maker reports strong demand from China for the electric model derided for its unconventional design. There's a whole bunch of interesting tidbits in here in the Financial Times article. Ferrari has hit this year's sales target for its first electric vehicle on the back of strong demand from China, despite a polarizing design that drew backlash from investors and enthusiasts. Remember, even the former CEO, former chief design officer, former executive came out and said, 'This is not a Ferrari.' There was a lot of back and forth in the timeline. I could have designed a better one with ChatGPT. A lot of people threw out different designs. But this one is selling, at least according to the Financial Times. The Italian group has not disclosed its target for the Luce. But two people with knowledge of the matter said it had aimed to sell this year just under 500 units of the EV. Not a lot, but their goal by 2030, over the next four years, is to get to 2,500 units. So 500 a year for four or five years — that's where they want to get here. And they say they're on track. And so this is priced at €550,000, $650,000 USD, something like that. One of the two people said the target had been reached earlier in July, just two months after its controversial launch, when critics on social media derided its unconventional styling. And there's also an interesting line in here that Ferrari says they gave strict instructions to dealers not to force traditional petrol-loving collectors to switch to electric cars. They said, 'If you want a Luce, we'll give you a Luce, but we're not going to make you buy a Luce in order to get in line for an SP3, SP4, some special F80 thing that's more limited. Just if you want it, it's here. It's a choice.' And a lot of people made that choice. Cara, according to these insiders, what do you think? Give me the pushback and then we'll debate it.
First, let me tell you about Console. While you think about that, Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets.
I was pausing because I wanted to start with something nice. I think...
So the videos that I've seen of it on the road, it does look even more strange than in the images, I'll say.
Wait, what is this photo? That's not... What is that?
The team just accidentally put in some random car.
Okay. So the car is strange. I love the interior, but the car overall is strange. It's still unclear to me who it's really for, but they are finding buyers. The idea that buying the Luce — like Ferrari is saying we're not forcing any dealer to push this car or whatever — but the idea that whether or not you bought a Luce is going to have some weight on your future allocations, to me is just insane. There's just no way that that's true, because every single dealer is going to look at their client list. They're going to look at what cars have they purchased. It's like we know...
We know multiple Ferrari collectors that are buying two separate cars that they don't even want of the same style in order to gain status within the dealership and show that they're a proper collector and they're properly sort of cherishing the brand. Yeah. And so I'm not at all surprised that they've sold 500-ish units. That's about as many as I would have expected for 2026. I actually, I guess if you asked me, I maybe would have thought they would have done more. Like to me this was a car that was so different than the rest of their cars. It serves a wildly different use case. I would have expected their sales targets to be quite a bit higher, simply because when it comes to their really special cars, they make about 500 of them. I would have expected at least 500 sales guaranteed, and then you would hope there was a bunch of incremental buyers that are like, yeah, I actually don't want a Ferrari sports car, but I do want a daily, and why not go for a Ferrari daily, right? So you would have thought that there would have been like — 500 for the first year was like my very base case, and I would have expected a bunch more on top of that. So I think they're positioning this as a win. I think people are going to love the Luce if you ignore the price. But I don't think it's the win that... I love the Pope in the Luce. So good.
Okay, three points in response to yours. First point: the design is absolutely growing on me. Like watching these videos here, it just looks way better than when we first saw the first pictures. And I don't know if it's just distance and I'm becoming more familiarized with it, but it looks a lot better. Even the exterior — I've always agreed on the interior. I think everyone agrees on that, but the exterior is looking better to me somehow. I don't know if this is just like I'm getting used to it. Two: yes, there are the 4D chess Ferrari collectors who are saying, I know I'm not getting pressured, but I'm buying one anyway because I think it'll help me jump the line, and I'm doing that independently of any pressure that's coming. But there's also just collectors that are like, this is going to be a piece of Ferrari history regardless of if they — what if they never make another EV again? What if the Luce can't sell next year and Ferrari literally for 30 years never makes an EV? This thing is important historically. It's an interesting thing to have in your collection. And then there's also just people that are like, I'm a true collector. I want every possible Ferrari experience. Give me the SUV because I want to see what that's like. Give me the mid-engine. Give me the front-engine. Give me the electric. Give me everything. Give me a vintage. Give me a new one. Give me a road car. Give me a track car. I want a Ferrari F1 car. Like I just want all of the experiences because I just want to experience everything Ferrari because I'm that deep with the brand. So there — and are there 500 of those buyers? Maybe. And then lastly, the question of what was their goal. I don't see this as their Urus. I don't see this as they were trying to make a mass-market daily. I think that they were trying to make a very iconic, very iconoclastic, very contrarian car that was bold and weird and different, and it happened to be dailyable. And the fact that it is dailyable is what is weird about it. Like they're not known — the brand is not known for being able to be dailyed, and yet they made one, and that makes it weirder. And I don't think that they were going after this as something that will be like a Lamborghini Urus, which they can sell in mass volume and completely change the profile of their sales curve.
My view is that — I think those are all great points. My view is that I don't think they should have done a car like this because I do think it hurts the brand, unless it was going to drive so many sales that it could make the rest of the cars that they make better, right? Like what the Cayenne did for Porsche or what the Urus does for Lamborghini, right? This sort of workhorse product that can fund a lot of the other cars that are more halo cars. Sure. And so yeah, I just don't think it makes sense if they sell 2,500 units, other than it solves their sort of European emissions standards issue, right?
Yeah. I do think a lot of people buy a Urus because they want a Lamborghini, but they need a daily, and so they go with that because it says a lot. It's a very bold choice, and I don't think that's the calculus here because you're paying so much more, whereas I believe the Urus is cheaper than most of the Lamborghini sports cars, whereas this is substantially more than just going and getting a...
That is cheaper than this.
Is significantly cheaper, and it is much more desirable to the Ferrari clientele because it has a naturally aspirated V12.
Yes. But at the same time, the Purosangue, the SUV that you're referring to, had a lot of pushback when it came out. So did the SF90, and they all look really great now. And so the anti-halo halo car theory also holds. But I think it looks good, and I think it's going to turn heads in a weird, weird way that very few cars — as we are collapsing and we're getting to everything is white, everything is gray, everything is black, everything is the same rounded. Can you tell the difference between a McLaren or a 296 or a Lamborghini? They're all sort of starting to look the same, like supercars. And then there's the whole hypercars, the Battista and the...
Yeah, I'll give you this. I think certain specs of the Luce are going to look funky, but they're going to look cool, and they're going to be a joy to drive.
I just like that Jony Ive in the Financial Times — they sent this photo of him looking over. We got to pull up the — go to the Financial Times article. Scroll down to the picture of former Apple designer Jony Ive who designed the Luce. And he's just like — he's like, 'What did I tell you? You doubted me. You doubted me. You doubted me. You doubted me. We couldn't sell 500 of these. We did.' It's a true victory lap photo, and I love it.
Anyway, and I can't wait to see one in person.
I'm excited too. Big moment. Tell you about Figma first, and then we'll bring in our next guest. Agents meet the canvas. Your AI agents can now create and modify your Figma files with design system context. And we have Guillaume Verdon from Extropic. Long overdue, but very excited to have him on the show. GV, how are you doing? Welcome to the show.
Doing great. Doing great. Longtime fan, first-time caller. So great to be on.
Thank you. Thank you. Long time coming.
Give us the state of the union on the company, and then the news and the deal that's going on, the letter of intent.
Yeah, I mean, so at Extropic we've been pioneering this new form of computing from the ground up called thermodynamic computing. You know, our whole thesis is that right now everyone is focused on scaling things up, buying more GPUs, bigger data centers. We want to scale things down. We think there's going to be a race to densification of intelligence. So we started that about four years ago, which was very early to worry about energy. It sounded crazy to say we're going to run out of energy four years ago, but here we are. And so we've reinvented how to leverage the transistor for the era of generative AI. So generative AI, for those not familiar, is a probabilistic algorithm, right? You're sampling from these distributions of like, okay, if I type in the word cat, what sort of image do I get out? It's kind of random at the output. And so that is a probabilistic workload. It turns out you could run electronics probabilistically and at much lower power. And we pay a huge price to maintain our electronics in a deterministic state. So it doesn't make sense to run probabilistic workloads on deterministic electronics if we're going to pay a huge tax to know the state of our computer at all times. So our insight is if you use very little power, essentially your computer can operate probabilistically, and you can train it directly just like a neural network, and you can map your algorithm — that is a generative AI algorithm like a diffusion model — directly onto the physics of the hardware. And so that's kind of the vision. And it's been a journey. Essentially, a year, a year and a half ago we had our first prototype in silicon — there you go, X0 — and this one we taped out with TSMC. We've also now done GlobalFoundries. And that demonstrated that actually, because when you operate electronics probabilistically, you can use far less transistors for these algorithms. You don't actually need the smallest transistors. And right now the reason people have to go offshore is because the best cutting-edge fabs are all offshore, and they have the smallest transistors. And so what we demonstrated is that in principle we could manufacture these chips in the US, and that was very interesting to all sorts of folks higher up. And you know, we had some conversations, and there's a very strong history of the government supporting the silicon industry. I mean, Silicon Valley was...
Back to DARPA, they created the internet. Right.
Exactly. Right. And now, you know, we're trying to go with — or I guess they're going, we're going with a modern twist where if the taxpayers are going to support R&D, they might as well get some upside as well and get some equity. And so this is what this announcement is about. It's with the CHIPS R&D office, and you know, we'll get into it, but...
In terms of where you want the first applications to be, it feels like different models are maybe sometimes designed around certain architectures. You see certain — like GPT-OSS runs really well on Cerebras, and then you'll have another model that runs on an NVL72, it's like rack scale, and then there's other models that can sort of run locally or run on a Mac mini or Apple silicon, and there's all these different pieces of the puzzle. And increasingly we're starting to see where like voice models might be really good for those to run locally, but some crazy long-running AI agent might be fine to send off the server. Do you have an idea yet of where you think, if things go perfectly, the first application might be? Like where's the sweet spot?
Yeah, we've been exploring a lot of applications. Obviously, running the model itself, the transformer, is a big one. What we found is that — and we have a blog post coming soon. I guess it's a scoop here, but we have a blog post coming on a sparsity scaling law. So if you sparsify your model, because our chips have a sort of sparse structure. So sparse is just — you know, GPUs, they have big matrices which are these grids, and basically every entry in the grid is busy, right? Like there's something going on. Sparse means there's a lot of zeros. There's a lot of stuff not going on, that's not activated. So our chips are really good at sparse operations. That's not originally what we're designed for, but if you really want to shove a transformer in our chips, you can do it. And if it's sparse, as long as it has the same number of parameters, you could reach the same sort of performance, which is interesting because for us, the FLOPs are much cheaper per watt. So even though you use more FLOPs, you get more intelligence per watt. So we have something coming out soon on that, and teasing it here a little bit. But overall, we're really interested in diffusion models, models for predictive control, for defense, physical intelligence, signal processing — you know, stuff you would use an FPGA for, you want to use one of these chips. So you know, we have RTX as investors for that reason. And similar defense primes, very interested. Yeah.
How are you feeling about the retrospective on effective accelerationism? Because it feels like my interpretation is that you were 100% right at the time, and that class of models didn't really pose any real risk. There was clear need for a buildout and diffusion of this technology. But now we're in this new — the last few months there's been a new discussion over literally slowing down, literally decelerating, and it's coming from inside the leading labs, not just like doomers who are outside. How have you reflected on what EA was, its role now, just AI optimism versus pessimism, risk — how are you thinking about all that now?
Yeah, I mean, originally I wanted to just bring balance to the force. Of course it was kind of a monoculture in Silicon Valley. It was just the doomer camp, and you know, I was seeing the writing on the wall — if we just had the doomers, then that would eventually affect policy. And to me it was a very biased view of the world. And so I think we kind of stretched the Overton window. It was okay to say no to doomers and say that's a ridiculous model of the future, you know, gray goo, foom, all that stuff. But now, yeah, we've reached a point where the models are very capable. And you know, the EA view to me is just like viewing the whole world as a complex, self-adaptive system. And ever since the dawn of time, it's basically been PvP. Every form of life is becoming more complex and harder to predict, and then its adversary has to step up, get smarter in order to predict it, in order to compete. And if you do that slowly, then basically you can have a whole ecosystem and everything's good. But if one thing becomes much smarter and much more complex than others, then it can kind of control others, and that's bad. And so I think pace stopping makes absolutely no sense. Pacing can make sense, but really it has to come with a really strong investment in, for example, cybersecurity and hardening things, because you can't pause things forever. Essentially you just want to be adversarially robust, and I'm all for that personally.
It is very interesting reflecting on the doom arguments, how they were — like to your point about gray goo and foom — it was not as precise as cybersecurity risk, which some people were calling out. But the conversation was definitely distracted from the more practical difficulties around slop and overinvesting and bubbles and all these different things that are much more short-term that America will need to grapple with to actually deliver a positive outcome here. How is your development process being accelerated by AI?
Oh, absolutely. Yeah. No, absolutely. I mean, we use all the big model providers in-house, and we try them all. We have agents constantly running. You know, for us it's like we have to speedrun deep learning. It's like you teleport back to 2011. You know, and there's no AlexNet paper, and you have to speedrun 10 years, 15 years of deep learning progress as fast as possible, right? But now it's like new game plus in a video game. You have all the power-ups, you have the AI, and so we're speedrunning progress very quickly. And you know, we have online learning agents. We're training our own models as well, or post-training rather, sorry, as the cutting-edge companies do. And so that's very exciting to me, that current AI can help bootstrap and kickstart the next paradigm, the next substrate. And going back to your comment about the cycle, you know, my thesis was that GPUs are not the endgame. And so you know, we want to invest in the current cycle, but we should hedge our bets. You know, as we saw today with Leopold RIP, you know, you got to hedge your bets. And I think if we're all in on GPUs and we don't invest in the next generation, next paradigm of computing and have a couple bets there, then when that pops up, you're going to get wiped out, and we don't want that. And so to me, we're starting the next S-curve. We're starting the next cycle. I know people are not over the current cycle, but everything is cyclical. But you know, it's my responsibility to make sure we can have buildouts in the future that are really power-efficient, and you know, it gets you better return on investment on your capital, whether it's on Earth or in space.
Or, you know, at the edge.
Yeah, it feels like one of the — you know, you've been one of the loudest, biggest voices on X for the entire time that we've been running the show. We've, you know, had a bunch of your posts on the show over the years. And it feels like you picked a product category that just by the nature of it was going to take time to evolve. And I think like when I think back, it's like the challenge of showing progress. I mean, this moment is massive, right? It's a vote of confidence, and it's very exciting. But everything that it took to get here — meanwhile, you're getting like hundreds of millions of views from some of like... Yeah. Billions of views. And I'm curious, like, reflecting on the last two years, do you feel like you've found the right balance between like you want to be getting attention and you want to maintain your voice, but you also want to be moving the business forward? And at any point, you know, X and Tech Twitter specifically, if they feel like those two things are not balanced, they're going to like pounce and take shots and all that stuff. And I know you've gone through that, which is why I'm excited for this moment. But yeah, how have you processed it all?
Yeah. No, it's been a journey. It's been a lot. It was — despite, you know, allegations, it was never planned to get doxxed. And you know, I knew I was going after very much a deep tech moonshot that was going to take some time to cook. But you know, exponentials are slow at first and then they compound over time. Could I have had, you know, a product that would have had a shorter time to market and used the heat to make it grow? Probably. But you know, I was just dedicated to this one mission. I think it's the most important thing I could be working on, and I've been all in since the beginning. But yeah, there's certainly a tension like building in deep tech where you're supposed to be in stealth. You know, you have nation states trying to reverse engineer your technology, and you're trying to just reveal the minimum because that's your secret, that's your edge. But at the same time, you want to get buy-in. You want to shape people's beliefs like, hey, actually this technology is here. It's coming. Here's a hint. Here's a prototype. Here's a Nature paper. Here's this. Here's that. Here's some signal. We're going to have a lot more coming up next week. A big announcement for Z1 and all those products around it, including our stack. So...
I love that framing of new game plus. It feels like there's so many founders from the previous era who grew big like SaaS companies and are now sort of on new game plus, moving a lot faster. It's a great formulation, great philosophy, and thank you so much for coming on the show and breaking it down.
Yeah, congrats on all the great. We'll talk to you soon. Have a good one. Goodbye.
Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web apps, servers, databases, and more, while Railway automatically takes care of scaling, monitoring, and security. Did you see that California startup Satyres unveiled a 2-meter-tall centaur robot designed to enter disaster zones too dangerous for humans?
There's no way this is a real picture. Sorry. Go to the website though. I mean, Satyres. S-A-T-Y-R-E-S.
They made the demon robot to go into disaster. Imagine you're in a disaster zone, John, and through the fire and flames, you see...
Yeah. Okay. Okay. I actually am in Hades.
I'm in Hades. If this thing says, 'No job is worth your limbs,' that doesn't mean a job isn't worth doing. Do you think this is a prank? Do you think this is a joke, or do you think this is a real company?
I think if it's not a joke, then they got to go back to the design minds and do something else.
Humanlike torso and arms allow for a joke. On the website, it says this robot is more dangerous than it looks. Like, no, it looks dangerous.
It looks very dangerous. I think this was vibe-coded, and I don't think it's a real company.
Well, it's burning up the timeline. People are having a lot of fun with it. Anyway, we have Lulu Meservey from Rostra in the waiting room.
Let's ask Lulu if she thinks this is a real company.
Yeah. And the value of rage bait, the value of like fake launches.
Lulu, please look at this product, this new robot company, and tell us — give us your read. Is it real? Team, can we pull it up?
Show her the centaur robot. That is absolutely...
I saw — yeah, I saw a glimpse of it.
This is like the stuff from nightmares. Like imagine that you're in a natural disaster and this thing comes to you through like the smoke and the flames. I think you would probably die of a heart attack if it could save you.
Don't worry, human. I will protect you.
Yeah. Really fascinating. I think if they changed the head, you know, maybe you could get it a little bit better, but still canonical demon head. Like this looks like the canonical Satan head when he appears in form on Earth. How did this originally hit the timeline? Like where did this — this just came out as if it's...
Well, the whole world gets their news from prediction market social media accounts now. So...
I guess, but yeah, it did go out on like a PR newswire. I guess Satyres robots, like, to answer the question.
No, I'm curious. I want to have them on now.
They say the horns are necessary because they place cameras there so they can see the position of all limbs and tools. It gives a layer of redundancy in case there's issues with encoders and on the actuators.
That's what will make people like AI more, actually, is that...
The horns don't have enough cameras. But once the horns are for surveillance, then people saw an open letter.
Yeah. They gave the demon robot a chainsaw, and they say the horns are completely necessary.
What do you make of the current startup launch playbook? Because there is a world — like there's all this speculation that Nathan Fielder is working on a new season of The Rehearsal with that boy band called Boythro. Have you heard this, Jordi?
So it's a real band, a boy band that has been detained, and they have a whole — they've been making news. But people have been suggesting that like Nathan Fielder is sort of puppeteering behind the scenes. And I'm wondering how far off we are from a startup just launching something that's like intentionally rage bait or viral, and then sort of pulling off the mask and saying, 'Oh, well, we actually are doing a quadruped. It doesn't look as crazy as this. This was just to draw a bunch of attention. Here's the final design.' It feels like that would probably be a rough go because you've sort of violated the trust of the audience on day one. But do you think the right founder could pull it off?
I think it's a really bad idea to just optimize for going viral.
Because viral comes and goes.
Yeah. Yeah. Why? It feels like it's such a bad...
Viral comes and goes. Like you're viral for like 24 hours, 36 hours, good or bad. You know, I think for Leopold, maybe we could get to that. Like by next week, people could be saying he's the GOAT again. People are just very fickle, and everyone is at least...
No, I do think people have this idea of like, if I can just go viral once, then I'll have a big business. And I think back with just building our show — like we had so many, we had some big moments, you know, you can think of like the Sohan Peri, and there were big guests, and there were things like that. But the amount of times that we needed to go like moderately viral on X to build our audience was like 20 times more than I would have thought that you needed to build a media business. And so yeah, just trying to go — doing something that will get you viral once and sort of in a shortcut kind of like almost cheating, which rage bait is like...
I wouldn't call it cheating, but it's kind of close to there. It's just short-term, and it doesn't actually get you the thing that you want, which is like sustained attention.
Yeah, I agree with you. But it's also not just about viral versus not viral. It's actually about consistency. So if you are consistent over the course of a year, there will be times when you go more viral, less viral. There will be times when there's like some Tuesday afternoon thing that you do that catches the attention of 50 people, but it happens to be that one of the exact right people is among them. You just have to keep doing the same thing kind of forever. It's like getting fit. You don't get fit by eating nothing for two days and then lifting for two days straight. It's like you just have to go to the gym kind of every day for the rest of your life in order to be fit. And the same thing is true of relevance. Consistency in just keep doing it. It's not a one-time thing. But also consistency in like what are you going viral for? It's like Ilya said, what are we scaling? What are you going viral for? Because if it's not going viral for the thing that you...
actually consider your true identity to be, now you've actually just gone in the wrong direction. Now you have to like overcorrect back. And so if you were consistent with the story and then with doing it over and over, that's a lot better than trying to put everything into one moment.
Do you think we are at peak? I forget what the word was. I got asked this recently. It was like, are we at peak transparency in terms of CEO communications? Because we're certainly at peak volume in terms of if I go back and I'm like, ah, I'd love to just like listen to 20 hours of podcasts from Jeff Bezos while he was in the middle of building Amazon. That's not really an option. There might be like a 10-minute hit on Charlie Rose and then a four-minute hit on Letterman or something.
Imagine Steve Jobs on Joe Rogan.
Yeah, that would be it.
Imagine, imagine like Steve Jobs on Lex Fridman for the third time and you're like, ah, that one I didn't actually get through because it's like now we're in hour nine. But do you think we're at like peak, like maximum amount of just founder content and founder like interviews in terms of just the volume?
It feels like this should be the peak, but I don't think it is.
Yeah, I think people consider it to be an arms race where now the peak is the new benchmark. I did, for a fun fact, I learned English from a book called English. So sometimes I still do stumble into like...
She read the book on English.
I did a book. It's called English. I read it and then I knew English.
So forgive benchmark. The new peak is the current benchmark where it's like everybody is loud all the time. So if I'm launching and talking, I have to talk more and louder. But to your point, it's very different from actually learning new things.
So I've seen the same CEO maybe go on 10 different podcasts and I still don't know what they're doing.
Or like is the sixth podcast marginally useful compared to the fifth? There was one week when Demis went on like a whole podcast tour. That's right. And this is no shade to Sir Demis. This is like planned around he released a book or whatever, but it still was like, am I going to get anything new from the sixth podcast that I didn't get from the fourth podcast that I already listened to? And each one of these is like two hours. I'm really interested.
Well, the other criticism there is like that podcast tour and the book launch and everything coincides with them like completely falling behind the frontier. So now in hindsight, it's like maybe you should have been working on...
There was no Rogan or Theo involved because that was the real.
Yeah. No, but do you agree that if you're advising a founder and the choice is between 10 tech podcasts that are going to ask roughly the same questions or 10 varied podcasts where one might be cultural, one might be more political, one might be business, one might be finance focused, one might be technology focused, that having more of a breadth? Because there was a moment where the CEO started going on Rogan and Theo and I was like, ah, this is so fresh and different. It's just a very different conversation for them to have. It's not the same as a serious interview, but it just shows me a different side and I like to see how they interact. Not every CEO is set up for it, but how would you solve that problem?
So I would approach it from the obverse, like almost the opposite of what you said, which is if you have five podcast opportunities in front of you or five whatever publicity opportunities and they're all different, think about who do you actually need to reach? Like what is the best use of your time? And if you don't need to reach the general population and if you don't need to show some fun side of your personality, then it's less about finding a way to show your range and it's more about focusing on what do you actually need. Instead of doing five podcasts that are all different, just cut the other three, the three that are either duplicative or are reaching people that you don't need to. Literally, just don't do them. I think CEOs need to be thinking about what is like minimum viable effort where I can get my vision and persona and build trust with the people that I need to without actually going on a press tour. Because at a certain point, the same way that it wastes my time to listen to the fifth podcast, it wastes you, the CEO's time to do the fifth podcast because the marginal value of one hour of your time there is like $30,000 or more. I mean, what's an hour of a CEO's time worth, right?
The only thing that I would push back is I think the reason that you see some CEOs once they just get on this podcast treadmill and then it's like every week new episode is when you actually think of if it's a virtual recording and they're using one hour of their time and even if all that happens, let's say the podcast is like decent size, right? Like top 25 tech podcasts in the world. If you have a hundred customers or potential customers that listen to that, I would argue and they'll listen to the whole episode. But I would argue that is like well worth the time to go on even if that's the only group that actually... And so that's why you end up... And it's like unless you're in a period where it's like you actually don't want to share anything, then I think every incremental progress is like, is it a better use of time for the CEO to talk to like a warm lead, you know, that's just coming in, talk to your customers or talk to a hundred customers and they listen to you talk for an hour in the sort of like passive sales motion, right?
Yeah. See, that's very true and I agree with you there. And the difference is we're making the assumption that it is one of the top podcasts that your customers are listening to and you're saying something that is going to be worthwhile to them. That is an amazing use of time because you use one hour or maybe on TVPN you use like 20 minutes if you're a busy CEO and then you get clips and clips and clips out of it and you just were able to amplify the equivalent of having a thousand meetings into this setting where someone is able to pull better responses from you than you might have been able to give on the spot. That's true. But what a lot of CEOs also do and founders in general is just try to maximize for volume at any cost. And that leads to going on a podcast that nobody listens to.
