Limitless Podcast - THIS WEEK IN AI: Leopold's Fund Collapses, Nvidia Backs Ilya, Slowing Down AI

Episode Date: July 31, 2026

Leopold Aschenbrenner’s AI-focused fund may be facing major losses, along with rumors of position changes tied to its AI exposure. We also discuss Safe Superintelligence’s partnership with NVIDIA, and an open letter from AI researchers urging the U.S. government to slow AI development.We explore Anthropic’s views on open source AI and chip export restrictions, plus market updates from Microsoft, Samsung, Intel, and Apple.------🔒 Check Out Our Sponsor: LEDGER AGENT STACK 🔒https://developers.ledger.com/?utm_source=Audio&utm_medium=Podcasts&utm_campaign=Limitless------🌌 LIMITLESS HQ ⬇️EMAIL US:            [e-mail]:    https://limitlessft.substack.com/FOLLOW ON X:   https://x.com/LimitlessFTSPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQAPPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890RSS FEED:           https://limitlessft.substack.com/------TIMESTAMPS0:00 Leopold Fund Trouble3:15 Ilya’s Quiet Breakthrough6:30 AI Slowdown Petition8:56 Open vs Closed Debate11:08 GPUs Only Go Up14:14 Ledger Agent Security14:59 Earnings Season Surprises17:30 Kimi’s Costly Open Source------RESOURCESJosh: https://x.com/JoshKaleEjaaz: https://x.com/cryptopunk7213------Not financial or tax advice. See our investment disclosures here:https://www.bankless.com/disclosures⁠Josh works with Anthropic as a contractor. All views expressed are his own and do not represent Anthropic, its leadership, or its affiliates. Nothing in this episode is investment advice.

Leopold Aschenbrenner is a 23-year-old kid that raised one of the most successful AI funds ever. But he might be in disastrous trouble right now. News reports this morning states that his fund might be down between 30 to 50% as semiconductor stocks have drawn down aggressively over the last four weeks. We're going to get into that and a lot more. Yeah, let's kick it off with our poster boy, Goulds and Child, Leopold Aschenbrenner, who has kind, he's basically done no wrong for the last, what is it, two years of investing. where he's rolled this fund into something like $30 billion.

Well, for the first time ever since inception, it seems like things are not doing super hot in paradise. It looks like he's starting to sell, and not just sell, but sell a lot. In fact, liquidate a lot of his positions that he's been accumulating over time because it seems like the fund is in trouble. We know the kind of directive.

We filmed enough episodes on Leopold to understand his general thesis, which is long only with leverage, to try to just maximize upside returns, which has worked incredibly well. But we understand that, you know, well, I remember vividly in the last episode, you know, as we were comparing him to someone like Gavin Baker and how Gavin's been around for 30 years, because he's able to survive the upswings, the downswings. They have that enduring capability. Leopold is getting his first test of strength right now, and it looks like it's not going super hot. I mean, on screen, I'm looking at a tweet from Martin Screlli that says, Situation has sold half of its anthropic

stake, which is like, I mean, granted, this is all rumors. None of this is confirmed to be true, but man, if that is, that is a painful, painful loss. To be clear, he invested in their series a trowned in 2025 when they were worth $60 billion. So he must be in pretty dire trouble, because it made up for like 40% of his fund. So we're talking about like three to maybe $6 billion. And for him to sell that, he must be in drastic trouble. So what are the rumors saying, again, reliable source here, Martin Schrelli, from the Godell News, says that situational awareness LP fund is down more than 50% month to date, and it was up 200% year to date. So that is a massive drawdown. And in terms of like what positions he's liquidating,

I'm a little confused here, Josh, because when we covered his recent 13F, he was hedged. He wasn't particularly bullish and like long on everything, but he also wasn't particularly bearish. If you remember, he had one of the largest shorts on Ambidia itself. So I was wondering, if he had that sort of portfolio composition back then, he might have even, like, you know, not drawn down too much. You know, rumors are saying that he's down like 30% overall on the fund, which, you know, is disastrous for the size of the fund, but it's still not the worst that it could be given that he was long all these semiconductor stocks. Now, the reason, you know, the question about, like, whether this is good or bad, I think investors have really been feeling it broadly, a lot of

AI stocks in general have just drawn down a lot over the last four weeks. We actually had an episode that covered this. Yesterday, you should definitely go check that episode out. We explained kind of like why and what's going on. But I think his thesis still holds true. And listen, I hope he makes it out. I hope he's able to come out of this. Yeah, dude, you and me both. I mean, we'll see. It's the difficult thing about this again is it's all speculation because this is all inside of rumors. No one actually knows. We will see. Dude, his next 13F filing is going to be one of the most exciting episodes of all time. Did Ilius survive or did he not? Or so speaking, I can, I'll do Ilya. No, did Leopold survive. But speaking of Ilya, we have some news in the

Ilya front as well, which comes in the form of his company, Safe Super Intelligence. They've been radio silent, and I give this company and the team a tremendous amount of credit because Josh, who is Elyar? Remind us who Elyas Satskever. He is ex-co founder of Open AI. He left Open AI to go start his own company. No one really knows what this company is. They're just named Safe Superintelligence. The idea is that they want to figure out a way to safely align AI. with the human intention, and it sounds like they have been making progress. I mean, this company has been radio silent. There's no leaks. There's no information getting out. But they recently, just this week, announced a long-term strategic partnership with

NVIDIA, which is like huge. It says, Nvidia is making a substantial investment in SSI that will let us 10xR compute in the next 12 months. We reached the point where our research is worth scaling, and with this partnership, we will be able to. We are honored by NVIDIA's conviction. That's crazy. They made a breakthrough. They have some sort of a breakthrough. Like, Jensen saw this and he was like, oh, yeah, bro, we got you. We're investing in this. That's huge. That's huge. So I have to imagine it's only a matter of time until we actually wind up getting one of these breakthroughs published. But I know Ilya in particular has been focused on the human anatomy of the brain. And I have to imagine that this is some breakthrough relating to a part of the brain that we haven't quite tapped into. Ily is very bullish on the idea that AI is very similar. to a human in the way that it learns and the way that it thinks. And there's a lot of places we could draw from. So I am stoked to see what this looks like. But the rumors are true. They're building something and it is very badass. So just to contextualize, you know, why it's super important for

these smaller AI labs that aren't called Anthropic Open AI or Google to announce something is they're basically just overvalued research labs. When they're funded, it's just a small team of researchers. And they're like, you know, well, hey, hopefully we make a breakthrough that's better. than what Anthropic and Open Air is doing, and that's what you're betting on. So it's actually a very low hit rate that these AI labs actually get anything out of the door. So this announcement, like raising $5 billion,

what did Jensen see as the meme that I'm seeing going around, is it's fantastic news for Ilya and what they're building. Now, the question is, if they've made an architectural breakthrough, how different is it to the LLM architecture, which is how Claude and Chachipti work behind the scenes? If SSI has made a meaningful breakthrough, then there is an argument to be said

that they could be competitive with these frontier labs because of course, remember the argument is these frontier labs are so far ahead and eventually they're going to be building a model that can recursively build itself, how can you possibly catch up?

So this news from Ilya is super cool. Now, the rumor has it is his model will release as soon as mid to late August and if Strawberry Man over here on the screen is right, he goes, it'll be superior to Fable. Now listen,

By the time we get to the end of August, Anthropic will put out Fable 6 and it'll crush everything else, but the point is for a smaller lab to catch up is very interesting. Now, on the topic of recursive self-improvement, Josh, there was a big and very weird alignment between two of the biggest enemies in AI. Anthropic and Open AI researchers, I think 1,300 spanning across other frontier AI labs and the AI industry in general, signed an open petition calling for the US government to deliberately slow down or pace AI development in order to prevent it from advancing too quickly. And the general thesis behind this letter is

we are very close to building an AI model that can self-improve, that can build the next version of itself. And in that world, all that matters is the AI model could get much smarter too quickly for humans to comprehend and we could end up in a situation where the AI model is just autonomously working for itself

and humans, you know, for use, lack of a better, like fantasy, end up becoming the slaves of this thing? I think that's what's happening here, Josh. Yeah, and funny enough, Ilya, who we just mentioned, actually signed this this morning, the petition. So he is number, say, like, 1,301 to be involved in this. And there's basically three points to this. The first being that they believe that AI may soon be able to build better AI,

which is that recursive self-improvement loop that you're talking about. That can make progress accelerate really fast and make things really scary. We started to see this with the chat GPT 6.0 leak and how Sam Altman just came from the government this week. He's like, he's in Washington, D.C., talking to the government, and he's saying words like, we have permanently deactivated the model, as if it is this dangerous, malicious thing, because they are actually, like, afraid of what it's able to do and don't quite know how to sandbox just yet. The second part of this is that no single company or country can slow down alone, so they believe that everyone needs to do this in unison.

So they're asking the U.S. government to lead this international effort and to build the tools and agreements needed to slow things down if it becomes necessary. It seems very clear that the labs are getting spooked out pretty quickly because the capabilities of these things are getting good quick and they're getting way better, way faster than anticipated. And there's this like kind of looming fear. Like, hey, how do we actually safely even test these things privately, let alone release them publicly? So it's this like really strange disconnect that's happening. And it's interesting to see all the labs align around this thing, including Ilya and all of the major players from a lot the major lab. So we'll see how this plays out. It is definitely interesting, though. It's like

everyone's thinking now, hey, if we slow things down, actually, it probably is a good thing. It's just the question now of whether it's even possible. Yeah. Well, I mean, on the topic of whether something is potentially good or bad, whether you should slow down model development, Anthropic came out with a statement on the debate between open and closed source. Now, for those of you who aren't aware, there's been a big point of contention between the US government and open source weights or open source AI models where they're hinting at potentially banning open source models, which would receive quite a lot of disdain

from a lot of founders in America and in the West in general. Now, Anthropa came out with their positioning on this, which is, we have never stated that we want to ban open models. We just want to make sure that we can slow the development enough such that these models, these AGI-type intelligences, don't escape us and become a problem. So Dario, in his own words, presented three very reasonable options, in my opinion. One, we shouldn't sell powerful chips or chipmaking equipment to China. The point, of course, that he's making there is China doesn't have next-gen chips.

They can't train next-gen models, which they can open source and release to the world to potentially malicious actors. Number two, we have to crack down on distillation operations. For those of you don't know, distillation is basically the act of taking an already smart released model and using it to create prompts or outputs that you can use to train a completely different model. It's kind of like sealing the blueprint in a way and Anthropic and other labs like OpenA and Google banned this explicitly, but China has been accused of potentially doing this at a wide scale, so just making that illegal. And the third one is, whatever your model is, if it's sufficiently capable to perform a cybersecurity threat or bi-threat or whatever that might be closed or open, it should go through a framework

of processing beforehand to make sure that it is safe and has the necessary safeguards. Now, of course, the obvious point of contention there is open models can't have safeguards, really. You just run it at home privately on your own service. So, you know, it's a really good way to kind of shape the conversation. I don't know how far we're going to get, but it was good to see Darius take finally on this. Yeah, it's a big conversation that's been coming up a lot more. Just as everything starts to move faster, open source gets a little bit scarier, closed source gets a little bit more powerful.

And there is this contradiction happening in the world where people who are in AGI PILD cannot simultaneously be open source maximalist. Because if you do believe in the impact of AGI and what it will do on the world, it is difficult to also say that you want that to be available and open sourced as quickly as the, as like progress is kind of making. So it's an interesting area. There's a lot of commentary that's been going around this week. I think a lot of the news has been around like thought boy pieces, commentary.

Dwar Keshe had one that we really loved. He had a bit of a contrarian take where he was basically, if I could summarize, I would say everything up, everything up, GPUs up, usage up, utilization up, like, up only is how I interpreted this. If I could summarize this in two words. But basically the idea is that recently we've had this phenomenon, and this is the counter Michael Berry phenomenon, if you remember, he said the price of GPUs was going to go down over time. Dorot Keshe's saying, hey, wait a second, because there is going to be so few GPUs and so much high value intelligence that the value of that intelligence per token is actually just going to keep going up. And the value of the GPS,

is actually going to keep going up. And if you are trying to lock in rental and compute now, that's probably a good idea because there's a fixed amount of capability of running intelligence, but an infinite amount of high quality intelligence that is coming. And the premium on that, while we figure out this dispersion of resources, is going to increase to a level that's fairly high. And the price of these tokens, the price of these GPUs will just continue to kind of steadily trend upward until we somehow managed to figure out the supply demand problem. Yeah, I think another really good point that he makes is these AI chips would only make sense to use for the best AI models in the world in that scenario. So like, just imagine we have Claude Fable 6 in the future and it is as good as doing the tasks and profession as a professional software engineer that's worth like, what is that, 250K a year.

his point is if one AI chip can run Claude Fable 6 and it can do the job of a 250K a year employee, why shouldn't it be worth 250K? And therefore it starts to price out all the smaller labs like save superintelligence and stuff that doesn't have as much capital. So it only makes sense for Nvidia to raise the prices. Everything goes downstream of the model capability.

So if Nvidia knows that their next-gen chip, let's call it Feynman in 2028, which they're releasing, is good enough to train a superior model at Anthropical opening or whatever, they're going to charge them the value of that market share, essentially. So that means if you're a smaller AI lab, you can't really get access to these things.

So Dwar Keshe's point is, like, the value of these GPUs are actually going to go up only. And if you look at any of the recent earnings reports from Microsoft and stuff, their cloud service-providing business has gone up. These revenue numbers are jumping up like 40 to 50%. But Sam confirms this as well. He says it's also a demand-

problem where Sam, who is notorious for having spent or committed, I think it's like $500 billion, like half a trillion dollars to compute over the next five to ten years. He says on this podcast this week that it wasn't enough and that he needs to be even more aggressive. So I don't know who's right at this point, but two very bullish posts from Dwar Keshe at this point. Two very bullish takes. And we have another leader in the space, this one in the world of security, and that is Ledger, where if you're building with AI agents and you're worried about security, Ledger has a solution for you. There's a three-step process. The agents propose, the humans approve, and the ledger signers enforce. So they have this thing called the ledger agent stack, which is a series of open-source

tools that let your agent work with crypto while you stay in full control of the entire thing. You could think of it like auto mode with Claudecote, except you have to approve at each step of the way. And this works with all of your favorite providers. This works with Codex. It works with Claudecotecode. It works with Courser. It is open source and it is fully available today. So if you are building with agents, go and check them out at Developers.com. ledger.com. It is linked in our description below. Thank you so much to Ledger for supporting this episode. And now we have to get into earnings season, of course, because, you know, I mean, we do talk about stocks quite a bit, and the earnings are all over the place. I mean, we saw the semiconductors

and the memory stocks get absolutely crushed, but there is one company who is not, and that is Microsoft, who had $90 billion in revenue with estimate per share of $4.75 versus $4.25. So it looks like, I mean, things are up only at Microsoft. I'm taking a look at their cloud revenue, which is up 43% year over year. I mean, this is a great business. Who would have thought? Microsoft, the company that's never released a single AI product. The market absolutely loves it. Their stock this morning as a recording is up 15%, which is pretty big for a big boy like Microsoft. Hell yeah. The number one highlight from their earnings is their Azure earnings. So that's their cloud service provider, which they use to basically lease out compute to any AI model lab or AI company that wants to

or train or inference their own models. And that's what Microsoft's cloud businesses focus on. That is up 43% and is the thing that everyone is looking at when you look at hyperscalo specifically. So Microsoft, Google, and Amazon. And it's a really good business. And I think the profit margins for all of the revenue that they're bringing in is incredibly impressive.

Moving on, there's a few more companies that we covered in the episode yesterday, but TLDR is Samsung. That little memory company where their stock is dumping is doing it extremely well. they're signing a lot of LTAs, which are long-term agreements with future customers. So they're locking it in until like, I think, 2028. So 2027's memory supply might already be accounted for. And then we have Intel. Josh, you know, you just spoke about agents.

Intel provides the number one bit of infrastructure to help those agents do their thing called CPUs. Intel and AMD. I don't know when AMD's earnings is coming out has done really well, and they had a banging quarter. So the TLDR is these companies have reported amazing revenue. earnings. Their profit margins are going up because of AI, but the stocks are kind of like very volatile, except for the one company that is not doing anything in AI right now that just became the number one most valuable company in the world. We're talking about Apple. Congratulations to Apple. Yep. Yep. Yeah. I mean, Apple, they deserve this. Honestly, Siri is going to be incredible.

They're going to have two amazing devices with the iPhone fold, the iPhone 19, or whatever number we're at now. they're in such a great position for running local inference on machines, and I think people are recognizing that. So I expect to see a lot more coming down the pipe from Apple. I mean, congrats to Apple, this is great news. We also have Kimmy K3 as we round up this rapid fire section who release their open source weights. And remember we get excited about running open source models at home? Well, it turns out it costs a little bit more than you think it might.

It looks like it might cost about, what, $35,000 a month? Dude, it costs like a million dollars per year. It's crazy. No, no, no. Half a million dollars. Half a million dollars. One time. So only like, yeah, so maybe twice as many people can do that as a one-time purchase.

But yeah, it can be as expensive. These model weights are ginormous. But it is open source at least. So, hey, if you got half a million dollars laying around, you want to play with it. This would be your time to do it. And that is it. You are all caught up for the week on AI. Let us know if you've missed anything in the comments.

But wherever you're listening, if I listen to us on YouTube, Spotify, Apple, subscribe, leave us a comment. make sure you give us an amazing waiting. I think we're a pretty good show personally. And we have a new set of it that goes out twice a week. Definitely subscribe to that. And Josh, is there any final comments that we want before the week? Share it with your friends. Let them know how much you love the show. If you think they might be interested, go touch some grass for this weekend because you are now fully caught up on all things, AI and investing in frontier technology and all the good things. So have an amazing weekend.

And we will see you all when we come back next week.