Jordi Visser
The AI Bubble Didn’t Pop. Hedge Fund Leverage Did.
🇬🇧 EN🇪🇸 ES
Ask the tutor about this video
Sign in with a verified email and ask the tutor anything about this video.
TL;DR
Research summary
▶ Market panic and the speaker’s framing
The speaker opens by recounting experience trading through Mexico’s 1994 devaluation, the 1997–99 emerging-market crises and LTCM, the dot-com bubble, the financial crisis, and COVID. He says the present situation is not yet an equity-market or economic bubble bursting: the S&P 500 was 2.7% below its all-time high at Friday’s close, compared with much larger historical drawdowns. In his view, the bubble that has popped is hedge fund leverage, which is undergoing a sustained dismantling. He calls capital the main risk to AI infrastructure and says CoreWeave’s fundamentals are not currently the issue, while allowing an AI bubble could emerge in a year or two. He questions the importance of forecasts about AI infrastructure impairment by 2030, arguing that rates, data-center costs, and compute demand could change; he also says the forecast does not mean CoreWeave will go to zero or that an unwind is inevitable. He contrasts the current market with 2007, noting that credit spreads and financial stocks showed broader stress before the financial crisis, while current international market moves are mixed and the AI trade is particularly weak.
▶ Leverage, crowded trades, and fund drawdowns
The speaker describes hedge funds, especially multi-strategy funds, as leveraged vehicles that depend on diversification, volatility, and low correlations. He argues that information-driven stock selection has become less rewarding: Nvidia’s earnings beat expectations while its share price remained effectively unchanged over roughly a year and its forward P/E fell, a pattern he says has spread across the AI trade. He cites Fed and OFR figures indicating record hedge fund and prime-broker leverage, including gross long-short equity exposure near two to one, and says leverage growth accelerated after 2023. He argues that market structure has changed and that funds may need less leverage to pursue similar returns, though generating a good Sharpe ratio is harder. He cites reported severe losses at several funds, including a 44% fall for a $20 billion flagship AI fund and a reported reversal of Lone Pine’s 44% year-to-date gains; he estimates the latter could imply gross losses exceeding 55% under an assumed 20% fee on AUM. He explicitly says Lone Pine is not in trouble; the drawdown matters because it can make managers reluctant to add risk and can pressure investor confidence. He characterizes the scale and speed of the losses as unprecedented in his career and LTCM-like.
▶ Deleveraging spreads beyond AI equities
The speaker points to sharp, broad moves in crowded hedge fund positions, including sectors beyond semiconductors, and a sudden rise in the Mexican market after it neared 52-week lows. He says hedge fund crowded longs had already fallen in July, but the selloff spread more widely during the current month. He cites a 15% decline over four months in an unlevered crowded-stock measure and argues that leverage could make losses larger; he distinguishes these equity measures from credit stress. He says the speaker’s thematic portfolio fell 11% in the prior week, worse than any week in July, and notes that credit and convertible securities also weakened. A synthetic credit position of long HYG and short IEF had given back much of its summer return; he says this matters for multi-strategy funds because credit losses can affect other strategies. He notes large AI-related capital raises and argues investors may need to sell Treasuries or equities to fund them. Despite the sharp moves, he reports the S&P was down 1.95% for the week, 350 S&P 500 stocks rose, the Nasdaq was down 5.5%, and Bitcoin was down 2.3%. He says the selloff appears concentrated in equities. On his own portfolio, he says the majority has moved into crypto; he still holds Marvell, a small amount of Micron, Nvidia, and Eli Lilly, and has significantly reduced his AI portfolio. He says he may add AI exposure for trading, but expects it not to exceed crypto in size.
▶ The AI bear case and uncertainty about future revenue
The speaker challenges bearish interpretations of AI demand and long-term infrastructure forecasts. He argues that the dot-com buildout anticipated useful technology, but lacked sufficient demand and revenue at the time, whereas AI already has users and GPU rental demand; he says the oil analogy is limited because AI revenue may come from many uses rather than one commodity. He objects to forecasts extending to 2030, saying analysts cannot know how large humanoids, consumer agents, coding, robotics, drug discovery, or other uses may become. Citing Andrej Karpathy, he says most people have little exposure to AI, some use chatbots casually, and relatively few are frontier professional users, which in his view makes confident long-term revenue claims premature. He says agent swarms have only recently emerged and describes them as a potential source of rising token and GPU demand. He cites an estimate that agent workflows are used by about 2% of companies, while over 50% use coding agents, and presents broader workflow adoption as a possible future development. He says he has sold most of his positions in the past and argued that the first-quarter rally would not recur, but still believes AI can generate alpha and that its future business opportunities remain uncertain. He recommends listening to people working directly in AI, including Tibo, Dylan Patel, and others, and describes SemiAnalysis as an AI-focused research business whose revenue, according to his account, grew from $20 million in 2025 to $100 million.
▶ Investment thesis: compute, agents, and AI-linked assets
The speaker argues that AI is changing the investment landscape and that returns from a tokenized digital economy could outpace those from conventional public equities. For the next 12 months, he identifies AI agents and Vera Rubin-related agentic compute as areas he wants to invest in, mentioning a list of 15 names, mostly in optics, without naming the companies in this segment. He says the recent selloff may offer lower entry prices, but waiting for hedge funds to finish liquidating could take another month or more; he also says it could end sooner and that further weakness may continue until hyperscaler results. He advises investors to do their own research and says entry points are improving if there is no bubble, while stressing that it is difficult to know when hedge fund selling has ended. He argues that tokenization may change hedge fund market structure, covariance calculations, and margin provision, and says his process narrows 100 names to 15 using thematic research, fundamentals, technicals, and earnings progression. He cites claims that AI capex could approach $2 trillion or 5–6% of GDP next year, and that OpenAI and Anthropic together could potentially reach $600–700 billion in revenue; he contrasts that with a combined estimate of about $170 billion for this year and says Anthropic was on pace for $90–100 billion versus $9 billion the prior year. He argues the revenue opportunity may extend beyond subscriptions and per-token API fees to knowledge work in coding, robotics, autonomous vehicles, drug discovery, health care, and chip design. He rejects simple judgments based on a single quarter’s growth and says future revenue depends on new products and uses.
▶ Demand evidence, valuation, and the case against premature bubble calls
The speaker says repeated AI-related selloffs have not stopped semiconductors from rising about 50% year-to-date, and argues that ignoring semiconductors is akin to ignoring major technology companies. He advocates long-term compounding and cites S&P annualized returns of 17% over 10 years, 13% over 20, 12% over 30, and 13% over 40, as well as his example that investing $10 each month for 50 years would grow to $478,000 assuming a 2.5% dividend. He says corrections of 20% are possible but argues that being at an all-time high is not by itself evidence of a bubble. He sees public equities as representing innovation, while expecting tokenized investments to become important; he is skeptical that venture capital can reliably select the winners. He argues that long-term uncertainty disrupts hedge funds that rely on short-term arbitrage and leverage. He compares the rapid growth and valuations of AI companies with public-market returns and says AI-linked businesses could grow faster, while cautioning that he cannot predict precisely where revenue will come from. He recommends an interview with Mark Zuckerberg and Alexander Wang, citing discussion of compute, the economics of neoclouds, and the value of older GPUs. He says GPU prices rising makes it difficult to argue that the AI compute bubble has burst, while distinguishing that view from bearishness on CoreWeave’s equity or bonds. He also discusses possible robotics advances, the potential value of frontier models in biotech, materials, and robotics, and the possibility that model providers may limit releases to prevent distillation. These are presented as his views and possibilities, not established outcomes.
▶ Hedge fund pressures and lessons from leverage
The speaker reiterates that the market decline is a hedge fund deleveraging event. He says funds with large drawdowns may reduce positions to avoid further losses, while prime brokers and investors may become concerned. He presents AI infrastructure as a lasting re-architecture of global compute, citing a discussion in which Jensen Huang reportedly answered “forever” when asked about its duration. He contrasts this with the dot-com buildout, which he says expanded supply before demand existed, and argues that AI progress is driving demand and token use. He says capital exists to fund a large AI buildout, but that it must come from somewhere; investors may sell assets to fund allocations and raises. Asked to choose between CoreWeave equity and debt at current levels, he says he would prefer the bonds. Drawing on LTCM, he lists leverage risks: a correct thesis with poor timing, volatility shifts, financing costs, crowded trades, falling liquidity, breaks in correlations, and position sizing. He says these factors can force managers to exit before a thesis plays out, especially when investors judge them over a 12-month period.
▶ Economic risks and labor-market concerns
The speaker describes a distribution of potential business outcomes in which some companies lose growth to AI, a smaller group wins, and others may move between those categories over time. He notes that Amazon, Meta, Google, and Microsoft had reached new all-time highs as a group. He also raises economic concerns: consumer confidence was at its lowest level since 2013, below its level during the pandemic, while the unemployment rate was 200 basis points higher in 2014 than it was at the time of the presentation. He says employment historically drives consumer confidence and argues the labor market may be weaker than economists suggest. He cites shrinking financial and technology payrolls and says health care is an exception, with the relevant jobs measure negative for about 18 months. He expects adoption of agent workflows could create a labor-market effect, which he frames as a profit-margin, political, and possibly Federal Reserve issue rather than necessarily a recession signal. He says the S&P’s earnings growth was then at 29% and compares it with year-over-year job creation.
▶ Crypto outlook and closing view
The speaker closes with his crypto work and says he is continuing to explain tokenization, including privacy, interoperability, middleware, settlement, and smart-contract rails. He describes crypto as a potential distinct opportunity for the following year and reports that the crypto index was down 7% month-to-date while Bitcoin was down about 2.5% since August; he says Bitcoin had still outperformed since June. He calls the market a bull market despite weakness, noting only 20 names above their 50-day average but favorable 200-day measures and slopes. He does not expect Bitcoin to rise until hedge fund conditions stabilize and says it could drift toward the 75 area. He says he plans to publish a paper on a crypto-related figure’s possible $107 billion in net interest income next year, attributing the number to that entity. He ends by advising viewers to stay calm about an AI bubble and asserting that hedge fund leverage will not return as a rising trend.
Full transcript
[0:00] We got a lot to cover. A lot of slides, a lot of market internals. Hopefully, I'm going to be able to shed some uh uh some experience on what have you. See, I'm I'm hoodies up this week because I don't know if you guys read the news, but the air bubble has popped. That is the news that is coming out at the end of last week. And I have never received more panic emails from people I've known over the years in terms of
[0:31] does this remind you of '08? Does this remind you of '01? Is this the same type of situation? And the answer for everyone is uh my career began in 1994 with Mexico. I trade started trading a book 2 months before the devaluation. I was at Morgan Stanley. I was a young person involved in a market going through things I've never seen before. Then I was part of '97 and '99 in emerging markets trading everything through the EM crisis, including the closure of LTCM was on both sides of their business, meaning on
[1:01] the other side of their trades as they were doing things in Brazil, and then part of the shutdown, and then a bunch of them worked at Morgan Stanley, and it was like working with the greatest people I've ever been with because I got to ask them, "How could you how could what happened happened occur?" And you notice LTCM leverage lessons. I'll be going through those. I came back to the US. I took over the S&P. Then I was there for the dot-com bubble, managing ETF business, index businesses. I was a part of all of that on the other side of billion-dollar
[1:31] transactions every day. Then I traded a billion-dollar macro book throughout the great financial crisis, took money out of the banks weekly and put it into gold. I have traded all kinds of things and been involved as a CIO during COVID. I bring this up because this is not a bubble popping. It might be in the future. It's not now. So, with hoodies up, I'm going to give you guys a lot of different things to go through. Again, I'm doing this for your information,
[2:01] your education, and hopefully not this week. We didn't make money this week, but over the long term, the goal of this is to make sure that in weeks like this, you're paying attention to the stuff that's actually happening and not to the noise that's going on. There is a bubble being popped. It is not in the equity market, and it is not in the economy. This is a true number. This is the rolling drawdowns of the S&P 500 over time. Here is the Great Financial Crisis. Look at that, almost 60% peak to
[2:33] trough. Here's the dot-com bubble. Eventually got to 50. Here is COVID, got close to 40. Here is when the Fed's raising rates. Here is the tariff tantrum. Here we are now. We are 2.7% off the all-time highs in the S&P 500 as of the close of Friday. Guys, please. Here's the bubble that has popped. I will go through this. Hedge fund leverage has popped. It will never ever ever ever ever, in my opinion, exceed.
[3:03] And what we are going through is a constant deleveraging that needs to happen. CoreWeave, not a problem. Hedge funds, big problem. I'll go through the reasons why. This is multi-strat leverage, which peaked, and this is from the Treasury. Peaked, started to decline. You're going to see more of this. I'm going to go through why. Hedge funds leverage is in a complete and utter dismantling.
[3:33] There will never be a time where you will see it go up, in my opinion, based on the things that I'm going to show you, because at the end of the day, you have to provide a sharp ratio that makes sense. And we are seeing extraordinary moves. And remember, as I go through the leverage, the leverage in the hedge funds dramatically bigger than anything going on in AI right now, not even close. Here's the Coreweave chart. This is where we closed the week. I just want to show this because this is the fifth time the equity price has been here. Every
[4:04] single bubble I have ever traded when it has popped, the equity is down massively in the bubble popping situation way way way way further than this. By the time we get to the bubble popping, when you see the bubble popping. And again, maybe there will be an AI bubble in a year, in 2 years. We definitely are constrained like I said last week, capital is the biggest risk. We continue to tap the markets for more capital. This is not about non-payments, this is not about not earnings, this is not about GPU
[4:34] prices falling. So, we're ahead of the game on the fundamentals, which means like there's been in Coreweave where we've had these panics, we get a bounce and we get a bounce. Now, the bounces seem to be getting shorter and I don't have a problem like I said on the weekly video for subscribers with the equity price of of Coreweave building in a higher risk premium. But, look at what started this whole thing. And again, hedge funds were losing money from July in a historic fashion. They got a bounce somewhat in September,
[5:04] but as I'm going to show you, some news came out after I did the video last week that shocked me. I can't believe it wasn't more of a story. Gartner says over 3 and 1/2 trillion of AI infrastructure spending could be exposed to impairment by the end of 2030. First of all, we don't have this number yet. So, I I just want to make sure you realize what this is saying. I've never in my history of trading markets seen a forecast out to 2030 that anyone cared about. We're talking about 4 and change years from now could be exposed. And yet, people sent
[5:35] this to me going up, this is a problem for Coreweave. Coreweave might be in trouble then, but we have no idea what's going to happen. If interest rates go down, it's not a problem for them. If the cost of memory and everything goes down to building a data center, but the compute demand stays high, you can have massive endless compute demand. And as Jim Chanos has showed, it may not be a good equity business. It doesn't mean it's going to zero. It doesn't mean we're going to have an unwind. Here's a chart of the S&P, again, sitting near the all-time highs.
[6:05] Here this one's also astonishing. So, here are Friday's returns around the globe. So, the S&P's down 1%. Big deal. The VIX has been trading 16. So, we're barely getting one standard deviation falls. It had been like 100 days that we had gone without getting a 1% move. We had two of them last week. The Nasdaq's down two. On the same day that that's happening, look at these. I mean, again, Brazil's up, Mexico's up, Canada's up, Europe's up. Every single
[6:36] one of them is up. If you go look at the last 5 days, S&P down, so this is for the week, down 2%, down 4 and 1/2. Again, Europe's up across the board. Look at Mexico. Uh Nikkei's down. The AI trade is what's down. Again, the AI trade is down. So, year-to-date, here's where we they are. The S&P's up 12, Nasdaq's up 13. Again, bubbles popping. If people are right on this, they are the best forecasters I've
[7:06] ever seen. So, in January of 2007, just to give you one example, you can do this throughout the dot-com bubble. It did not become obvious what the problems were in the dot-com bubble until well after the equity market had started to fall, well after. In the case of here, we were getting defaults already. Like, the housing bubble was unwinding in 2006, 2007, way before. Probably 2007 is the better one. I think March of 2007 is when I associate as when at the fund we
[7:38] were at, we were short, we had the big short trade on as well. We started seeing these things which we knew were horrible mortgages that were being written. As I said in my subscriber thing this week, when I go through the core weave clients, remember the clients of the subprime mortgages? The story and thing strippers in Florida were getting mortgages. So, again, this is not the same scam that was done. There's demand for all of this. Here's AIG. You can go back and look at AIG stock price. I
[8:09] advise you to go look at Sienna stock price, which is another AI chart right now. But, when this thing finally fell, again, before we got to Lehman Brothers, which was in here, all of financials were down massively. The credit spreads were widening. Here's where we finish when people said, "How do I watch credit? How bad is credit?" Here is the high yield side of option adjusted spreads within side the cash bonds. So, again, are we doing anything other than what we did across
[8:40] so many different times here? Of course not. Is this COVID? No. Is this the fracking situation in by the [clears throat] time we got again to a serious problem in oil? No. And again, it's not to say that we can't get up to those levels. We can. We're not there yet. And now I'm going to take you through what this is from my perspective. I was heavily involved with Long-Term Capital Management. For those of you who haven't read this book, it's probably a good time to pull it out and go understand it.
[9:11] Hedge funds at the end of the day are leverage vehicles. They are making bets to some degree on volatility and correlation amongst a diversification of assets. For the most part, a multi-strat in particular, it's across different types of strategies. Hundreds of them in some of the largest places with thousands of investment professionals hoping that there'd be a low correlation. The problem is, since I've been involved with markets, hedge funds have focused their attention
[9:41] for the most part, the way that you really get diversified is by having high trading activity, which means you depend on stocks to move based on information. So, if you get really good information and you think companies are going to beat the quarter cuz you've worked really hard, you've done all the channel checks, and you've done things that a retail investor can't do historically, and they beat the the quarter, you should make money. Well, again, I show this all the time.
[10:11] Here's Nvidia. The stock price is effectively unchanged since October of last year. 1 year, it's effectively unchanged. Their earnings have blown away consensus in a way that nobody could have expected, and what we've had is a complete crash in their PE, which is now down off next year's earnings to 15, and depending on where you go, if you take the higher area, it's probably closer to 10. That's what's happened is as they beat numbers, the stock's not reacting that
[10:42] way. And it was fine when it was just Nvidia for most of this year, but it has spread to the entire AI trade, and this is why when I said back the fireworks show is over, this is directly what it was referenced to. You will never go through what you went through in the first quarter again, ever again in public stocks, ever. As far as I'm concerned, we have destroyed terminal value. That terminal value means that any company's earnings past 3 years, we have no idea whether they're going to survive or not.
[11:12] The hedge fund industry is at all-time record leverage. I'm going to move this over so you can see it. This is from the fret the the the Fed, and this is through Q1 of this year. Okay? All time, but more importantly, look at the leverage effectively from 2018 to 2023, it was unchanged. We've now gone up dramatically and more than doubled. The leverage provided by the prime brokers of the banks, another place where crypto is going to have an
[11:43] impact, is at all-time highs. Here it is from the OFR report. Breaks it out by category, multis. These are big numbers. Here's long-short equity. Close to two to one. You're borrowing money. So, hedge funds make money on leverage. And there are multiple ways that you can use when you use leverage. When you're not taking directional bets in the market. And again, the net exposure across the board for multis is very low. For long-short equity, it's been on the
[12:14] decline for most of this year, but the gross has remained high. Now, this is all through Q2. We know that July must have had an impact and these numbers must have come down because the vol has gone up so significantly as I showed on the factor side. It just means the instability within side all parts of the market and most importantly, there are very limited hedges for the AI trade because it is a cross sectors. There is no S&P put, as I will show you. Here's the leverage growth as I'm
[12:45] showing, 15% in hedge funds and then the BDs, which again, the broker-dealers, heavily involved with the prime brokerage, massive growth. When you're looking for leverage, like I said, we're talking about 40 to 60 billion in core wave debt over the course of 6 years. Hedge funds have money, as we learned with situational awareness, that can become an issue in a matter of weeks. On that point,
[13:16] we all forget and everyone started bringing out X jokes, like this was the issue. This was just one of the issues in July. So, yes. To get a number here, again, we're caring about CoreWeave whose debt is spread out over the next seven years, six years. And at the same time, the clients they have are the biggest clients in the world with some of the best balance sheets in the world. Companies like Microsoft and Meta
[13:47] dominating their stuff. Are they not going to get paid by them? Is that really going to be an issue? Are they stripper-related? Here is what happens. These are the numbers that were reported by the Wall Street Journal that everyone kind of backed into. I don't really care whether or not this is for illustrations purposes. I do know one thing. They're not far off. So, if there's $11 billion of equity and $45 billion of positions, the borrowed capital
[14:17] is 33 billion. And as long as you're making money, there is not an issue. So, hedge funds are the problem, guys. Mark my words. They are the problems. In my world of terminal value being eradicated in the bank apocalypse and the pushback that I get, the pushback I got with Micron last year, this is a scenario that hedge funds are not being honest with themselves about how much the market structure has changed. And it's not just hedge funds that are
[14:47] feeling the pain. Mutual funds are feeling it, too. That's why I did the demo that I did last week to show the types of market structure changes that I'm trying to help people with on how bad this is going to be. You don't need as much leverage to make the same amount of money as we're going to see. So, the leverage is going to come down. That doesn't mean the hedge fund industry won't be producing the same types of returns. I think it's possible they can generate similar returns. The money might aggregate even more into the the bigger places that have an advantage on the
[15:17] risk management side, but I definitely think the market structure has changed making it more challenging to have a good sharp ratio. So, Vara, another multi-billion dollar hedge fund, $20 billion dollar flagship AI fund, the losses paired the year-to-date returns from to 65% through July. So, again, they fell 44% in a month. This was in August. So, this is not now, but these types of drops, I don't think people It wasn't just situational awareness.
[15:49] This one happened and came out after the bell last Friday. So, I didn't include this in my video. So, if you asked me to name the 10 best non-multi-strat hedge funds of this century, Lone Pine would absolutely be in there based on my experience with them, based on the way people talk about Steve Mandel. Just an incredible hedge fund. So, this is not even saying anything negative about them. They were up 44% and it erased the gains.
[16:19] It is the staggering drawdown that happened in a quarter. So, to be down, to give up 44% gains, assuming they're charging 20% on their AUM, that means that they lost over 55% in gross returns. That's That's a drawdown that again, I showed you in the S&P 500, that's equivalent to the great financial crisis drawdown by a hedge fund.
[16:49] One of the greatest hedge funds of all time. Here is their returns that I have basically on public data through Q2 relative to the S&P. So, basically, they move a lot with the S&P. Based on what you're seeing over the course of the last 3 years, there is a high correlation to the S&P returns. This is what happened during the quarter that they lost 55%. This kind of stuff has never happened
[17:19] before in my career. I've never seen anything like this. This is LTCM-ish. That's what it is. Now, uh again, that was through September. Yes, this month has been a disaster. The reason I'm bringing this up is not because Lone Pine is in any way, shape, or form in trouble. That has nothing to do with this. This has to do with when you're short-stacked and you have that kind of a drawdown, you're already figuring out what's going to happen, what your people are going through, what the investors
[17:49] are going to say. There's a lot of work that goes on. And so, to put on risk at that point, it is is just very unlikely. And more importantly, to deal with any more losses come up, and that's what I would say is going on right now with inside the marketplace. You're seeing things. Jared Kubin is is someone who's great to watch on the quant side and just to highlight. But, he bloody last 21 days, this is through the middle of last week, and it's just showing how many names are up violently in terms of sectors. Here it is here.
[18:20] Again, what's down? It's not just the semiconductors. It is fertilizer and crop inputs. How often did we hear how easy it would be that we'd have food shortages and that uh these are all hedge fund trades that people had gotten into. You obviously have short SAS up here. You have short software up here. You just have a lot of different components that are not just related to that one thing. And the reason I know that's the case is this kind of stuff
[18:50] also doesn't happen. This is the Mexican stock market. Trading at 52, well, close to 52-week lows at the end of the quarter. It has gone straight up. Here is the monthly or the weekly return of the Mex Bolsa relative to the Qs going back over the last 3 years. Here is the chart of the Goldman Sachs crowded hedge fund longs relative to the S&P. So, again, this month here
[19:21] Vara down 44%. Situational awareness down I don't remember how much they were down, 60, 70, whatever to get to the to get bailed out by Ken. We barely got a bounce here. And this is the most important thing is the hedge funds, these are crowded long, so this includes the fertilizers, includes everything. It just never got a movement on it, very similar to momentum, which I'm not going to bring up. Here's what's happened so far this month, and it's been a straight line down. So, again, these are the equity positions that hedge funds are holding
[19:52] that are dominated by the AI trade because that's the only thing that works. That is about 25% of the market, which I've talked about. Everyone's in the same names. So, it's still up big for the year, and as I'm going to show you, semis are still up about 50% year-to-date despite all of this going on. Nvidia's PE has collapsed. This is deleveraging, guys, and this started in here, and then as I show you
[20:22] all these hedge funds that had losses, and then you get these bounces as we head into hyperscaler earnings and the midterm elections. So, if you ask me what the pain trade is, the pain trade for this is to continue into the hyperscaler earnings and maybe all the way into the midterms. Take everyone out, they will not take risk on, and then in the end of the year, assuming I'm right and there's no AI bubble and this is just a positioning shakeout, you get a snapback. Here is the hedge fund
[20:52] that that chart now with monthly numbers. So, in July, it was down 10%. So, the crowded hedge fund names relative to the S&P. Hedge fund leverage was down already when this occurred in 2008. That's the thing I want to make sure you know. This does not include leverage. This is an unlevered number. This is just market neutral the stocks that are the most crowded according to Goldman Sachs relative to the S&P 500. So, these are
[21:22] market neutral, dollar neutral scenarios. There's beta in this to some degree, but as you can see like over time there's the movements this is an outsized move and again this is the Lehman post-Lehman Brothers move. So, hedge funds were already delevered at this scenario. Right here not delevered as much as they should have been based on all the metrics that I got. Deleveraging has come down. Here are the rolling four months. Again you're talking 15% over the last
[21:54] four months of that chart. Again, unlevered. So, if you add in the leverage that we've said, that would mean I mean a significant amount of losses. Here's here, but that really is all equities. So, I just want to make sure you know everything I'm showing you right now is equities. The real problem came in when it started to spread this month. That didn't happen in July. Here is the AI, the Goldman Sachs AI winners versus losers, which is the orange the white line and then the other line is that Goldman Sachs hedge fund. So, you
[22:24] can see this trade is basically at the end of the day it's the AI trade that every hedge fund has on. Everybody that it's it's the pair trade that everyone has on the winners versus losers and because they have a month-to-month bogey, because they have a drawdown level, because the majority of hedge funds don't have five-year locked up money, they've got short-term locked up money, they can't have drawdowns without their business becoming a risk. And I think this is the puking of a lot of businesses that had
[22:54] significant drawdowns that can't handle it anymore. Here is my thematic portfolio for the week. Last week was down 11%. This is worse than any week during July. The month isn't as bad as July, but the week is. That felt a lot like capitulation, but here's the place that it started to really become an issue. So, this is where you really start getting into the pods. The pods, the big multi-strats, they're going to be losing on the equity long short like everyone,
[23:24] but when credit starts to become an issue, that's when it really gets bad because credit starts to impact a lot of other strategies. I heard biotech was a major, major problem in September. Again, the weakening kind of spreading the contagion within side pods is always a big story, but when credit has this kind of week, this is HYG relative to IEF, which was sitting near all-time highs at the end of the quarter, which meant if you were running a credit strategy, everything was great. And this
[23:54] is a synthetic credit strategy. If you're just long HYG, and then you're short IEF, which is a treasury, that's theoretically taking out your rate exposure, and the only thing left is the credit credit side. And so, if spreads stay unchanged, you're just collecting the differential. So, it basically, since February, went up in kind of a straight line. Well, this gave back effectively all of the returns from the summer. So, in the same way that we had the July numbers, look what happened
[24:24] in July? Nothing. Nothing happened to credit. Here it is on a a weekly basis. Not anything like what happened with the tariff situation, but still, a big week. Here is the convert index. So, converts have joined the party. Why? Well, converts ex- extend from CoreWeave, uh MicroStrategy. It's a very concentrated world as well in in in in converts, but again, if the only
[24:54] place raising capital is AI, which is what's been happening. The size of the capital raises for AI, and not just SpaceX, not just Anthropic, not just OpenAI, but Broadcom coming to the market with over a hundred billion dollars this year. Google coming to the market between equities and bonds for over a hundred billion. We're talking about massive massive capital raises, which again the market has, but if you come to the market, the cash that people have might be from maturing bonds. They have to sell something to go
[25:24] buy things. People don't just have cash in most cases just sitting around endlessly. They got to sell something. Part of that was treasuries. Part of that ends up being some equities, whatever the case is, but converts have now started to get hit. And again, they were getting hit in July. They do have an equity one, but this is a very different fall. This is more of a waterfall. This was more of a reluctant fall that immediately bounced back and and made back most of the fall. Um the VIX did have a big move higher. Again, we're still at at at fairly reasonable
[25:56] levels, but the VIX was up 25 points in a week. VIX is the vol of the VIX. So people are buying call options. The VIX itself closed just below 20, which again is shocking that we could still be below 20, but that's like I showed you, the S&P is not moving that much. Uh the Qs were down 5 and 1/2% for the week. The S&P, like I said, down 1.95, and over the last year that's one of one, two, three, four, five, six seven moves of a similar move. Um during
[26:29] the week, 350 of the S&P 500 stocks were up on the week. So just like in the first chart, this was a unwind where we had been watching as equal weight and breadth had been getting hit while cap weight had been doing well. Well, that unwound in a vicious way so far last week. IWM M about 3%. Again, a big week but not uh one thing I will mention, it's been 6 weeks in a row for small cap and eight of the last nine with one little
[27:01] blurb, one little minor move higher. Uh Bitcoin for the crypto people didn't budge. Down 2.3%. So again, everyone who's continuing to say that or in the past has said Bitcoin's too volatile an asset to own. For everyone watching this that's deciding how much should be in their portfolio I'm going to say this over and over again. The majority of my portfolio moved into crypto. I'm very, very happy about that. I still have Marvell. I
[27:31] still have a little bit of Micron. I still have Nvidia. I still have Eli Lilly. I still have names like that, but the size of my AI portfolio is still significantly down and as I've said, it will never be as big as crypto again. Almost certainly based on my views and what's going to happen. There may be times where I add to AI for trading purposes. But the way Bitcoin is trading and hanging in there when everything else is blowing up this is a different kind of story as an asset class. The move index
[28:02] rates, you didn't get a big move in there. So this was really isolated to the equity market. So now let's go through the AI bear talk. So again, this is something that A16Z put out and I'm going to keep saying it. This index people have no idea how to read it. They just continue to say things about it. Then they mock out Jevons paradox. No matter who you talk to, no matter where it is, if you lis- listen to Jensen Huang
[28:32] and I guess people have this viewpoint that you can't listen to any of the people involved in it because they're going to lie to you. This is not 2007 guys and this is not 2000. You actually need to listen to this cuz here's the one thing I will tell you with absolute certainty that occurred through the dot com bubble. The dot com bubble was built to have things like your smartphone, to have your app store, to have all of those things. That's what it was being built for. It ended up being right. The problem was we didn't have any money coming in to offset it. So, everyone's
[29:04] trying to find a way to create an image that the money's not coming in and yet GPU rental for the old ones, not the new ones, it just continues to sit there. And I'll as I'll show you This is This is a completely different world. So, Kuppy, who wrote a piece last year that I highlighted in terms of the ROIC risk, and I I bring this up I don't know the dude and I really don't care. This is not to pick on people. This is just basically to say people believe to compare this 100%
[29:37] to in this case the shell situation. Shell and open anything re- regarding intelligence. There is a similarity, which is the price goes down and the build out depends to some degree on the prices being up. The token prices are not the way to do this. The revenue is because in the case of oil, guess what? If oil goes down, well then revenues
[30:07] have to go down. In the case of AI, that is not the case. And so far, that has not been the case as this has gone down. The revenues can come, as I'll show you, from so many different places that this is not a one commodity thing. So, to compare it to oil is honestly, aside from the hey, this price is going down, I don't really use this thing that much and I'm not saying he doesn't. I'm just saying the majority of people, based on the data that I have, don't use it. They don't see any benefit for it. They
[30:38] don't think solving cancer matters. There's There's money with that. They don't think that solving math problems that go to physics that allow us to do things to make more efficient that no one would want to pay for that. They would like to pay for open source and cheaper models. It's just not true. I don't know what's happened to Carl Quintanilla, but it seems like he's just having a fun time putting out um bearish stuff. I mean, almost every It's almost like he's become a a bear. Must be some kind of way to get more people to pay attention. I have no idea, but I've been yet presented a credible
[31:08] framework as to where this money is going to come from. The reason I'm bringing this up, and this is not from Carl, this is from Barclays desk. The reason I'm bringing this up is nobody in 2009, so this is after the dot com bubble, nobody knew this was coming. Nobody did. When the iPhone came out, nobody knew where the revenues were coming from. And right now, the one thing I know with certainty is Barclays desk has no clue as to where the revenues are coming from. Where the revenues are coming from has a
[31:40] huge impact. Do they know where humanoids are going to be? Do they know how big consumer agents are going to be? I added everything here. We don't know. So, to say in 2030, which again is what we're getting to, that the revenues won't be here is just impossible for them to be right on or for the market to trade on. So, you follow the data and the extrapolation of the most recent numbers. Everyone else is jumping forward and saying, "I've been through this before. This is what it'll look like. This is too much being built." It's just not the case. So, here's Andrej Karpathy this week
[32:12] saying what I completely know, and the reason why I don't listen to macro podcasts anymore because I'm listening to people speak about something they know nothing about because they don't use it. So, he's talking about the fact that there's two groups of people. Most people on the planet have little or no exposure. The remaining of them, and not a high percent, the remaining people are casual users using it for chatbots. There are very few frontier professional users. So, how could anyone
[32:44] be able to say with certainty what's going to go on? This is far more complex than oil. This is more demand right now than the dot-com bubble ever saw on the build-out. So, again, I can keep showing you guys this, but this is coming from Andrej Karpathy. He's been honest throughout the process. He's the one who got it wrong. Agents weren't going to be here a year ago. They They're not only here, it's gone through. So, the AI debate the people on X that are bearish is being conducted by billions of people
[33:15] who have barely experienced the technology, millions who are beginning to understand its economic power, and a few thousand who are already living in a future the rest of the world cannot yet see. That is what exponential innovation is. Aaron Levie, if you want to know where all the tokens will get consumed in the future, it's agent swarms. We only started having agent swarms in the last 2 months. That's why we've had consumer agents, and that's why I keep highlighting. We are at a step-up function in demand right now, which is why you're seeing more and more interest
[33:46] in GPUs at this current time. DRAM prices made new highs after consolidating for weeks. All we're seeing with prices on demand is going higher. And that's coming because there's more money showing up in terms of demand. They're booked out as I'm going to show. Uh Patrick Collison talked about agents. I'm not going to spend all this time on it, but thanks to personal agents, we will probably get somewhat more rational in how we engage with companies. Basically, you're getting to the point
[34:17] where companies are starting to see the ability to use AI agents, and that's coming across. I just listened to a pod uh an interview with AI engineers this week, where they estimate at this point that agent workflows at companies, agent workflows, are only being used by about 2.2 or 2% of the companies. 2% of the companies. Now, over 50% are using coding agents. But, using agents for your workflows, which is when the job
[34:47] situation starts to deteriorate, which I'll talk about later, that is a different story. So, when you're thinking in bets again, and you listen to people, the prospect of a bet makes us examine our beliefs. The the whole point of this is in almost every case with people saying they're bearish, and especially people that have been bearish for over a year, which is almost everyone. I challenge you to go into X, find someone who's bearish right now, who wasn't bearish a year ago, 2 years ago, 3 years ago on the bubble trade. And not people that
[35:18] aren't taking a short-term, I think the market is overdone here and going down. That is the reason why I wrote the mid-cycle slowdown in May. That is the reason why I sold that of my the majority of my positions. That is the reason why I wrote the fireworks show was over, and we will never see that again. We will never see what we saw in the first quarter again. So, I'm being rational with the fact that to make alpha, you need a gen to compute. Everything else to me is up for risk. So, unfortunately, when you're listening to people say there's a bubble, by definition, they can't be Alexander Wang. He hasn't been through one.
[35:48] You can't listen to him talk about it. You can't listen to most people in AI talk about it because they haven't been through it. Now, Gavin Baker has, and Brad Gerstner has, but people choose to ignore them. The reason I bring this up with what's going on is this people should listen to this week. This people should listen to this week. This is Tibo from OpenAI. These are all ones I listened to this week. This one absolutely you should listen to. Dylan Patel
[36:18] absolutely has the most and best information of any research place on the street, because he has set up his business to be vertically integrated, meaning he covers only the AI situation. He covers the data centers. He covers the semiconductors. He covers Anthropic and OpenAI and models all this stuff out. They spoke this week. Go listen to them. Go hear how things are. This was on Alex Kantrowitz, who's great. The problem is for people, Alexander is 29 years old. Tibo is in the early 30s.
[36:50] It's always good when you can find someone who's so mysterious that you can't even find his proper age. Dylan Patel is 30. When you're listening to people say the AI bubble has broken, they're not wearing a hoodie like me. They're telling you I've been through this. You don't know what you're missing. These kids don't matter. They've never seen this. The reality is these guys have built good businesses and their research has been dead on. Dylan Patel founded the company, SemiAnalysis, if you don't know this about it. Has 90 employees. It's in 11 different countries. In 2025, they had $20 million
[37:21] of revenue. Now they have $100 million in revenue. 90 employees, guys. This is the way AI is changing the world for real. So he not only knows this from his business being uncovering it, unlike the public banks, he's spending millions of dollars on AI and he's openly talking about it. So it's vertically integrated versus Wall Street. Exponential AI needs a new system. Why this wins. This is what my belief is is
[37:54] that all public companies are at risk. That's why I did the Stripe versus Citibank thing during the video last week. You're going to see this over and over again for me because we went from the industrial economy to the digital economy. From an investment standpoint, you only wanted to invest in the digital economy equity. Well, now no more private equity, no more private credit, no more VC, no more bonds, and no more public stocks. No more. You'll get some returns in equities for the next whatever 3 years, but the
[38:24] returns will be dwarfed in terms of the tokenized digital economy. That's what we're going through. The tokenized digital economy, places like this business. He has a hundred million in revenue. It's a big business, guys. That means his business I I I don't know where they would put it right now, but I do know this. You're probably talking about a market cap, I don't know if he went to market, billions of dollars. That's how fast these businesses are growing. And what he talked about was a year ago he was spending a hundred
[38:54] thousand dollars on AI. As of August it was up to 10 to 11 million. So, again, you're really talking about someone who's doing the work on it, but is also using it against his revenue that's come in. So, the question is how should you think about it? So, I wrote this. I went through it. People are not looking to make money right now. I'm just telling you the thing that I put out last week or earlier this week when the cost of intelligence collides with the demand for intelligence.
[39:24] AI agents and Vera Rubin are the place you want to be investing in for the next 12 months within side the agentic compute. And I put together those 15 names. Most of them are in the optical side as to who benefits. This week has been a blood fest, so you can go in and you're getting things at much lower levels. If you want to wait and wait for stability and wait for the hedge funds to be done liquidating, that might take another month. Um I don't know. It could end quicker than that, but based on how bad the numbers have been, would not surprise me to see more
[39:55] of this until we get into the hyper scalar numbers. Uh I did want to make sure that if you guys are interested in Jev, I had a lot of people that reached out. What is Jev? This video was fantastic, had over a million views uh as of last week. Very simple. It'll be in the uh all of these podcasts because there's so many of them. I still have I think two more to show you guys. I listened to at least seven or eight this week. All on the AI chip trade. Again, there's no way to forecast the exponential. So, you have to deal with
[40:25] the fact that you're going to have these ups and downs within side something that's going to go on. That's why in this week's AI video of what I did, I went through this to kind of give people the ability to dip jab. I know I did a lot of over the last 2 weeks, a lot of videos, a lot of stuff came out. The Vera Rubin report, remember you can go get this off the website. You can do this yourself. I did the Agentech consumer inflection. You guys can go do this on your own. This is the time to be doing your homework on these names if you're not in the position the hedge funds are. So, if you're sitting at
[40:55] home, even if you've had a drawdown in terms of things going, your entry point is getting better assuming there's not a bubble. And that's my job here is to basically go through and say we're not in a bubble phase, but we are in a deleveraging phase. It is a very, very difficult game to know when the hedge fund pain is gone because when you're dealing with $14 trillion of of positioning there and the growth has been enormous, it could take a while even if the the fundamentals are still
[41:25] strong and that's part of the leverage thing. I also included the technical side. Again, you guys that are new to it, I'm showing you this if you didn't go through it because there's a lot to use each week on this stuff. All of the technical sheets, all of fundamental sheets, all of the stuff in AI you want to be on top of. And again, I go through the different ways the resources and the stuff that are there. I give you the prompt on it so you can go through it. And that's basically what it does. It's taking 100 names, it's going through, you're going through the Jeff process
[41:55] which again, go look at the video I put on the subscriber site. It will show you how you can do this. And effectively, you are doing what hedge funds are doing and that's the other reason why the market structure is changing. The competition for hedge funds in terms of the speed to ideas is gone. I am about to list something and put money into something with my crypto. So, the tokenized world is beginning, guys. The market structure has forever changed. It is a deleveraging event. You're going to hear me talk about this more and more. Tokenization is the final
[42:25] part of the deleveraging stage for hedge funds. So, hedge fund risk managers I actually don't want to talk to the risk managers. Hedge fund senior people that want to reach out and talk, I'm happy to go through at least my thoughts on it so you guys can go through it. But, tokenization is going to change the way covariance matrices work and the way that margin is going to be provided. So, again, I go from 100 names down to 15 using this process. You guys can set it up yourself. And that way you are effectively doing what hedge funds do, taking thematic research, whittling it down, combining
[42:56] it with fundamentals and technicals and earnings progression, getting an answer within seconds. Watch that video. This is the one that you need to watch this week aside from the other two that I'm going to go through. But, to get back to it of his side in there, AI capex is already bigger than most people think. He says this could be close to $2 trillion or 5 to 6% of GDP next year. The big debate in whether revenue can catch up the infrastructure. He says OpenAI and Anthropic alone could potentially reach 600 to 700 billion.
[43:27] The reason I wanted to show this, they're going to finish combined this year somewhere around 170 billion. All he's saying and again, this is the the thing to me that I just don't think people understand. So, everyone was going through the Anthropic thing this week. So, Anthropic is on pace for 90 to 100 billion for this year. Last year they finished the year at nine. At nine. So, they're 10 times again and 10 times the year before that and 10 times the year before that. And right now he's saying combined for OpenAI and Anthropic
[43:58] maybe 600 700. So, let's take the bottom end, 600. Well, that would mean between the two of them they're doing about three times. So, he's talking about a rapid deceleration. This is already built in he's saying 600, 700, and that's basically what David Sacks and Gavin Baker and Chamath gave as their answers about a month and a half ago. Sorry to the guys who are focused on the way oil trades in the dot-com bubble they went through. If you're not in Silicon Valley
[44:28] and you're not actually talking to the people heavily involved. Brad Gerstner's an investor. He was at an investment at the investment day with Anthropic this week. He saw everything. He spoke on TV. I I I just don't even know. Um the biggest revenue opportunity may not be today's token pricing. Jordan says people are too narrowly focused on cloud cloud subscriptions. This is a critical thing. And per token API revenue. If frontier AI systems can perform knowledge work across coding, robotics, self-driving, and drug discovery, health care, chip
[44:58] design, and other industries, the revenue pool could be much larger than today's software-like products. Again, this is the part of what I said at the beginning. When you have exponential growth, nobody has any idea. The Barclays desk is not including any of this in where the revenue's coming from. They are doing basic human linear garbage work. End of story. You didn't predict the dot-com the the post-iPhone era. And trust me, if you did, you would have made a fortune and not be working anymore cuz you just would have gone out
[45:29] and bought the Mag 7. The Mag 7 dominated the world. We are going to see a similar thing in AI companies. I don't think it'll be OpenAI and Anthropic. I have no problem if people say to me that I don't think they'll ever be worth 10 trillion. Great. So, if they're worth one and a half right now to two, and next year they're still worth one and a half to two, they've won. That's the part I don't understand that people don't get. If they just stay at the same level, they've won. These companies are up so much from where they where they were before. AI
[46:00] could supercharge productivity in industries and flat head count and with flat head count and I think that's what we're going to see. So again, I bring this Barclays desk up cuz I just I think it's so stupid for people not only to write it, but then for a reporter to put it out there. It's only one reason to put it out there. To be bearish. That's it. Like what's the purpose for you guys reading that? It means nothing. They can't predict what this is going to be. Go look at their trading desk and see if they got Micron going from $10 a share to $80 a
[46:31] share. The answer is no. I guarantee you they were off by a factor of five at least. So you can't predict this stuff. No one can. And I'm not saying I can. I'm just saying that the path continues to move at this pace and unless we see a change in the revenue growth, unless we see a change in and I don't mean the revenue growth of a oh my gosh, in August it was slightly lower than July. That means they've lost it. That is complete garbage in terms of the Anthropic and Open AI stuff. Looking at a quarter return for a company and saying that
[47:01] that's the end of it is just ridiculous. Especially when we just launched Grokbot, we just launched Muse, Meta. You're going to have more and more people using other parts of AI where revenues are going to be coming through. Here's the other problem I have and again, we look at this buildout. Canals. We're comparing this stuff 5 years, 20 years. 20 years. 17 years. 7 years. We're near two of this and yet we're saying the bubble is here. We're saying
[47:31] there's been 10 trillion, this is a bubble. We haven't even started the bulk of the CapEx. It's just starting. If there's a bubble, I'm guessing. In fiber it took 7 years of buildout. In this case we're saying in year two while everything still looks positive. It's just insane. And again, we're doing everything out to 2030. I've never seen forecast in my career to reiterate out to 2030. We we've been dealing with what's the
[48:01] economy going to be in a quarter? Is there going to be a recession next year? Now we're talking about 2030? How many things can be different than what's expected in 5 years with exponential innovation? That's why I'm bringing this up all this 2030 stuff. This got brought out again. Bains has AI needs 6 trillion. They're going to be short by 800 billion. I mean, this is insanity. Out to 2031 you're giving me a forecast on how short you're going to be? So again, here's here's CoreWeave. Uh trading this is before Friday. This
[48:31] is after Thursday. Um again, trading in this range they've been in. Here are their clients Microsoft, OpenAI, Meta, Anthropic, Mistral Jane Street. Like are they going to default? I I mean, maybe down the road, not in the next couple years. Uh we cared about this last week. We got this. It's only two trill the two trillion could be 1.9 uh blah blah blah. Again, Jared Cuban came out and basically did the math after everyone was going through it saying, "Oh,
[49:02] they're coming in light." And he just went through it and I love what he said, "You have to do the work and run the math." Let me clear some things up. 100 billion end of year seems pretty straightforward. I I there's no way to look at it and not be able to come up with at least 90. More importantly, we're talking about whether they come in at 90 or 100 when as open as a 16Z put out the labs have added more revenue in 2026 than all of public software combined and we're calling them a failure or slowing
[49:32] down. Um Broadcom, this was a headline by the media likens many AI investors to casino gamblers. It's not what he said. Gavin Baker volatility on misunderstood news flow. Everything I've shown you today is misunderstood news flow plus the the the chest beating of the bears coming out seeing the bubbles pop then weak at the S&P is two two and a half percent off the all-time highs. We had deep seek day. We had liberation day. We had Kimmy. We had blah blah blah
[50:02] everything out and then a mistranslation of Hak Tan's comments yesterday because he basically did not say what they said he did. He did not say people were gambling on AI. He actually said that OpenAI and guessing whether they had a it was 70 billion or 50 billion and jumping to the gun and making decisions off that. That's gambling. That is hedge funds. That is AI agents trading. Make sense to me. Um you get all these jumps and all these moves. You also had CXMT this week and
[50:34] the fact that they were coming out with stuff. So all memory names were going to get hit. So you had all of this stuff happening and again I I feel bad for the people who do all this research and then something comes out and everyone immediately sells the stocks and then they get called up. I go through this as well. And Elon Musk basically came in. He got involved and he said, "Yeah, problem is RAM supply." So you know, whatever. And remember, here are the facts. Four weeks ago, Micron estimates signed agreements represent more than 35% of
[51:04] its revenue through 2030. So remember the 2030 forecast for what CapEx will be and how this will be the biggest bubble in history in five years. How we have a situation where Bain says we're not going to get the revenues in five years. How Gartner says the neoclass could blow up in five years. Here are the contracts. So the these are not forecast by those guys that have no no facts to them. That's just their guess, okay? And that's a fact. That is just their guess.
[51:35] Micron has signed agreements of its revenue through 2030. ASML AI chip demand is stretching into 2030. It already expects supply shortages through 2028. Nvidia-backed Lumentum sold out of AI parts till 2029. You have to literally go through and just realize what I'm showing you. Pe- papers writing how in 5 years there could be problems at the same time that real business is being constructed that
[52:05] is sold out until then. Amazon CEO Andy Jassy told him AI computer sold out beyond 2028. Nvidia just guided for 70% revenue growth. This is from a month ago or a month and a half ago. 70% revenue growth. It was expected to be I think 40. It It came in at 70 for next year. Jensen Huang guided to 70%, which is
[52:37] going to be higher than that just based on his track record. Uh >> [sighs] >> Meta Muse just came out. So again, as people are saying there's not They're forecasting where revenues are going to be and we're getting new progress. As progress comes out, as we break cancer research, as we do robotics, and as I'm going to show you everything is accelerating, revenues will accelerate, too. As the progress gets better, it gets easier for people to pay them to use it. Plain and simple. I don't know why this is hard to
[53:07] understand. This is the video I did uh on Thursday putting the If you haven't seen it, go watch it. Um again, it's a lot of it is is in this or some of it is. This is where we closed on Friday. This is the most bizarre thing to me. I'm listening to bubble talk and how this is ending and this is like I said, do you know how many AI bubble prickings we've had in this period? Like it's non-stop. Here's where semis are after all of the panic of July and all of the panic of this month.
[53:38] The SMH is up 50% year-to-date, slightly more than last year. I think the most important thing is the we've If you've ignored semis, you're ignoring basically the mag seven now. Let's assume that it started in 2019 when machine learning and the cloud and everything really started to to to become more of a a thing. The transformer papers, everything. All you had to do is invest in semiconductors. You've had seven of the last eight years effectively be 40% or greater.
[54:08] And yet people are saying the bubble has popped. You're just missing the opportunity. Regardless of what the charts look like of how fast things went up this year, Oh God. And here's the same thing in the S&P. And I just wanted to show this for people who have joined here to avoid the news, send this to your family members, send it to everyone. Compounding is why I got involved in the markets. My father was a construction worker, didn't get didn't finish high school, left school in the ninth grade, taught me handicapping, taught me math, taught me everything to go through. But
[54:38] what he really taught me was, I don't want to be a construction worker based on your hearing, based on your back, based on the electrocution times, based on me needing to siphon gas. I didn't want to do that. My mother handed me a book and it was on compounding. This is what the S&P 500 has done. If you were in the last 10 years, you got 17% annualized. If you were in the last 20, you got 13. The last 30, 12. The last 40, 13. This is not cherry-picking numbers. This is just using the last decade, 20 years, 30 years, 40 years, 50
[55:08] double digits in every case. Back to 90 years. This is what's happened, guys. It's been getting better recently. But this is what happened. Let's just take the last 50. So if you put $6,000 in 50 years ago, I'm sorry, if you put $10 in every month, just $10 for 50 years and ignored everything in the papers, the value of that $10 each month for 50 years, $10 is now $478,000.
[55:39] And this has got in a 2 and 1/2% dividend to just use it in there. Guys, compounding is your friend. And the S&P 500 and all investments, effectively equities represent innovation. So, if you believe AI is going to continue to have an impact on the world for now, you want public stocks, but you also want to involve in tokenized things as soon as you can. I don't think VC and any of that stuff is going to be able to monetize because they're cherry-picking out of the billion of winners, and I don't think you can do that anymore. Here's the log
[56:10] chart of the S&P 500. Are there periods of time where it goes sideways? That's why I just showed you what the returns are. But the reality is when you're at all-time highs, you're at all-time highs for the last 90 years. Could there be a correction that takes us back 20%? Yes, we've seen a bunch of those since 2019. It shouldn't be part of it. This is the thing that got me going. When I read this in a book my mother gave me, a penny doubled every day from day one to the end of the month, by the time you get to day 31, there's $10.7 million.
[56:41] That's why I got interested in not being electrocuted. This is the chart of where we are. I show this time and time again. And the one thing I know is that this is incredibly disruptive to hedge funds and leverage because it makes certainty gone. Hedge funds invest in certainty, end of story. They trade their book. They get out of it. I was in one, had my own for 20 years. It's about short-term arbitrage. It's
[57:13] about arbitraging something in the future, but not carrying long positions for long periods of time. We are entering a period right now of complete uncertainty, so leverage has to come down. For all the negative things you've heard on OpenAI and Anthropic, here's where they were 2 years ago valuation-wise versus today. And I'm this easy. Anthropic is more like a bit a peer. We're talking about companies that have gone from less than a hundred billion dollars in Anthropic's case, less than five billion dollars
[57:44] to two trillion, and we're saying they're not successful, and you're trying to pick the top of this. It's just moronic. It really is. SpaceX, how many people continue to say that SpaceX blah blah blah blah blah. Here's the valuation. These are just the facts. If you invested in them, you're doing great. Here's what the S&P did over the last year, 16%. Boring. Public equities are in trouble on a comparable basis to things that are moving on the hyper the the exponential. The exponential are AI trades. I gave
[58:15] you Anthropic and OpenAI, but this could have been Cursor. This could have been so many different companies, OpenRouter, that have been purchased. SemiAnalysis goes from a zero-based company or 200 million to 2 billion, whatever you want to say. If you do 20 times their revenue. Somewhere within here at this point because of the growth, you have to understand that public companies just can't grow this fast with AI. And you don't know where the revenues are going to be coming from because you're only focused on a few companies. That's not the case. Very important interview, listen to Zuckerberg and Alexander Wang. First of
[58:46] all, we had heard a lot about the relationship. Um this was a great interview on on Cheeky Pawn. I highly recommend um listening to the entire thing. It's about two hours. Um he talks about the compute and the spin not working out. He talks about the neo clouds and why they don't have as it is easy as they do. He talks about the fact that GPUs are basically higher than they were a year ago. The value of even an H100, the old ones, at this point is
[59:17] higher than it was last year. This is him saying this when talking about all of this. So, again, if GPUs remain productive longer than expected, then they can generate more revenue from the same hardware with an AI. This is just someone who's giving you the positive readout. If GPU prices are going higher, it's very hard to be bearish on CoreWeave and not bearish. You can be bearish on the equity. You can be bearish on the bonds. To say it's an AI bubble bursting is just very hard when the GPU prices are going up. OpenAI's GPT-6 may lead to
[59:49] robotics breakthrough as soon as next year. So, Unitree uh which as I was told by Emad Mostaque, who I I was at dinner at this week and shocked me. I didn't realize Unitree had only sold uh about 11,000 robots this year. Humanoids, robots. Um So, when the I This is really a big deal because it's a Chinese company, very important to the government, saying OpenAI's latest advances have
[60:21] accelerated prospects for a GPT moment in embodied intelligence. Think about what that means. It's two things. For all of you who say the Chinese are at the same level, which they're not, this is about Astra. Okay? We are two models ahead of Astra right now, which they haven't released, and my guess is, especially since we're not seeing as many Chinese new models come out, distillation may be over and the benefits of OpenAI and Anthropic, which was talked about by Dylan Patel, if you
[60:51] listen to the podcast, may not be released. This was also talked about on the Zuckerberg one. We might be reaching a point where we never see, or at least for a long time, the models better than Astra and Opus 5.5. That may be it for us. Because there may be a reason now to not have them come out to avoid distillation, and that would create a really big gap between the Chinese and the US, and this whole open-source thing would be in trouble because one of the major reasons for open-source is catch staying close
[61:22] is you release the model and then there's distillation. Well, if we're not doing that and if we're leading to robotics, why wouldn't they keep the models for robotics to themselves at this point? So, Frontier Model Intelligence, as was said by the Semianalysis guys, reaches a point where their revenues are not going to be about consumers paying for it or even honestly some enterprises paying for it. The majority of the big enterprises are going to continue to pay for Frontier models.
[61:52] But, the reality is most of their business going forward may be from biotech. It may be from from nano technology and materials. It may be from robotics. It may be from That's probably where it's going to be and that revenue hasn't even started yet. Astro makes a robotics breakthrough closer because it potentially solves the hardest remaining problem in humanoid robots, moving from a robot that can perform a program task to one that can understand a goal, reason through unfamiliar situations, and adapt actions to the physical world. This is how
[62:22] important it is for the humanoid explosion. So, again, personal agents, agent swarms, solving math problems, and robotics are all together. Do not get bearish AI compute with all of this hitting right now, guys, as I have my hoodie on. Brad Gerstner, Jensen Huang, Clark Tang, great thing to listen to if you want to feel better about AI. From the guy who's been right the entire time. Uh Frontier Labs are roughly a third of
[62:52] the business, not the whole story. He's basically saying, "Why is so much pressure Why is everyone focused so much on Open AI and Anthropic?" And he goes through the math of where the 10 trillion What it includes, like where the revenues are going to come from. And one of the things he says right off the bat is you have to understand it's 500 billion a year just for new compute the way it's always been. So, again, to listen to him go through this and to break down from someone who's been right
[63:23] consistently, and he talks and makes jokes about it, whose business is booming and he's involved with every company in there, to listen to him answer all the questions. He doesn't care about where CoreWeave's debt is. That's why he came out and said, "They're fine. Their cost of capital is actually lower." Which again, then people laugh at. They don't even go do the work, which arguably is true. So, again, as you go through this and you don't figure it out and you don't think about it, you're just responding to things. And I get it when prices are
[63:53] going down, your mind runs to places that don't make sense. I'm just telling you hedge funds are doing this. This is a hedge fund deleveraging event. Now, the Korean market was involved in the first part of the year. Retail got in. It led to hedge funds going down. This is another leg of hedge funds, but remember, hedge funds that are near closure don't have a stack anymore. Hedge funds that have been beaten down, like the ones I showed, that are flat for the year and now could be down 10 or 15% if they didn't get out of their positions, they're going to take things
[64:23] off because they don't want to be down 40 and they've already seen a down 50. So, you start running into issues and primes start getting worried and people start coming in and investors are going through it. It's not a fun story, especially when it's happening across the board because that makes the primes even more worried. So, again, the frontier labs are a slice, they're not the engine. Bucket one, classical compute. I'm just giving you the three that he mentions here. You should really go through it on your own. I'm going to try to start to speed this up a little bit cuz this is a long
[64:53] video, but there's a lot of important things in there and hopefully you guys are getting it from it. Uh this is talking about his track record where he's basically been right throughout the entire time of doing the interviews in terms of how much compute would be spent, how things would do, what would happen to GPU prices. All of that's been there. And then Brad on the inevitability of AI. this is the most important exchange in the conversation. Brad is pushing back on Jensen's own answer to Ezra Klein, where he gave something like there's two or three years of visibility before the bubble question becomes live. Brad reframes it.
[65:25] This is not a CapEx cycle with a visible end. It's a fundamental re-architecture of global compute permanently. Huang's one-word answer was forever. So, I completely agree with this and this is the reason why the dot-com bubble was a mismatch between the demand not being there while the supply was being built on leverage by the companies that had nothing to do with the demand. This was about the fiber companies
[65:55] putting up the rails. This is very similar to crypto. Crypto didn't have the users, they didn't have the demand. Now, they're going to have the demand from the AI agents, which is why I say it's the same thing to me as the end of the dot-com bubble for crypto and why my positions are in crypto. His point, which I completely agree with, there is no end to this build-out. There is no end. And because the progress is driving the demand and the progress keeps accelerating, as I've shown you, I cannot make it any clearer. Token use has gone higher as progress
[66:27] moves forward. As the models get better, they're able to do more things that generate revenues. That is not a guess, so you guys know. That has been a fact. That's why listening to people saying they know what the revenues are going to look like in 2030 when they didn't even get Micron's revenue is literally ridiculous to listen to these people, including Gartner Group. It is a joke. Here have been the revenues for Nvidia relative to the consensus. And what Brad Gerstner was showing on
[66:57] here is where it was just a year ago. This was the consensus, again. This is what the consensus was for Look how far off the street was. And yet the stock price is unchanged as I said. Think about it. If you're a hedge fund and you thought the street thought that for 2029 it would be 305 and now it's at 982 a trillion dollars and you watch this happen and made no money.
[67:27] Of course you have to get out of things. Of course you do. This is the only thing right now and this is more of a slowing thing. I think what the market is going through is that the capital coming to the market is just causing the market to have trouble dealing with this stuff. As more money needs to be raised and people go I'd rather I got to make sure I have money for open AI. I have to make sure I have money for Anthropic. And whether it's mutual funds or this they have to raise cash.
[67:58] And I think all of this stuff that's been going on for SpaceX, for the bonds, for everything, the way that rates have moved, it's all been related. There's plenty of capital in the world for the AI build out. If it ends up being 20 trillion and not 10 trillion, there's plenty of capital but it will have to come from somewhere. If the revenues are coming in, people would rather own bonds at some point in CoreWeave than they would the Treasury. Absolutely. Someone this week said would you rather buy the equity or the bonds and I said I'd rather buy the bonds.
[68:29] That's where my mentality is at this point. I just think at this point with the panic that's going on and what people are doing if I had to have a choice, I'd rather have the debt at this point at these levels. When Genius Failed. So again, we're going out to 2030, we've lost all certainty. I watched as LTCM had trades on like Brazilian Telebras preferreds versus ordinaries that traded at a spread. The preferreds traded at a higher price but every day the privatization was going to happen, that was all going to converge to zero, the spread. So if
[69:00] you're trading at a 5 to 10% spread and 6 months from now you're going to have a privatization event, that should collapse to zero. So, if you put that trade on along with a bunch of other spread trades on, which is what LTCM did, you think you're going to make money. But, here's the problem. This is the lessons from leverage. And I want to make sure that for those of you this is even for people managing money at a freaking hedge fund who've never thought about this stuff. Lessons from leverage. Your fundamental thesis can be correct.
[69:32] But, the timing. If the AI trade, if all of Nvidia, if all of a sudden next year Nvidia is up 300% and goes back to where it was before, nothing changed. It's just the timing you got wrong. You were in it, you couldn't wait because you've got investors who are measuring you on a 12-month basis. The volatility. We've had a dramatic shift in the Vixy Q. Single name vol has exploded. Index vol has barely budged. You have no hedges. So, you can be wrong on volatility of the markets and the relationship of
[70:02] volatility to the index. The financing. You can have rates move higher. That can cause a problem on things. You've got to borrow more money. Maybe the banks can't lend as much or maybe your cost of capital is going up and your investors are like, "Hey, I can now invest in Libor plus 400. I don't need you guys anymore." Crowding. There can be too many people in the trade, which I think is one of the major problems, while liquidity is dropping off, which again, we've seen. Correlation. You can have breaks in correlation for periods of time. Well, we're definitely seeing that
[70:32] right now where the Mex Bolsa is going higher at the same time that the Qs are going down. So, if you've been sitting there riding a winner of Okay, I'm going to be long Qs. I need some shorts on it. Oh, the Mex Bolsa is trading at 52-week lows. That's a good short. Blah blah blah. Liquidity. Obviously, this was a major one with LTCM. Once it went through, they couldn't get out of it. And then the position sizing and I think situational awareness. Gone through it. That's why I did this stack on moral philosophy. I've done presentations mainly for mutual funds at this point. This is the way I see this going. We
[71:02] went from a normal distribution curve where we knew that 5 years from now Ford would still be in business. Again, unless there's some serious recession, but in this case with growth going, we are getting a weird world where we've got losers, think retail mall names, winner, think Amazon, and then we have a bunch of names right now like bank stocks that are sitting in the middle here that I think will gradually go into here leaving very few AI winners, and I think over time more of these will come over here.
[71:32] The losers will be have no growth. They will have their growth eradicated from AI. And that's just the fact. Here are the hyperscalers. Remember, arguably the most important space aside from Nvidia and their GPUs is what the hyperscalers stock price is doing. If capital's the issue, their stocks made new all-time highs as a group. And again, this is Amazon. This is Meta. This is Google. This is Microsoft. This is those four names.
[72:02] I did want to bring this up. Uh one one data point um which I think is important and becoming more important. Consumer confidence made the lowest level since 2013. Yes, the lowest level. We are below when there was a pandemic and millions of people were dying. We're lower than that. That's how bad things are in the way people view it as we go into the midterms. Since 2014, the unemployment rate here was 200 basis points higher than it is today. To show the importance of this from the
[72:32] jobless situation, jobs is what drives this historically. This is the unemployment rate, and this is now that consumer confidence number inverted, and this is back to 1970. It is the jobs market that drives this historically. Now, we've seen the jobs market slightly weaken, but this year has been fine. But like I've said before, the jobs market is far weaker than economists will tell you. This is inside the number of that consumer confidence. This is the jobs plentiful versus hard to get inverted
[73:04] again. So, hard hard to get jobs versus plentiful, you've got, again, the It says the unemployment rate should be closer to six, just like it was back in 2013. Uh Warren Pies put this out uh last week. And again, financials and tech payrolls are shrinking. Aside from nursing jobs, health care jobs, we are negative now for whatever it's been, 18 months. And he's just highlighting that this is
[73:34] a dramatic change in terms of what has gone on in the past. And we're entering a new period where the agent workflows are about to be adopted. So, I do think you're going to have a labor kind of situation. You can view how that matters. It's not a recession type thing. It's a profit margin type thing, but it's also a politics thing. It could be a Fed thing. Uh we'll just have to see. And remember, another way that we've seen this is we're right now looking at uh earnings growth in the S&P at 29%. So, up here.
[74:06] And this is the year-over-year jobs creation relative to it. So, to finish up, very short week in crypto. I did this video. I continue to go through learn the stack. For those people who have joined, remember to watch the videos. I'll have another one come out this week on It's been a busy week, obviously, with the video I did on the AI side. I will continue to go through it. Um last week, we went through the Oracle data and interop middleware and the settlement and smart contract rails. For those of you interested, especially in tokenization, I think you should
[74:38] sign up if you haven't already uh particularly on the hedge fund and mutual fund side and really start to get these. Uh I think you should be doing your homework now. As you're getting as you're watching the horrible trade in the AI thing, crypto is going to be a very different thing next year. Um privacy is the theme that I covered last week, very important. Here is the month-to-date. So, the index is down 7% Bitcoin's down 2 and 1/2 since August again, still crushing it since June.
[75:08] Same thing as AI has worsened, they have done well. Uh still bull market here. Did see some weakness in the names on the 50-day. Only 20 names above the 50-day, but in terms of the 200-day and what's rising in the slopes, good. And again, a correction within side a very large market. Bitcoin is hanging in here. I don't think we're going to see a move higher in Bitcoin until we see some stabilization in the hedge fund scenario. Uh but hopefully it'll be able in here. It would not surprise me if it
[75:39] kept meandering back down towards the 75 area. Uh just in there. And the final piece of information, which I will cover in a paper this week, related to crypto. I think it is shocking this number that their net interest income could reach $107 billion according to them next year. This money, as I showed last week with Stripe, is about to come out of the market. All right. Long one, an important one. Um I need to keep you guys happy. I gave you a hoodies up
[76:11] uh week. I'll see you next week. And if big things are happening during the week, uh I will keep you up with one, two, three, who knows how many videos I might have to do this week because of all the news going on. Stay calm. Don't worry about the bubble, but for hedge funds, this is the end of leverage in hedge funds uh as a rising situation. See you.
[0:31] does this remind you of '08? Does this remind you of '01? Is this the same type of situation? And the answer for everyone is uh my career began in 1994 with Mexico. I trade started trading a book 2 months before the devaluation. I was at Morgan Stanley. I was a young person involved in a market going through things I've never seen before. Then I was part of '97 and '99 in emerging markets trading everything through the EM crisis, including the closure of LTCM was on both sides of their business, meaning on
[1:01] the other side of their trades as they were doing things in Brazil, and then part of the shutdown, and then a bunch of them worked at Morgan Stanley, and it was like working with the greatest people I've ever been with because I got to ask them, "How could you how could what happened happened occur?" And you notice LTCM leverage lessons. I'll be going through those. I came back to the US. I took over the S&P. Then I was there for the dot-com bubble, managing ETF business, index businesses. I was a part of all of that on the other side of billion-dollar
[1:31] transactions every day. Then I traded a billion-dollar macro book throughout the great financial crisis, took money out of the banks weekly and put it into gold. I have traded all kinds of things and been involved as a CIO during COVID. I bring this up because this is not a bubble popping. It might be in the future. It's not now. So, with hoodies up, I'm going to give you guys a lot of different things to go through. Again, I'm doing this for your information,
[2:01] your education, and hopefully not this week. We didn't make money this week, but over the long term, the goal of this is to make sure that in weeks like this, you're paying attention to the stuff that's actually happening and not to the noise that's going on. There is a bubble being popped. It is not in the equity market, and it is not in the economy. This is a true number. This is the rolling drawdowns of the S&P 500 over time. Here is the Great Financial Crisis. Look at that, almost 60% peak to
[2:33] trough. Here's the dot-com bubble. Eventually got to 50. Here is COVID, got close to 40. Here is when the Fed's raising rates. Here is the tariff tantrum. Here we are now. We are 2.7% off the all-time highs in the S&P 500 as of the close of Friday. Guys, please. Here's the bubble that has popped. I will go through this. Hedge fund leverage has popped. It will never ever ever ever ever, in my opinion, exceed.
[3:03] And what we are going through is a constant deleveraging that needs to happen. CoreWeave, not a problem. Hedge funds, big problem. I'll go through the reasons why. This is multi-strat leverage, which peaked, and this is from the Treasury. Peaked, started to decline. You're going to see more of this. I'm going to go through why. Hedge funds leverage is in a complete and utter dismantling.
[3:33] There will never be a time where you will see it go up, in my opinion, based on the things that I'm going to show you, because at the end of the day, you have to provide a sharp ratio that makes sense. And we are seeing extraordinary moves. And remember, as I go through the leverage, the leverage in the hedge funds dramatically bigger than anything going on in AI right now, not even close. Here's the Coreweave chart. This is where we closed the week. I just want to show this because this is the fifth time the equity price has been here. Every
[4:04] single bubble I have ever traded when it has popped, the equity is down massively in the bubble popping situation way way way way further than this. By the time we get to the bubble popping, when you see the bubble popping. And again, maybe there will be an AI bubble in a year, in 2 years. We definitely are constrained like I said last week, capital is the biggest risk. We continue to tap the markets for more capital. This is not about non-payments, this is not about not earnings, this is not about GPU
[4:34] prices falling. So, we're ahead of the game on the fundamentals, which means like there's been in Coreweave where we've had these panics, we get a bounce and we get a bounce. Now, the bounces seem to be getting shorter and I don't have a problem like I said on the weekly video for subscribers with the equity price of of Coreweave building in a higher risk premium. But, look at what started this whole thing. And again, hedge funds were losing money from July in a historic fashion. They got a bounce somewhat in September,
[5:04] but as I'm going to show you, some news came out after I did the video last week that shocked me. I can't believe it wasn't more of a story. Gartner says over 3 and 1/2 trillion of AI infrastructure spending could be exposed to impairment by the end of 2030. First of all, we don't have this number yet. So, I I just want to make sure you realize what this is saying. I've never in my history of trading markets seen a forecast out to 2030 that anyone cared about. We're talking about 4 and change years from now could be exposed. And yet, people sent
[5:35] this to me going up, this is a problem for Coreweave. Coreweave might be in trouble then, but we have no idea what's going to happen. If interest rates go down, it's not a problem for them. If the cost of memory and everything goes down to building a data center, but the compute demand stays high, you can have massive endless compute demand. And as Jim Chanos has showed, it may not be a good equity business. It doesn't mean it's going to zero. It doesn't mean we're going to have an unwind. Here's a chart of the S&P, again, sitting near the all-time highs.
[6:05] Here this one's also astonishing. So, here are Friday's returns around the globe. So, the S&P's down 1%. Big deal. The VIX has been trading 16. So, we're barely getting one standard deviation falls. It had been like 100 days that we had gone without getting a 1% move. We had two of them last week. The Nasdaq's down two. On the same day that that's happening, look at these. I mean, again, Brazil's up, Mexico's up, Canada's up, Europe's up. Every single
[6:36] one of them is up. If you go look at the last 5 days, S&P down, so this is for the week, down 2%, down 4 and 1/2. Again, Europe's up across the board. Look at Mexico. Uh Nikkei's down. The AI trade is what's down. Again, the AI trade is down. So, year-to-date, here's where we they are. The S&P's up 12, Nasdaq's up 13. Again, bubbles popping. If people are right on this, they are the best forecasters I've
[7:06] ever seen. So, in January of 2007, just to give you one example, you can do this throughout the dot-com bubble. It did not become obvious what the problems were in the dot-com bubble until well after the equity market had started to fall, well after. In the case of here, we were getting defaults already. Like, the housing bubble was unwinding in 2006, 2007, way before. Probably 2007 is the better one. I think March of 2007 is when I associate as when at the fund we
[7:38] were at, we were short, we had the big short trade on as well. We started seeing these things which we knew were horrible mortgages that were being written. As I said in my subscriber thing this week, when I go through the core weave clients, remember the clients of the subprime mortgages? The story and thing strippers in Florida were getting mortgages. So, again, this is not the same scam that was done. There's demand for all of this. Here's AIG. You can go back and look at AIG stock price. I
[8:09] advise you to go look at Sienna stock price, which is another AI chart right now. But, when this thing finally fell, again, before we got to Lehman Brothers, which was in here, all of financials were down massively. The credit spreads were widening. Here's where we finish when people said, "How do I watch credit? How bad is credit?" Here is the high yield side of option adjusted spreads within side the cash bonds. So, again, are we doing anything other than what we did across
[8:40] so many different times here? Of course not. Is this COVID? No. Is this the fracking situation in by the [clears throat] time we got again to a serious problem in oil? No. And again, it's not to say that we can't get up to those levels. We can. We're not there yet. And now I'm going to take you through what this is from my perspective. I was heavily involved with Long-Term Capital Management. For those of you who haven't read this book, it's probably a good time to pull it out and go understand it.
[9:11] Hedge funds at the end of the day are leverage vehicles. They are making bets to some degree on volatility and correlation amongst a diversification of assets. For the most part, a multi-strat in particular, it's across different types of strategies. Hundreds of them in some of the largest places with thousands of investment professionals hoping that there'd be a low correlation. The problem is, since I've been involved with markets, hedge funds have focused their attention
[9:41] for the most part, the way that you really get diversified is by having high trading activity, which means you depend on stocks to move based on information. So, if you get really good information and you think companies are going to beat the quarter cuz you've worked really hard, you've done all the channel checks, and you've done things that a retail investor can't do historically, and they beat the the quarter, you should make money. Well, again, I show this all the time.
[10:11] Here's Nvidia. The stock price is effectively unchanged since October of last year. 1 year, it's effectively unchanged. Their earnings have blown away consensus in a way that nobody could have expected, and what we've had is a complete crash in their PE, which is now down off next year's earnings to 15, and depending on where you go, if you take the higher area, it's probably closer to 10. That's what's happened is as they beat numbers, the stock's not reacting that
[10:42] way. And it was fine when it was just Nvidia for most of this year, but it has spread to the entire AI trade, and this is why when I said back the fireworks show is over, this is directly what it was referenced to. You will never go through what you went through in the first quarter again, ever again in public stocks, ever. As far as I'm concerned, we have destroyed terminal value. That terminal value means that any company's earnings past 3 years, we have no idea whether they're going to survive or not.
[11:12] The hedge fund industry is at all-time record leverage. I'm going to move this over so you can see it. This is from the fret the the the Fed, and this is through Q1 of this year. Okay? All time, but more importantly, look at the leverage effectively from 2018 to 2023, it was unchanged. We've now gone up dramatically and more than doubled. The leverage provided by the prime brokers of the banks, another place where crypto is going to have an
[11:43] impact, is at all-time highs. Here it is from the OFR report. Breaks it out by category, multis. These are big numbers. Here's long-short equity. Close to two to one. You're borrowing money. So, hedge funds make money on leverage. And there are multiple ways that you can use when you use leverage. When you're not taking directional bets in the market. And again, the net exposure across the board for multis is very low. For long-short equity, it's been on the
[12:14] decline for most of this year, but the gross has remained high. Now, this is all through Q2. We know that July must have had an impact and these numbers must have come down because the vol has gone up so significantly as I showed on the factor side. It just means the instability within side all parts of the market and most importantly, there are very limited hedges for the AI trade because it is a cross sectors. There is no S&P put, as I will show you. Here's the leverage growth as I'm
[12:45] showing, 15% in hedge funds and then the BDs, which again, the broker-dealers, heavily involved with the prime brokerage, massive growth. When you're looking for leverage, like I said, we're talking about 40 to 60 billion in core wave debt over the course of 6 years. Hedge funds have money, as we learned with situational awareness, that can become an issue in a matter of weeks. On that point,
[13:16] we all forget and everyone started bringing out X jokes, like this was the issue. This was just one of the issues in July. So, yes. To get a number here, again, we're caring about CoreWeave whose debt is spread out over the next seven years, six years. And at the same time, the clients they have are the biggest clients in the world with some of the best balance sheets in the world. Companies like Microsoft and Meta
[13:47] dominating their stuff. Are they not going to get paid by them? Is that really going to be an issue? Are they stripper-related? Here is what happens. These are the numbers that were reported by the Wall Street Journal that everyone kind of backed into. I don't really care whether or not this is for illustrations purposes. I do know one thing. They're not far off. So, if there's $11 billion of equity and $45 billion of positions, the borrowed capital
[14:17] is 33 billion. And as long as you're making money, there is not an issue. So, hedge funds are the problem, guys. Mark my words. They are the problems. In my world of terminal value being eradicated in the bank apocalypse and the pushback that I get, the pushback I got with Micron last year, this is a scenario that hedge funds are not being honest with themselves about how much the market structure has changed. And it's not just hedge funds that are
[14:47] feeling the pain. Mutual funds are feeling it, too. That's why I did the demo that I did last week to show the types of market structure changes that I'm trying to help people with on how bad this is going to be. You don't need as much leverage to make the same amount of money as we're going to see. So, the leverage is going to come down. That doesn't mean the hedge fund industry won't be producing the same types of returns. I think it's possible they can generate similar returns. The money might aggregate even more into the the bigger places that have an advantage on the
[15:17] risk management side, but I definitely think the market structure has changed making it more challenging to have a good sharp ratio. So, Vara, another multi-billion dollar hedge fund, $20 billion dollar flagship AI fund, the losses paired the year-to-date returns from to 65% through July. So, again, they fell 44% in a month. This was in August. So, this is not now, but these types of drops, I don't think people It wasn't just situational awareness.
[15:49] This one happened and came out after the bell last Friday. So, I didn't include this in my video. So, if you asked me to name the 10 best non-multi-strat hedge funds of this century, Lone Pine would absolutely be in there based on my experience with them, based on the way people talk about Steve Mandel. Just an incredible hedge fund. So, this is not even saying anything negative about them. They were up 44% and it erased the gains.
[16:19] It is the staggering drawdown that happened in a quarter. So, to be down, to give up 44% gains, assuming they're charging 20% on their AUM, that means that they lost over 55% in gross returns. That's That's a drawdown that again, I showed you in the S&P 500, that's equivalent to the great financial crisis drawdown by a hedge fund.
[16:49] One of the greatest hedge funds of all time. Here is their returns that I have basically on public data through Q2 relative to the S&P. So, basically, they move a lot with the S&P. Based on what you're seeing over the course of the last 3 years, there is a high correlation to the S&P returns. This is what happened during the quarter that they lost 55%. This kind of stuff has never happened
[17:19] before in my career. I've never seen anything like this. This is LTCM-ish. That's what it is. Now, uh again, that was through September. Yes, this month has been a disaster. The reason I'm bringing this up is not because Lone Pine is in any way, shape, or form in trouble. That has nothing to do with this. This has to do with when you're short-stacked and you have that kind of a drawdown, you're already figuring out what's going to happen, what your people are going through, what the investors
[17:49] are going to say. There's a lot of work that goes on. And so, to put on risk at that point, it is is just very unlikely. And more importantly, to deal with any more losses come up, and that's what I would say is going on right now with inside the marketplace. You're seeing things. Jared Kubin is is someone who's great to watch on the quant side and just to highlight. But, he bloody last 21 days, this is through the middle of last week, and it's just showing how many names are up violently in terms of sectors. Here it is here.
[18:20] Again, what's down? It's not just the semiconductors. It is fertilizer and crop inputs. How often did we hear how easy it would be that we'd have food shortages and that uh these are all hedge fund trades that people had gotten into. You obviously have short SAS up here. You have short software up here. You just have a lot of different components that are not just related to that one thing. And the reason I know that's the case is this kind of stuff
[18:50] also doesn't happen. This is the Mexican stock market. Trading at 52, well, close to 52-week lows at the end of the quarter. It has gone straight up. Here is the monthly or the weekly return of the Mex Bolsa relative to the Qs going back over the last 3 years. Here is the chart of the Goldman Sachs crowded hedge fund longs relative to the S&P. So, again, this month here
[19:21] Vara down 44%. Situational awareness down I don't remember how much they were down, 60, 70, whatever to get to the to get bailed out by Ken. We barely got a bounce here. And this is the most important thing is the hedge funds, these are crowded long, so this includes the fertilizers, includes everything. It just never got a movement on it, very similar to momentum, which I'm not going to bring up. Here's what's happened so far this month, and it's been a straight line down. So, again, these are the equity positions that hedge funds are holding
[19:52] that are dominated by the AI trade because that's the only thing that works. That is about 25% of the market, which I've talked about. Everyone's in the same names. So, it's still up big for the year, and as I'm going to show you, semis are still up about 50% year-to-date despite all of this going on. Nvidia's PE has collapsed. This is deleveraging, guys, and this started in here, and then as I show you
[20:22] all these hedge funds that had losses, and then you get these bounces as we head into hyperscaler earnings and the midterm elections. So, if you ask me what the pain trade is, the pain trade for this is to continue into the hyperscaler earnings and maybe all the way into the midterms. Take everyone out, they will not take risk on, and then in the end of the year, assuming I'm right and there's no AI bubble and this is just a positioning shakeout, you get a snapback. Here is the hedge fund
[20:52] that that chart now with monthly numbers. So, in July, it was down 10%. So, the crowded hedge fund names relative to the S&P. Hedge fund leverage was down already when this occurred in 2008. That's the thing I want to make sure you know. This does not include leverage. This is an unlevered number. This is just market neutral the stocks that are the most crowded according to Goldman Sachs relative to the S&P 500. So, these are
[21:22] market neutral, dollar neutral scenarios. There's beta in this to some degree, but as you can see like over time there's the movements this is an outsized move and again this is the Lehman post-Lehman Brothers move. So, hedge funds were already delevered at this scenario. Right here not delevered as much as they should have been based on all the metrics that I got. Deleveraging has come down. Here are the rolling four months. Again you're talking 15% over the last
[21:54] four months of that chart. Again, unlevered. So, if you add in the leverage that we've said, that would mean I mean a significant amount of losses. Here's here, but that really is all equities. So, I just want to make sure you know everything I'm showing you right now is equities. The real problem came in when it started to spread this month. That didn't happen in July. Here is the AI, the Goldman Sachs AI winners versus losers, which is the orange the white line and then the other line is that Goldman Sachs hedge fund. So, you
[22:24] can see this trade is basically at the end of the day it's the AI trade that every hedge fund has on. Everybody that it's it's the pair trade that everyone has on the winners versus losers and because they have a month-to-month bogey, because they have a drawdown level, because the majority of hedge funds don't have five-year locked up money, they've got short-term locked up money, they can't have drawdowns without their business becoming a risk. And I think this is the puking of a lot of businesses that had
[22:54] significant drawdowns that can't handle it anymore. Here is my thematic portfolio for the week. Last week was down 11%. This is worse than any week during July. The month isn't as bad as July, but the week is. That felt a lot like capitulation, but here's the place that it started to really become an issue. So, this is where you really start getting into the pods. The pods, the big multi-strats, they're going to be losing on the equity long short like everyone,
[23:24] but when credit starts to become an issue, that's when it really gets bad because credit starts to impact a lot of other strategies. I heard biotech was a major, major problem in September. Again, the weakening kind of spreading the contagion within side pods is always a big story, but when credit has this kind of week, this is HYG relative to IEF, which was sitting near all-time highs at the end of the quarter, which meant if you were running a credit strategy, everything was great. And this
[23:54] is a synthetic credit strategy. If you're just long HYG, and then you're short IEF, which is a treasury, that's theoretically taking out your rate exposure, and the only thing left is the credit credit side. And so, if spreads stay unchanged, you're just collecting the differential. So, it basically, since February, went up in kind of a straight line. Well, this gave back effectively all of the returns from the summer. So, in the same way that we had the July numbers, look what happened
[24:24] in July? Nothing. Nothing happened to credit. Here it is on a a weekly basis. Not anything like what happened with the tariff situation, but still, a big week. Here is the convert index. So, converts have joined the party. Why? Well, converts ex- extend from CoreWeave, uh MicroStrategy. It's a very concentrated world as well in in in in converts, but again, if the only
[24:54] place raising capital is AI, which is what's been happening. The size of the capital raises for AI, and not just SpaceX, not just Anthropic, not just OpenAI, but Broadcom coming to the market with over a hundred billion dollars this year. Google coming to the market between equities and bonds for over a hundred billion. We're talking about massive massive capital raises, which again the market has, but if you come to the market, the cash that people have might be from maturing bonds. They have to sell something to go
[25:24] buy things. People don't just have cash in most cases just sitting around endlessly. They got to sell something. Part of that was treasuries. Part of that ends up being some equities, whatever the case is, but converts have now started to get hit. And again, they were getting hit in July. They do have an equity one, but this is a very different fall. This is more of a waterfall. This was more of a reluctant fall that immediately bounced back and and made back most of the fall. Um the VIX did have a big move higher. Again, we're still at at at fairly reasonable
[25:56] levels, but the VIX was up 25 points in a week. VIX is the vol of the VIX. So people are buying call options. The VIX itself closed just below 20, which again is shocking that we could still be below 20, but that's like I showed you, the S&P is not moving that much. Uh the Qs were down 5 and 1/2% for the week. The S&P, like I said, down 1.95, and over the last year that's one of one, two, three, four, five, six seven moves of a similar move. Um during
[26:29] the week, 350 of the S&P 500 stocks were up on the week. So just like in the first chart, this was a unwind where we had been watching as equal weight and breadth had been getting hit while cap weight had been doing well. Well, that unwound in a vicious way so far last week. IWM M about 3%. Again, a big week but not uh one thing I will mention, it's been 6 weeks in a row for small cap and eight of the last nine with one little
[27:01] blurb, one little minor move higher. Uh Bitcoin for the crypto people didn't budge. Down 2.3%. So again, everyone who's continuing to say that or in the past has said Bitcoin's too volatile an asset to own. For everyone watching this that's deciding how much should be in their portfolio I'm going to say this over and over again. The majority of my portfolio moved into crypto. I'm very, very happy about that. I still have Marvell. I
[27:31] still have a little bit of Micron. I still have Nvidia. I still have Eli Lilly. I still have names like that, but the size of my AI portfolio is still significantly down and as I've said, it will never be as big as crypto again. Almost certainly based on my views and what's going to happen. There may be times where I add to AI for trading purposes. But the way Bitcoin is trading and hanging in there when everything else is blowing up this is a different kind of story as an asset class. The move index
[28:02] rates, you didn't get a big move in there. So this was really isolated to the equity market. So now let's go through the AI bear talk. So again, this is something that A16Z put out and I'm going to keep saying it. This index people have no idea how to read it. They just continue to say things about it. Then they mock out Jevons paradox. No matter who you talk to, no matter where it is, if you lis- listen to Jensen Huang
[28:32] and I guess people have this viewpoint that you can't listen to any of the people involved in it because they're going to lie to you. This is not 2007 guys and this is not 2000. You actually need to listen to this cuz here's the one thing I will tell you with absolute certainty that occurred through the dot com bubble. The dot com bubble was built to have things like your smartphone, to have your app store, to have all of those things. That's what it was being built for. It ended up being right. The problem was we didn't have any money coming in to offset it. So, everyone's
[29:04] trying to find a way to create an image that the money's not coming in and yet GPU rental for the old ones, not the new ones, it just continues to sit there. And I'll as I'll show you This is This is a completely different world. So, Kuppy, who wrote a piece last year that I highlighted in terms of the ROIC risk, and I I bring this up I don't know the dude and I really don't care. This is not to pick on people. This is just basically to say people believe to compare this 100%
[29:37] to in this case the shell situation. Shell and open anything re- regarding intelligence. There is a similarity, which is the price goes down and the build out depends to some degree on the prices being up. The token prices are not the way to do this. The revenue is because in the case of oil, guess what? If oil goes down, well then revenues
[30:07] have to go down. In the case of AI, that is not the case. And so far, that has not been the case as this has gone down. The revenues can come, as I'll show you, from so many different places that this is not a one commodity thing. So, to compare it to oil is honestly, aside from the hey, this price is going down, I don't really use this thing that much and I'm not saying he doesn't. I'm just saying the majority of people, based on the data that I have, don't use it. They don't see any benefit for it. They
[30:38] don't think solving cancer matters. There's There's money with that. They don't think that solving math problems that go to physics that allow us to do things to make more efficient that no one would want to pay for that. They would like to pay for open source and cheaper models. It's just not true. I don't know what's happened to Carl Quintanilla, but it seems like he's just having a fun time putting out um bearish stuff. I mean, almost every It's almost like he's become a a bear. Must be some kind of way to get more people to pay attention. I have no idea, but I've been yet presented a credible
[31:08] framework as to where this money is going to come from. The reason I'm bringing this up, and this is not from Carl, this is from Barclays desk. The reason I'm bringing this up is nobody in 2009, so this is after the dot com bubble, nobody knew this was coming. Nobody did. When the iPhone came out, nobody knew where the revenues were coming from. And right now, the one thing I know with certainty is Barclays desk has no clue as to where the revenues are coming from. Where the revenues are coming from has a
[31:40] huge impact. Do they know where humanoids are going to be? Do they know how big consumer agents are going to be? I added everything here. We don't know. So, to say in 2030, which again is what we're getting to, that the revenues won't be here is just impossible for them to be right on or for the market to trade on. So, you follow the data and the extrapolation of the most recent numbers. Everyone else is jumping forward and saying, "I've been through this before. This is what it'll look like. This is too much being built." It's just not the case. So, here's Andrej Karpathy this week
[32:12] saying what I completely know, and the reason why I don't listen to macro podcasts anymore because I'm listening to people speak about something they know nothing about because they don't use it. So, he's talking about the fact that there's two groups of people. Most people on the planet have little or no exposure. The remaining of them, and not a high percent, the remaining people are casual users using it for chatbots. There are very few frontier professional users. So, how could anyone
[32:44] be able to say with certainty what's going to go on? This is far more complex than oil. This is more demand right now than the dot-com bubble ever saw on the build-out. So, again, I can keep showing you guys this, but this is coming from Andrej Karpathy. He's been honest throughout the process. He's the one who got it wrong. Agents weren't going to be here a year ago. They They're not only here, it's gone through. So, the AI debate the people on X that are bearish is being conducted by billions of people
[33:15] who have barely experienced the technology, millions who are beginning to understand its economic power, and a few thousand who are already living in a future the rest of the world cannot yet see. That is what exponential innovation is. Aaron Levie, if you want to know where all the tokens will get consumed in the future, it's agent swarms. We only started having agent swarms in the last 2 months. That's why we've had consumer agents, and that's why I keep highlighting. We are at a step-up function in demand right now, which is why you're seeing more and more interest
[33:46] in GPUs at this current time. DRAM prices made new highs after consolidating for weeks. All we're seeing with prices on demand is going higher. And that's coming because there's more money showing up in terms of demand. They're booked out as I'm going to show. Uh Patrick Collison talked about agents. I'm not going to spend all this time on it, but thanks to personal agents, we will probably get somewhat more rational in how we engage with companies. Basically, you're getting to the point
[34:17] where companies are starting to see the ability to use AI agents, and that's coming across. I just listened to a pod uh an interview with AI engineers this week, where they estimate at this point that agent workflows at companies, agent workflows, are only being used by about 2.2 or 2% of the companies. 2% of the companies. Now, over 50% are using coding agents. But, using agents for your workflows, which is when the job
[34:47] situation starts to deteriorate, which I'll talk about later, that is a different story. So, when you're thinking in bets again, and you listen to people, the prospect of a bet makes us examine our beliefs. The the whole point of this is in almost every case with people saying they're bearish, and especially people that have been bearish for over a year, which is almost everyone. I challenge you to go into X, find someone who's bearish right now, who wasn't bearish a year ago, 2 years ago, 3 years ago on the bubble trade. And not people that
[35:18] aren't taking a short-term, I think the market is overdone here and going down. That is the reason why I wrote the mid-cycle slowdown in May. That is the reason why I sold that of my the majority of my positions. That is the reason why I wrote the fireworks show was over, and we will never see that again. We will never see what we saw in the first quarter again. So, I'm being rational with the fact that to make alpha, you need a gen to compute. Everything else to me is up for risk. So, unfortunately, when you're listening to people say there's a bubble, by definition, they can't be Alexander Wang. He hasn't been through one.
[35:48] You can't listen to him talk about it. You can't listen to most people in AI talk about it because they haven't been through it. Now, Gavin Baker has, and Brad Gerstner has, but people choose to ignore them. The reason I bring this up with what's going on is this people should listen to this week. This people should listen to this week. This is Tibo from OpenAI. These are all ones I listened to this week. This one absolutely you should listen to. Dylan Patel
[36:18] absolutely has the most and best information of any research place on the street, because he has set up his business to be vertically integrated, meaning he covers only the AI situation. He covers the data centers. He covers the semiconductors. He covers Anthropic and OpenAI and models all this stuff out. They spoke this week. Go listen to them. Go hear how things are. This was on Alex Kantrowitz, who's great. The problem is for people, Alexander is 29 years old. Tibo is in the early 30s.
[36:50] It's always good when you can find someone who's so mysterious that you can't even find his proper age. Dylan Patel is 30. When you're listening to people say the AI bubble has broken, they're not wearing a hoodie like me. They're telling you I've been through this. You don't know what you're missing. These kids don't matter. They've never seen this. The reality is these guys have built good businesses and their research has been dead on. Dylan Patel founded the company, SemiAnalysis, if you don't know this about it. Has 90 employees. It's in 11 different countries. In 2025, they had $20 million
[37:21] of revenue. Now they have $100 million in revenue. 90 employees, guys. This is the way AI is changing the world for real. So he not only knows this from his business being uncovering it, unlike the public banks, he's spending millions of dollars on AI and he's openly talking about it. So it's vertically integrated versus Wall Street. Exponential AI needs a new system. Why this wins. This is what my belief is is
[37:54] that all public companies are at risk. That's why I did the Stripe versus Citibank thing during the video last week. You're going to see this over and over again for me because we went from the industrial economy to the digital economy. From an investment standpoint, you only wanted to invest in the digital economy equity. Well, now no more private equity, no more private credit, no more VC, no more bonds, and no more public stocks. No more. You'll get some returns in equities for the next whatever 3 years, but the
[38:24] returns will be dwarfed in terms of the tokenized digital economy. That's what we're going through. The tokenized digital economy, places like this business. He has a hundred million in revenue. It's a big business, guys. That means his business I I I don't know where they would put it right now, but I do know this. You're probably talking about a market cap, I don't know if he went to market, billions of dollars. That's how fast these businesses are growing. And what he talked about was a year ago he was spending a hundred
[38:54] thousand dollars on AI. As of August it was up to 10 to 11 million. So, again, you're really talking about someone who's doing the work on it, but is also using it against his revenue that's come in. So, the question is how should you think about it? So, I wrote this. I went through it. People are not looking to make money right now. I'm just telling you the thing that I put out last week or earlier this week when the cost of intelligence collides with the demand for intelligence.
[39:24] AI agents and Vera Rubin are the place you want to be investing in for the next 12 months within side the agentic compute. And I put together those 15 names. Most of them are in the optical side as to who benefits. This week has been a blood fest, so you can go in and you're getting things at much lower levels. If you want to wait and wait for stability and wait for the hedge funds to be done liquidating, that might take another month. Um I don't know. It could end quicker than that, but based on how bad the numbers have been, would not surprise me to see more
[39:55] of this until we get into the hyper scalar numbers. Uh I did want to make sure that if you guys are interested in Jev, I had a lot of people that reached out. What is Jev? This video was fantastic, had over a million views uh as of last week. Very simple. It'll be in the uh all of these podcasts because there's so many of them. I still have I think two more to show you guys. I listened to at least seven or eight this week. All on the AI chip trade. Again, there's no way to forecast the exponential. So, you have to deal with
[40:25] the fact that you're going to have these ups and downs within side something that's going to go on. That's why in this week's AI video of what I did, I went through this to kind of give people the ability to dip jab. I know I did a lot of over the last 2 weeks, a lot of videos, a lot of stuff came out. The Vera Rubin report, remember you can go get this off the website. You can do this yourself. I did the Agentech consumer inflection. You guys can go do this on your own. This is the time to be doing your homework on these names if you're not in the position the hedge funds are. So, if you're sitting at
[40:55] home, even if you've had a drawdown in terms of things going, your entry point is getting better assuming there's not a bubble. And that's my job here is to basically go through and say we're not in a bubble phase, but we are in a deleveraging phase. It is a very, very difficult game to know when the hedge fund pain is gone because when you're dealing with $14 trillion of of positioning there and the growth has been enormous, it could take a while even if the the fundamentals are still
[41:25] strong and that's part of the leverage thing. I also included the technical side. Again, you guys that are new to it, I'm showing you this if you didn't go through it because there's a lot to use each week on this stuff. All of the technical sheets, all of fundamental sheets, all of the stuff in AI you want to be on top of. And again, I go through the different ways the resources and the stuff that are there. I give you the prompt on it so you can go through it. And that's basically what it does. It's taking 100 names, it's going through, you're going through the Jeff process
[41:55] which again, go look at the video I put on the subscriber site. It will show you how you can do this. And effectively, you are doing what hedge funds are doing and that's the other reason why the market structure is changing. The competition for hedge funds in terms of the speed to ideas is gone. I am about to list something and put money into something with my crypto. So, the tokenized world is beginning, guys. The market structure has forever changed. It is a deleveraging event. You're going to hear me talk about this more and more. Tokenization is the final
[42:25] part of the deleveraging stage for hedge funds. So, hedge fund risk managers I actually don't want to talk to the risk managers. Hedge fund senior people that want to reach out and talk, I'm happy to go through at least my thoughts on it so you guys can go through it. But, tokenization is going to change the way covariance matrices work and the way that margin is going to be provided. So, again, I go from 100 names down to 15 using this process. You guys can set it up yourself. And that way you are effectively doing what hedge funds do, taking thematic research, whittling it down, combining
[42:56] it with fundamentals and technicals and earnings progression, getting an answer within seconds. Watch that video. This is the one that you need to watch this week aside from the other two that I'm going to go through. But, to get back to it of his side in there, AI capex is already bigger than most people think. He says this could be close to $2 trillion or 5 to 6% of GDP next year. The big debate in whether revenue can catch up the infrastructure. He says OpenAI and Anthropic alone could potentially reach 600 to 700 billion.
[43:27] The reason I wanted to show this, they're going to finish combined this year somewhere around 170 billion. All he's saying and again, this is the the thing to me that I just don't think people understand. So, everyone was going through the Anthropic thing this week. So, Anthropic is on pace for 90 to 100 billion for this year. Last year they finished the year at nine. At nine. So, they're 10 times again and 10 times the year before that and 10 times the year before that. And right now he's saying combined for OpenAI and Anthropic
[43:58] maybe 600 700. So, let's take the bottom end, 600. Well, that would mean between the two of them they're doing about three times. So, he's talking about a rapid deceleration. This is already built in he's saying 600, 700, and that's basically what David Sacks and Gavin Baker and Chamath gave as their answers about a month and a half ago. Sorry to the guys who are focused on the way oil trades in the dot-com bubble they went through. If you're not in Silicon Valley
[44:28] and you're not actually talking to the people heavily involved. Brad Gerstner's an investor. He was at an investment at the investment day with Anthropic this week. He saw everything. He spoke on TV. I I I just don't even know. Um the biggest revenue opportunity may not be today's token pricing. Jordan says people are too narrowly focused on cloud cloud subscriptions. This is a critical thing. And per token API revenue. If frontier AI systems can perform knowledge work across coding, robotics, self-driving, and drug discovery, health care, chip
[44:58] design, and other industries, the revenue pool could be much larger than today's software-like products. Again, this is the part of what I said at the beginning. When you have exponential growth, nobody has any idea. The Barclays desk is not including any of this in where the revenue's coming from. They are doing basic human linear garbage work. End of story. You didn't predict the dot-com the the post-iPhone era. And trust me, if you did, you would have made a fortune and not be working anymore cuz you just would have gone out
[45:29] and bought the Mag 7. The Mag 7 dominated the world. We are going to see a similar thing in AI companies. I don't think it'll be OpenAI and Anthropic. I have no problem if people say to me that I don't think they'll ever be worth 10 trillion. Great. So, if they're worth one and a half right now to two, and next year they're still worth one and a half to two, they've won. That's the part I don't understand that people don't get. If they just stay at the same level, they've won. These companies are up so much from where they where they were before. AI
[46:00] could supercharge productivity in industries and flat head count and with flat head count and I think that's what we're going to see. So again, I bring this Barclays desk up cuz I just I think it's so stupid for people not only to write it, but then for a reporter to put it out there. It's only one reason to put it out there. To be bearish. That's it. Like what's the purpose for you guys reading that? It means nothing. They can't predict what this is going to be. Go look at their trading desk and see if they got Micron going from $10 a share to $80 a
[46:31] share. The answer is no. I guarantee you they were off by a factor of five at least. So you can't predict this stuff. No one can. And I'm not saying I can. I'm just saying that the path continues to move at this pace and unless we see a change in the revenue growth, unless we see a change in and I don't mean the revenue growth of a oh my gosh, in August it was slightly lower than July. That means they've lost it. That is complete garbage in terms of the Anthropic and Open AI stuff. Looking at a quarter return for a company and saying that
[47:01] that's the end of it is just ridiculous. Especially when we just launched Grokbot, we just launched Muse, Meta. You're going to have more and more people using other parts of AI where revenues are going to be coming through. Here's the other problem I have and again, we look at this buildout. Canals. We're comparing this stuff 5 years, 20 years. 20 years. 17 years. 7 years. We're near two of this and yet we're saying the bubble is here. We're saying
[47:31] there's been 10 trillion, this is a bubble. We haven't even started the bulk of the CapEx. It's just starting. If there's a bubble, I'm guessing. In fiber it took 7 years of buildout. In this case we're saying in year two while everything still looks positive. It's just insane. And again, we're doing everything out to 2030. I've never seen forecast in my career to reiterate out to 2030. We we've been dealing with what's the
[48:01] economy going to be in a quarter? Is there going to be a recession next year? Now we're talking about 2030? How many things can be different than what's expected in 5 years with exponential innovation? That's why I'm bringing this up all this 2030 stuff. This got brought out again. Bains has AI needs 6 trillion. They're going to be short by 800 billion. I mean, this is insanity. Out to 2031 you're giving me a forecast on how short you're going to be? So again, here's here's CoreWeave. Uh trading this is before Friday. This
[48:31] is after Thursday. Um again, trading in this range they've been in. Here are their clients Microsoft, OpenAI, Meta, Anthropic, Mistral Jane Street. Like are they going to default? I I mean, maybe down the road, not in the next couple years. Uh we cared about this last week. We got this. It's only two trill the two trillion could be 1.9 uh blah blah blah. Again, Jared Cuban came out and basically did the math after everyone was going through it saying, "Oh,
[49:02] they're coming in light." And he just went through it and I love what he said, "You have to do the work and run the math." Let me clear some things up. 100 billion end of year seems pretty straightforward. I I there's no way to look at it and not be able to come up with at least 90. More importantly, we're talking about whether they come in at 90 or 100 when as open as a 16Z put out the labs have added more revenue in 2026 than all of public software combined and we're calling them a failure or slowing
[49:32] down. Um Broadcom, this was a headline by the media likens many AI investors to casino gamblers. It's not what he said. Gavin Baker volatility on misunderstood news flow. Everything I've shown you today is misunderstood news flow plus the the the chest beating of the bears coming out seeing the bubbles pop then weak at the S&P is two two and a half percent off the all-time highs. We had deep seek day. We had liberation day. We had Kimmy. We had blah blah blah
[50:02] everything out and then a mistranslation of Hak Tan's comments yesterday because he basically did not say what they said he did. He did not say people were gambling on AI. He actually said that OpenAI and guessing whether they had a it was 70 billion or 50 billion and jumping to the gun and making decisions off that. That's gambling. That is hedge funds. That is AI agents trading. Make sense to me. Um you get all these jumps and all these moves. You also had CXMT this week and
[50:34] the fact that they were coming out with stuff. So all memory names were going to get hit. So you had all of this stuff happening and again I I feel bad for the people who do all this research and then something comes out and everyone immediately sells the stocks and then they get called up. I go through this as well. And Elon Musk basically came in. He got involved and he said, "Yeah, problem is RAM supply." So you know, whatever. And remember, here are the facts. Four weeks ago, Micron estimates signed agreements represent more than 35% of
[51:04] its revenue through 2030. So remember the 2030 forecast for what CapEx will be and how this will be the biggest bubble in history in five years. How we have a situation where Bain says we're not going to get the revenues in five years. How Gartner says the neoclass could blow up in five years. Here are the contracts. So the these are not forecast by those guys that have no no facts to them. That's just their guess, okay? And that's a fact. That is just their guess.
[51:35] Micron has signed agreements of its revenue through 2030. ASML AI chip demand is stretching into 2030. It already expects supply shortages through 2028. Nvidia-backed Lumentum sold out of AI parts till 2029. You have to literally go through and just realize what I'm showing you. Pe- papers writing how in 5 years there could be problems at the same time that real business is being constructed that
[52:05] is sold out until then. Amazon CEO Andy Jassy told him AI computer sold out beyond 2028. Nvidia just guided for 70% revenue growth. This is from a month ago or a month and a half ago. 70% revenue growth. It was expected to be I think 40. It It came in at 70 for next year. Jensen Huang guided to 70%, which is
[52:37] going to be higher than that just based on his track record. Uh >> [sighs] >> Meta Muse just came out. So again, as people are saying there's not They're forecasting where revenues are going to be and we're getting new progress. As progress comes out, as we break cancer research, as we do robotics, and as I'm going to show you everything is accelerating, revenues will accelerate, too. As the progress gets better, it gets easier for people to pay them to use it. Plain and simple. I don't know why this is hard to
[53:07] understand. This is the video I did uh on Thursday putting the If you haven't seen it, go watch it. Um again, it's a lot of it is is in this or some of it is. This is where we closed on Friday. This is the most bizarre thing to me. I'm listening to bubble talk and how this is ending and this is like I said, do you know how many AI bubble prickings we've had in this period? Like it's non-stop. Here's where semis are after all of the panic of July and all of the panic of this month.
[53:38] The SMH is up 50% year-to-date, slightly more than last year. I think the most important thing is the we've If you've ignored semis, you're ignoring basically the mag seven now. Let's assume that it started in 2019 when machine learning and the cloud and everything really started to to to become more of a a thing. The transformer papers, everything. All you had to do is invest in semiconductors. You've had seven of the last eight years effectively be 40% or greater.
[54:08] And yet people are saying the bubble has popped. You're just missing the opportunity. Regardless of what the charts look like of how fast things went up this year, Oh God. And here's the same thing in the S&P. And I just wanted to show this for people who have joined here to avoid the news, send this to your family members, send it to everyone. Compounding is why I got involved in the markets. My father was a construction worker, didn't get didn't finish high school, left school in the ninth grade, taught me handicapping, taught me math, taught me everything to go through. But
[54:38] what he really taught me was, I don't want to be a construction worker based on your hearing, based on your back, based on the electrocution times, based on me needing to siphon gas. I didn't want to do that. My mother handed me a book and it was on compounding. This is what the S&P 500 has done. If you were in the last 10 years, you got 17% annualized. If you were in the last 20, you got 13. The last 30, 12. The last 40, 13. This is not cherry-picking numbers. This is just using the last decade, 20 years, 30 years, 40 years, 50
[55:08] double digits in every case. Back to 90 years. This is what's happened, guys. It's been getting better recently. But this is what happened. Let's just take the last 50. So if you put $6,000 in 50 years ago, I'm sorry, if you put $10 in every month, just $10 for 50 years and ignored everything in the papers, the value of that $10 each month for 50 years, $10 is now $478,000.
[55:39] And this has got in a 2 and 1/2% dividend to just use it in there. Guys, compounding is your friend. And the S&P 500 and all investments, effectively equities represent innovation. So, if you believe AI is going to continue to have an impact on the world for now, you want public stocks, but you also want to involve in tokenized things as soon as you can. I don't think VC and any of that stuff is going to be able to monetize because they're cherry-picking out of the billion of winners, and I don't think you can do that anymore. Here's the log
[56:10] chart of the S&P 500. Are there periods of time where it goes sideways? That's why I just showed you what the returns are. But the reality is when you're at all-time highs, you're at all-time highs for the last 90 years. Could there be a correction that takes us back 20%? Yes, we've seen a bunch of those since 2019. It shouldn't be part of it. This is the thing that got me going. When I read this in a book my mother gave me, a penny doubled every day from day one to the end of the month, by the time you get to day 31, there's $10.7 million.
[56:41] That's why I got interested in not being electrocuted. This is the chart of where we are. I show this time and time again. And the one thing I know is that this is incredibly disruptive to hedge funds and leverage because it makes certainty gone. Hedge funds invest in certainty, end of story. They trade their book. They get out of it. I was in one, had my own for 20 years. It's about short-term arbitrage. It's
[57:13] about arbitraging something in the future, but not carrying long positions for long periods of time. We are entering a period right now of complete uncertainty, so leverage has to come down. For all the negative things you've heard on OpenAI and Anthropic, here's where they were 2 years ago valuation-wise versus today. And I'm this easy. Anthropic is more like a bit a peer. We're talking about companies that have gone from less than a hundred billion dollars in Anthropic's case, less than five billion dollars
[57:44] to two trillion, and we're saying they're not successful, and you're trying to pick the top of this. It's just moronic. It really is. SpaceX, how many people continue to say that SpaceX blah blah blah blah blah. Here's the valuation. These are just the facts. If you invested in them, you're doing great. Here's what the S&P did over the last year, 16%. Boring. Public equities are in trouble on a comparable basis to things that are moving on the hyper the the exponential. The exponential are AI trades. I gave
[58:15] you Anthropic and OpenAI, but this could have been Cursor. This could have been so many different companies, OpenRouter, that have been purchased. SemiAnalysis goes from a zero-based company or 200 million to 2 billion, whatever you want to say. If you do 20 times their revenue. Somewhere within here at this point because of the growth, you have to understand that public companies just can't grow this fast with AI. And you don't know where the revenues are going to be coming from because you're only focused on a few companies. That's not the case. Very important interview, listen to Zuckerberg and Alexander Wang. First of
[58:46] all, we had heard a lot about the relationship. Um this was a great interview on on Cheeky Pawn. I highly recommend um listening to the entire thing. It's about two hours. Um he talks about the compute and the spin not working out. He talks about the neo clouds and why they don't have as it is easy as they do. He talks about the fact that GPUs are basically higher than they were a year ago. The value of even an H100, the old ones, at this point is
[59:17] higher than it was last year. This is him saying this when talking about all of this. So, again, if GPUs remain productive longer than expected, then they can generate more revenue from the same hardware with an AI. This is just someone who's giving you the positive readout. If GPU prices are going higher, it's very hard to be bearish on CoreWeave and not bearish. You can be bearish on the equity. You can be bearish on the bonds. To say it's an AI bubble bursting is just very hard when the GPU prices are going up. OpenAI's GPT-6 may lead to
[59:49] robotics breakthrough as soon as next year. So, Unitree uh which as I was told by Emad Mostaque, who I I was at dinner at this week and shocked me. I didn't realize Unitree had only sold uh about 11,000 robots this year. Humanoids, robots. Um So, when the I This is really a big deal because it's a Chinese company, very important to the government, saying OpenAI's latest advances have
[60:21] accelerated prospects for a GPT moment in embodied intelligence. Think about what that means. It's two things. For all of you who say the Chinese are at the same level, which they're not, this is about Astra. Okay? We are two models ahead of Astra right now, which they haven't released, and my guess is, especially since we're not seeing as many Chinese new models come out, distillation may be over and the benefits of OpenAI and Anthropic, which was talked about by Dylan Patel, if you
[60:51] listen to the podcast, may not be released. This was also talked about on the Zuckerberg one. We might be reaching a point where we never see, or at least for a long time, the models better than Astra and Opus 5.5. That may be it for us. Because there may be a reason now to not have them come out to avoid distillation, and that would create a really big gap between the Chinese and the US, and this whole open-source thing would be in trouble because one of the major reasons for open-source is catch staying close
[61:22] is you release the model and then there's distillation. Well, if we're not doing that and if we're leading to robotics, why wouldn't they keep the models for robotics to themselves at this point? So, Frontier Model Intelligence, as was said by the Semianalysis guys, reaches a point where their revenues are not going to be about consumers paying for it or even honestly some enterprises paying for it. The majority of the big enterprises are going to continue to pay for Frontier models.
[61:52] But, the reality is most of their business going forward may be from biotech. It may be from from nano technology and materials. It may be from robotics. It may be from That's probably where it's going to be and that revenue hasn't even started yet. Astro makes a robotics breakthrough closer because it potentially solves the hardest remaining problem in humanoid robots, moving from a robot that can perform a program task to one that can understand a goal, reason through unfamiliar situations, and adapt actions to the physical world. This is how
[62:22] important it is for the humanoid explosion. So, again, personal agents, agent swarms, solving math problems, and robotics are all together. Do not get bearish AI compute with all of this hitting right now, guys, as I have my hoodie on. Brad Gerstner, Jensen Huang, Clark Tang, great thing to listen to if you want to feel better about AI. From the guy who's been right the entire time. Uh Frontier Labs are roughly a third of
[62:52] the business, not the whole story. He's basically saying, "Why is so much pressure Why is everyone focused so much on Open AI and Anthropic?" And he goes through the math of where the 10 trillion What it includes, like where the revenues are going to come from. And one of the things he says right off the bat is you have to understand it's 500 billion a year just for new compute the way it's always been. So, again, to listen to him go through this and to break down from someone who's been right
[63:23] consistently, and he talks and makes jokes about it, whose business is booming and he's involved with every company in there, to listen to him answer all the questions. He doesn't care about where CoreWeave's debt is. That's why he came out and said, "They're fine. Their cost of capital is actually lower." Which again, then people laugh at. They don't even go do the work, which arguably is true. So, again, as you go through this and you don't figure it out and you don't think about it, you're just responding to things. And I get it when prices are
[63:53] going down, your mind runs to places that don't make sense. I'm just telling you hedge funds are doing this. This is a hedge fund deleveraging event. Now, the Korean market was involved in the first part of the year. Retail got in. It led to hedge funds going down. This is another leg of hedge funds, but remember, hedge funds that are near closure don't have a stack anymore. Hedge funds that have been beaten down, like the ones I showed, that are flat for the year and now could be down 10 or 15% if they didn't get out of their positions, they're going to take things
[64:23] off because they don't want to be down 40 and they've already seen a down 50. So, you start running into issues and primes start getting worried and people start coming in and investors are going through it. It's not a fun story, especially when it's happening across the board because that makes the primes even more worried. So, again, the frontier labs are a slice, they're not the engine. Bucket one, classical compute. I'm just giving you the three that he mentions here. You should really go through it on your own. I'm going to try to start to speed this up a little bit cuz this is a long
[64:53] video, but there's a lot of important things in there and hopefully you guys are getting it from it. Uh this is talking about his track record where he's basically been right throughout the entire time of doing the interviews in terms of how much compute would be spent, how things would do, what would happen to GPU prices. All of that's been there. And then Brad on the inevitability of AI. this is the most important exchange in the conversation. Brad is pushing back on Jensen's own answer to Ezra Klein, where he gave something like there's two or three years of visibility before the bubble question becomes live. Brad reframes it.
[65:25] This is not a CapEx cycle with a visible end. It's a fundamental re-architecture of global compute permanently. Huang's one-word answer was forever. So, I completely agree with this and this is the reason why the dot-com bubble was a mismatch between the demand not being there while the supply was being built on leverage by the companies that had nothing to do with the demand. This was about the fiber companies
[65:55] putting up the rails. This is very similar to crypto. Crypto didn't have the users, they didn't have the demand. Now, they're going to have the demand from the AI agents, which is why I say it's the same thing to me as the end of the dot-com bubble for crypto and why my positions are in crypto. His point, which I completely agree with, there is no end to this build-out. There is no end. And because the progress is driving the demand and the progress keeps accelerating, as I've shown you, I cannot make it any clearer. Token use has gone higher as progress
[66:27] moves forward. As the models get better, they're able to do more things that generate revenues. That is not a guess, so you guys know. That has been a fact. That's why listening to people saying they know what the revenues are going to look like in 2030 when they didn't even get Micron's revenue is literally ridiculous to listen to these people, including Gartner Group. It is a joke. Here have been the revenues for Nvidia relative to the consensus. And what Brad Gerstner was showing on
[66:57] here is where it was just a year ago. This was the consensus, again. This is what the consensus was for Look how far off the street was. And yet the stock price is unchanged as I said. Think about it. If you're a hedge fund and you thought the street thought that for 2029 it would be 305 and now it's at 982 a trillion dollars and you watch this happen and made no money.
[67:27] Of course you have to get out of things. Of course you do. This is the only thing right now and this is more of a slowing thing. I think what the market is going through is that the capital coming to the market is just causing the market to have trouble dealing with this stuff. As more money needs to be raised and people go I'd rather I got to make sure I have money for open AI. I have to make sure I have money for Anthropic. And whether it's mutual funds or this they have to raise cash.
[67:58] And I think all of this stuff that's been going on for SpaceX, for the bonds, for everything, the way that rates have moved, it's all been related. There's plenty of capital in the world for the AI build out. If it ends up being 20 trillion and not 10 trillion, there's plenty of capital but it will have to come from somewhere. If the revenues are coming in, people would rather own bonds at some point in CoreWeave than they would the Treasury. Absolutely. Someone this week said would you rather buy the equity or the bonds and I said I'd rather buy the bonds.
[68:29] That's where my mentality is at this point. I just think at this point with the panic that's going on and what people are doing if I had to have a choice, I'd rather have the debt at this point at these levels. When Genius Failed. So again, we're going out to 2030, we've lost all certainty. I watched as LTCM had trades on like Brazilian Telebras preferreds versus ordinaries that traded at a spread. The preferreds traded at a higher price but every day the privatization was going to happen, that was all going to converge to zero, the spread. So if
[69:00] you're trading at a 5 to 10% spread and 6 months from now you're going to have a privatization event, that should collapse to zero. So, if you put that trade on along with a bunch of other spread trades on, which is what LTCM did, you think you're going to make money. But, here's the problem. This is the lessons from leverage. And I want to make sure that for those of you this is even for people managing money at a freaking hedge fund who've never thought about this stuff. Lessons from leverage. Your fundamental thesis can be correct.
[69:32] But, the timing. If the AI trade, if all of Nvidia, if all of a sudden next year Nvidia is up 300% and goes back to where it was before, nothing changed. It's just the timing you got wrong. You were in it, you couldn't wait because you've got investors who are measuring you on a 12-month basis. The volatility. We've had a dramatic shift in the Vixy Q. Single name vol has exploded. Index vol has barely budged. You have no hedges. So, you can be wrong on volatility of the markets and the relationship of
[70:02] volatility to the index. The financing. You can have rates move higher. That can cause a problem on things. You've got to borrow more money. Maybe the banks can't lend as much or maybe your cost of capital is going up and your investors are like, "Hey, I can now invest in Libor plus 400. I don't need you guys anymore." Crowding. There can be too many people in the trade, which I think is one of the major problems, while liquidity is dropping off, which again, we've seen. Correlation. You can have breaks in correlation for periods of time. Well, we're definitely seeing that
[70:32] right now where the Mex Bolsa is going higher at the same time that the Qs are going down. So, if you've been sitting there riding a winner of Okay, I'm going to be long Qs. I need some shorts on it. Oh, the Mex Bolsa is trading at 52-week lows. That's a good short. Blah blah blah. Liquidity. Obviously, this was a major one with LTCM. Once it went through, they couldn't get out of it. And then the position sizing and I think situational awareness. Gone through it. That's why I did this stack on moral philosophy. I've done presentations mainly for mutual funds at this point. This is the way I see this going. We
[71:02] went from a normal distribution curve where we knew that 5 years from now Ford would still be in business. Again, unless there's some serious recession, but in this case with growth going, we are getting a weird world where we've got losers, think retail mall names, winner, think Amazon, and then we have a bunch of names right now like bank stocks that are sitting in the middle here that I think will gradually go into here leaving very few AI winners, and I think over time more of these will come over here.
[71:32] The losers will be have no growth. They will have their growth eradicated from AI. And that's just the fact. Here are the hyperscalers. Remember, arguably the most important space aside from Nvidia and their GPUs is what the hyperscalers stock price is doing. If capital's the issue, their stocks made new all-time highs as a group. And again, this is Amazon. This is Meta. This is Google. This is Microsoft. This is those four names.
[72:02] I did want to bring this up. Uh one one data point um which I think is important and becoming more important. Consumer confidence made the lowest level since 2013. Yes, the lowest level. We are below when there was a pandemic and millions of people were dying. We're lower than that. That's how bad things are in the way people view it as we go into the midterms. Since 2014, the unemployment rate here was 200 basis points higher than it is today. To show the importance of this from the
[72:32] jobless situation, jobs is what drives this historically. This is the unemployment rate, and this is now that consumer confidence number inverted, and this is back to 1970. It is the jobs market that drives this historically. Now, we've seen the jobs market slightly weaken, but this year has been fine. But like I've said before, the jobs market is far weaker than economists will tell you. This is inside the number of that consumer confidence. This is the jobs plentiful versus hard to get inverted
[73:04] again. So, hard hard to get jobs versus plentiful, you've got, again, the It says the unemployment rate should be closer to six, just like it was back in 2013. Uh Warren Pies put this out uh last week. And again, financials and tech payrolls are shrinking. Aside from nursing jobs, health care jobs, we are negative now for whatever it's been, 18 months. And he's just highlighting that this is
[73:34] a dramatic change in terms of what has gone on in the past. And we're entering a new period where the agent workflows are about to be adopted. So, I do think you're going to have a labor kind of situation. You can view how that matters. It's not a recession type thing. It's a profit margin type thing, but it's also a politics thing. It could be a Fed thing. Uh we'll just have to see. And remember, another way that we've seen this is we're right now looking at uh earnings growth in the S&P at 29%. So, up here.
[74:06] And this is the year-over-year jobs creation relative to it. So, to finish up, very short week in crypto. I did this video. I continue to go through learn the stack. For those people who have joined, remember to watch the videos. I'll have another one come out this week on It's been a busy week, obviously, with the video I did on the AI side. I will continue to go through it. Um last week, we went through the Oracle data and interop middleware and the settlement and smart contract rails. For those of you interested, especially in tokenization, I think you should
[74:38] sign up if you haven't already uh particularly on the hedge fund and mutual fund side and really start to get these. Uh I think you should be doing your homework now. As you're getting as you're watching the horrible trade in the AI thing, crypto is going to be a very different thing next year. Um privacy is the theme that I covered last week, very important. Here is the month-to-date. So, the index is down 7% Bitcoin's down 2 and 1/2 since August again, still crushing it since June.
[75:08] Same thing as AI has worsened, they have done well. Uh still bull market here. Did see some weakness in the names on the 50-day. Only 20 names above the 50-day, but in terms of the 200-day and what's rising in the slopes, good. And again, a correction within side a very large market. Bitcoin is hanging in here. I don't think we're going to see a move higher in Bitcoin until we see some stabilization in the hedge fund scenario. Uh but hopefully it'll be able in here. It would not surprise me if it
[75:39] kept meandering back down towards the 75 area. Uh just in there. And the final piece of information, which I will cover in a paper this week, related to crypto. I think it is shocking this number that their net interest income could reach $107 billion according to them next year. This money, as I showed last week with Stripe, is about to come out of the market. All right. Long one, an important one. Um I need to keep you guys happy. I gave you a hoodies up
[76:11] uh week. I'll see you next week. And if big things are happening during the week, uh I will keep you up with one, two, three, who knows how many videos I might have to do this week because of all the news going on. Stay calm. Don't worry about the bubble, but for hedge funds, this is the end of leverage in hedge funds uh as a rising situation. See you.