Something Is Wrong With Bitcoin
- Bitcoin’s relative weakness leaves a short squeeze possible, but unconfirmed. Spot buying against persistent perpetual-futures selling and elevated token-denominated open interest creates squeeze conditions. But “complete apathy towards this space” remains the problem: shorts need a catalyst that changes their bearish thesis; otherwise, they can sell another bounce.
- The panel split between buying a bull-market pullback and preserving cash through year-end. Yan sees declining inflation, eventual Fed easing and a constructive one-to-three-month setup, with no current reason to expect a correction exceeding 20%. Kevin’s October models show deteriorating growth expectations, tighter policy and little liquidity cushion; he needs something like August’s policy intervention before overriding that caution.
- José is deploying cash into AI exposure because he believes accelerating adoption overwhelms short-term market noise. “I just can't fathom how you can sell assets during the singularity,” he said, citing Nvidia growth of 80% from a $200 billion base and hyperscaler capex growth of 75%. The valuation debate includes Nvidia at a quoted 12 times forward earnings, but also the warning that analyst estimates can lag falling prices; one panelist said Meta was his largest individual stock and saw its AI optionality as barely priced.
- The equity decline contains substantial rotation beneath the technology sell-off. The panel cited the Nasdaq down 11%, the S&P 500 down just 4% from its all-time high, and an equal-weight index reaching a new high. Nvidia’s additional uncertainty concerns the potential consequences of allegations around Jensen Huang; speakers disagreed over whether those risks would amount to anything material.
- Anthropic’s planned IPO will test the market’s capacity to absorb another enormous supply event. The discussion cited roughly $70 billion sought, a $70–78 share-price range and a valuation around $1 trillion; several speakers preferred waiting. Separately, disputed SPV interests could become “a chain reaction of lawsuits” if transfer restrictions are enforced, potentially constraining float while leaving investors fighting over ownership.
- Onchain equities need financial products that create durable reasons to move assets onto blockchains. Covered-call income, smaller position sizes and collateral reuse offer a stronger mechanism than activity-dependent trading fees. For the HyperCore-connected L2 under discussion, the test is whether developers build distinctive options and liquidity products; another launchpad cycle risks becoming “Robinhood 2.0.”
- Fresh crypto allocation depends on finding businesses with real demand and tokenholder returns. José still holds four or five longstanding positions, but wants another Hyperliquid-like business before actively adding. ETH’s store-of-value pitch drew neutrality, while MegaETH’s announced $30 million inference-marketplace initiative prompted a more interested discussion about whether an AI business could sustain attractive margins.
1. Bitcoin has squeeze conditions, but the macro panel remains divided
Yan’s immediate concern was Bitcoin failing to participate in a strong equity rebound: “when the markets start to fall, Bitcoin will fall with them.” The constructive counterweight was persistent spot demand against perpetual-futures selling, visible in cumulative volume delta and predating Saylor’s purchases. Token-denominated open interest was at annual highs or very high levels, suggesting substantial participation and accumulated shorts.
The squeeze still needs a catalyst. Yan argued that shorts can tolerate losses or add exposure if they retain their bearish thesis; a temporary bounce is insufficient. The desired sequence is bad news arriving without further price declines, followed by sustained buying that makes continued short exposure uncomfortable. His comparison was August’s forceful squeeze, which he said was triggered by buying long-term bonds and Treasury securities.
Yan remained constructive over one to three months and longer horizons, seeing a bull-market pullback and no present reason for a correction of 20% or more. He believes inflation has peaked, with some leading energy indicators declining, and expects Fed cuts next year, although December remains uncertain. Historical strength after midterms supports a six-to-twelve-month view.
Kevin’s pushback came from October’s models: global growth expectations had deteriorated materially, the political LEI indicated a shift toward tougher policy, and liquidity measures showed little capacity to cushion a drawdown. Those growth forecasts look six to twelve months ahead and can change with incoming data. He preferred retaining accumulated cash through year-end, while allowing that an abrupt policy change, like Bessent’s August announcement, could justify temporarily overriding the models.
2. Rotation supports the broader market while Nvidia carries extra uncertainty
Jason framed the Jensen Huang controversy around potential disclosure of sensitive information. One speaker leaned toward little ultimately happening, citing the claim that Trump had already been briefed in 2025. Others highlighted downside scenarios, a purported shareholder lawsuit and prediction-market odds around 40% for Huang leaving by 2028. The discussion treated the underlying espionage allegations as unresolved.
The panel’s broader-market evidence was rotational: the Nasdaq was down 11%, the S&P 500 only 4% below its high, and an equal-weight index had reached a new all-time high. At the index level, the market looked oversold, while more than 50% of constituent companies were reportedly overbought. That supported the argument that capital was moving from AI and technology into other sectors.
3. José sees accelerating AI adoption meeting increasingly modest expectations
José’s framing: “I just can't fathom how you can sell assets during the singularity.” He cited opportunities growing about twice as fast over the period from March 2024 to October 2025, Anthropic recurring revenue growing 10% month-over-month in October, Nvidia growing 80% from a $200 billion base, and hyperscaler capex rising 75%. He also cited a 50% increase in H100 rental prices. Several major AI stocks were down year-over-year before Claude Opus 4.5 came out, which he said gave the trend a boost.
A speaker highlighted Nasdaq forward valuations returning to pre-COVID, March 2020 levels and Nvidia’s quoted forward P/E of 12. The pushback: forward earnings depend on analyst forecasts, which can lag price declines and then keep falling. José countered that forecasts had consistently underestimated results over the previous three years. Valuation was presented as useful over 12–24 months, with limited value for timing the bottom.
A speaker’s depreciation exercise started with roughly $800 billion of capex this year. Maintaining earnings as depreciation rises requires EBITDA growth around 7–8% for Microsoft and Google, 12% for Amazon and 16% for Oracle. Under the speaker’s terminal-value assumptions, the additional growth priced above those hurdles was only 5.6% for Microsoft and 3.3% for Alphabet. The speaker considered that pessimistic relative to current growth and expected continued positive surprises.
One panelist said Meta was his largest individual stock, although modest relative to his former crypto concentration. He argued its advertising business already benefits from better targeting, while he sees almost no value assigned to AI execution through Muse or distribution across WhatsApp, Facebook and Instagram. A cloud business would add another option. José, having sold some crypto and accumulated cash, said he was now mostly buying indices, reflecting conviction over roughly twelve months to three years.
4. Anthropic’s IPO combines a supply test with uncertain SPV ownership
The panel cited a proposed raise around $69–70 billion, a newly reported $70–78 share-price range, and pre-IPO trading carrying a roughly 20–25% premium. Jason said a valuation around $1 trillion was more manageable than earlier discussion of $1.5–2 trillion. Strategically, speakers understood going before OpenAI: a weak competing listing could close the window. One speaker argued that “the IPO isn’t for you”—the company’s financing needs drive the decision.
Jason questioned how another large issue would be absorbed amid weak liquidity and heavy supply into early 2027. He thought Anthropic lacked SpaceX’s automatic index-related demand and preferred waiting. Another concern was prospective junk-bond status and access to borrowing, compared with hyperscalers’ investment-grade funding, with credit conditions potentially less favorable in 2027. Speakers differed on whether the first few weeks would reveal broader-market fragility or mostly reflect IPO mechanics.
The SPV concern was explicitly conditional: a speaker had not seen the shareholder agreement, but said Anthropic had warned against unauthorized share and derivative transfers. SPVs formed around the $150 billion and $330 billion rounds could represent billions of dollars. If the company invalidated interests or canceled shares under those restrictions, buyers might sue sellers, sellers might sue Anthropic, and layered SPVs could sue one another.
The proposed market consequence was unusually sharp: litigation lasting years could immobilize supply and support the share price, even while investors disputed ownership. “Low free float, high FDV” was the panel’s comparison with familiar crypto structures. Whether sellers retained shares or lost them depended on contractual provisions the speakers had not inspected. The closing joke captured the uncertainty: “I'm betting on lawyers.”
5. Onchain equities need useful products beyond another launchpad cycle
A speaker described improving onchain metrics in June, July and August giving way to disappointing activity and price charts across Robinhood-related projects and Solana. The earlier tokenized-equity proposition often depended on 3% AMM fees funding buybacks or yield. That mechanism was vulnerable to the same downturn that reduced trading activity: attractive distributions depended on continued volume.
The stronger example was a covered-call vault for a non-dividend-paying asset. Operators sell calls to generate income; when assets are called away, being forced to sell at a higher price affects the eventual return. The panel emphasized smaller position sizes, professional management of volatility and strike selection, and tokenized collateral that can be reused or “looped.” Those features could make holding shares onchain financially useful.
For the HyperCore-connected L2, the opportunity was direct access to Hyperliquid’s order books, detailed market data and liquidity for options strategies, pre-IPO products and expanded spot markets. A speaker stressed that launchpads are easy to replicate and monetize, whereas useful applications require sustained development. “We haven’t created anything truly valuable yet” captured the criticism of turning tokenized shares into another speculative issuance cycle.
The host pressed for a falsifiable test: what would demonstrate a different outcome? The answer was options vaults and applications using HyperCore in ways unavailable elsewhere. One speaker proposed monitoring the first one to three months after mainnet; another objected that meaningful applications might take six months or a year. They agreed that early development should point toward those capabilities, with launchpads alone providing little confirmation.
6. ETH’s narrative remains unsettled while crypto investors ask for real businesses
A speaker stayed neutral on Ethereum’s renewed store-of-value pitch, seeing both possible adoption next year and a “last hope” interpretation. The quantum-risk hedge argument faced competition from assets such as Zcash. In one conditional price discussion, a speaker paired ETH at $2,500 with Bitcoin returning to $85,000–90,000 and rejected ETH at $1,000 in that scenario. Direction remained uncertain.
José still owns four or five longstanding crypto assets and had trimmed one into unusually optimistic news. He could not identify what would bring him back to active allocation beyond another Hyperliquid-like product with genuine market fit and most fees returned to its token, preferably in AI. The pushback was economic: AI requires capital and faces deep-pocketed competitors, making that distribution model difficult.
Inference remained a theoretical possibility. The panel cited MegaETH rebranding to Mega and allocating $30 million toward an inference-provider marketplace, arguing that its optimization skills might transfer. Speakers called the pivot sensible while acknowledging that the original plan had struggled. Elsewhere, HYPE retained supporters because genuinely good crypto businesses were scarce; José’s broader AI conviction remained rooted in daily use, with individual investments still open to reassessment.
Full transcript
I just can’t fathom selling assets in the time of the singularity.
American stocks lost over $2 trillion in value.
When you see markets rising and Bitcoin stagnating, that’s probably the worst thing.
I don’t think anyone cares about the affair, do they?
Well, CEOs have affairs all the time. This is more likely a case of espionage for the CCP.
I think prediction markets are now giving it about a 40% chance of going away by 2028.
Nvidia’s forward P/E is 12. That’s it: 1, 2, 12.
If it looks like Robinhood and smells like Robinhood, then it’s probably a lot like the changes at Robinhood.
Everyone says that AI will take away lawyers’ jobs.
Just not with such a mess as now.
Some billionaire who doesn’t like to get up early just sits down at his computer at 10:30 instead of 9:30.
1. Is the Worst Behind Us?
Last time we recorded the previous 2 episodes, there was huge market news the next day that we unfortunately didn’t have time to cover. So I warn everyone listening to this: you know what happens.
Last week, we recorded the day after the election and briefly discussed the possible consequences. Now that we’ve had a week to think about it, I’m sure we have clearer thoughts on this.
The day after the release, we saw one of the biggest stock market drops in recent memory. American stocks lost over $2 trillion in value. Then, on Friday, there was a small rebound. Yesterday was not bad. Actually, yesterday was pretty good. Today, the market is starting to falter.
So, my first question is: now that you’ve had time to reflect on the price movements of the last few sessions, do you believe the worst is over?
Yes. I was going to say no, though. A slightly hesitant “yes” comes out.
I think there are things in crypto that look good, and there are things that don’t. The negative side is more obvious, especially yesterday. When the markets are going up and Bitcoin is stagnant, that’s probably the worst thing, because you know that when the markets start to fall, Bitcoin will fall with them. You always rely on salvation in the form of both sides growing at the same time.
So I think the strength of yesterday’s rebound relative to Bitcoin’s stagnation was quite shocking. That’s probably what I dislike most about this situation: the complete apathy toward this space.
At the same time, I believe we’re seeing a significant divergence between the spot market and perpetual futures. If we look at the cumulative volume delta, we see steady selling in futures against a backdrop of constant spot demand. This is partly due to the emergence of Saylor, but the trend was there even before he entered the market.
So I think this is a positive signal. Open interest is at annual highs, or very high levels, across tokens, which I believe is a better indicator of market participation. The conditions for a fairly serious short squeeze have already formed.
I think the scenario is this: enough bad news comes in, the asset stops falling, and then we see either a gradual, steady rise or something more dramatic, like what happened in August, that gives the rise momentum. That massive short squeeze was triggered by buying long-term bonds and Treasury securities.
The situation with BTC looks good. If you look at the liquidation charts, we see a lot of activity on the long side, but the accumulation of short positions is also significant. The conditions for a squeeze are there, but we need a catalyst that will make the market rise, causing traders to worry and close their short positions. Traders can hold losing shorts if they’re still bearish. They need a reason to believe that the price will continue to rise, not just bounce back temporarily. Otherwise, they’ll just build up their positions, because the basis for their thesis hasn’t changed.
So I think there’s a local support zone here. Looking ahead, the year after the midterm elections was historically successful, wasn’t it? That’s something you can rely on, although it’s a 6-to-12-month forecast, not something instantaneous.
I also believe the peak of inflation is behind us. If we look at today’s indicators, growth was mainly driven by high energy prices. And if we look at leading indicators, everything rests on energy and its secondary effects. Some leading indicators of energy prices are trending downward.
It’s hard to say for sure how the war will play out, since the president can still act through executive orders. But I think it’s quite logical to think that the worst is behind him, as funding will become increasingly difficult to obtain, the voter base is fragmented, and the issue is becoming unpopular and, accordingly, less attractive to promote further.
I think all of these things help. Stocks didn’t fall that much, but they’re still significantly down from their all-time highs, right? It’s still something, and there’s no reason to believe we should enter a large-scale correction.
Overall, I still believe we’re in a bull market. Bull markets are known for double-digit percentage pullbacks, so this is definitely one of them. I have no reason to think we’re heading for a deeper correction of 20% or more. For all these reasons, I think it’s entirely appropriate to take a constructive stance now.
The mood is also terrible. It’s probably exacerbated by the fact that we’re mostly in the crypto market, at least in the chats and all that. I think it’s particularly noticeable in crypto, so people are exaggerating it to some extent. But things seem pretty bad in the stock market as well, because people are mostly focused on the assets that have suffered serious losses.
All of this together creates a pretty good situation for now, but time will tell.
Are you referring more to the situation by the end of the year, rather than trying to make any further predictions? More like the short term, say 1 to 3 months?
Yes. In the long run, I’m definitely optimistic. I think next year will be difficult for many reasons. But I also fully expect the Fed to cut interest rates next year.
2. Why the Macro Models Say Wait
As for December, that’s a separate conversation given the latest data. But I’m in the camp of those who believe inflation will decline, so I do expect more policy easing. I’m constructive about both time horizons. I’d need new information to change my mind on this.
I’d say I’m a little more cautious. I’ll quickly go over a few points. This is what we discussed last week on Market Matters with Kev and Scott, so if anyone hasn’t had time to watch it yet, we broke it down in more detail there.
I’ll briefly go over what the market and macro models showed at the end of October, because macroeconomics is mostly a one-month game. You get data that comes in mostly monthly. We analyze what happened at the end of October and see that growth held, let’s say. Growth prospects looked good. They were tight for a while and then fell sharply during October, meaning our global growth outlook is not encouraging right now, given the events of recent months.
Again, this is a forecast for the future, from 6 to 12 months out. Of course, everything could change when new data comes in by the end of the year and in January. But by that point, the global growth forecast had deteriorated significantly.
Inflation, as Yan mentioned, remains fairly stable. I don’t think we should expect another big jump in inflation. Our political LEI indicator has declined significantly, meaning a lower score indicates tougher policies and a higher score indicates looser policies. So we’re clearly observing a change in political course.
And another chart—or rather, 3 charts: liquidity, the main driver of the markets, as well as the level of liquidity and the margin of safety—looks simply terrible. If you look at these 3 factors together, it becomes clear that the market has almost no resources left to support prices during a pullback. There’s very little margin of safety left in liquidity. Growth prospects are not the best. Inflation still remains a bit of a problem.
So my opinion is: I don’t know. I wouldn’t rush into the market now to take on unnecessary risks at the end of the year. If we get something like a policy change, like Bessent’s announcement in August, that would be a signal to ignore your models to some extent and at least get involved in short-term trades, because they don’t account for abrupt policy changes.
3. The Nvidia Spy Scandal
My overall opinion—and I stand by it—is that I don’t feel the need to take excessive risks until the end of the year.
I think the situation with Nvidia also significantly hinders attempts to assess the real risks to the markets, don’t you think?
I don’t think so. I’m sure you have your own thoughts on this, but hardly anyone cares about the affair itself, right? CEOs have affairs all the time. It’s more about the spying aspect for the CCP: whether something was passed on that shouldn’t have been, and what that means for Nvidia, its management, and the implications of the leak of secrets.
This is something you need to think about as a market participant. Although Nvidia shares have fallen significantly, when you analyze everything together, I don’t see a need to use the cash I’ve accumulated in recent months. And I don’t think I’ll regret having that cash in a month or two.
I feel the same way as Jason in that there’s no need to rush, though I still have all my assets left, so I’m not completely cashed out. It’s just that spot assets have sagged significantly right now.
Actually, this pullback made me feel more confident about the long-term prospects of the market. I think pullbacks like this only reduce the likelihood of a big bubble like the one in 2000. That’s still etched in the memory of many people.
Although José studied valuations, the market behaves differently. It simply doesn’t value assets that aggressively. If you compare it to the dot-com period, there was an increase of 30 times from the current level.
The difference in the scale of this bubble is obvious, isn’t it? I think a correction of 10%, maybe 20% to 25%, would be significant. The Nasdaq is down 11%, while the S&P 500 is down only 4%, so there’s still room for further decline.
The Nvidia situation is a dark horse. Nvidia had a pretty good relationship with China, right? They sold a lot to China. There’s so much we don’t know here. I think Trump was briefed on Jensen’s relationship with this woman in 2025, and he’s met with the president multiple times since then, but nothing has been done. If there were a real risk that he’d leaked serious information, someone would have reacted sooner. Of course, you can’t underestimate people’s incompetence, but this will most likely end in nothing.
Yes, it’s most likely that this will lead to nothing, but there are some interesting negative scenarios. I think there’s a real chance he’ll leave. It seems market forecasts now give about a 40% chance that he’ll go by 2028.
They already have a class-action lawsuit filed against them. Is this securities fraud?
Kind of like that, right?
Why is this securities fraud?
Because investors weren’t told that the CEO had a relationship with a CCP spy. That’s misleading investors. It’s information they should have known before investing. It’s all securities fraud.
Just read Matt Levine, will you?
But there’s already a class-action lawsuit against them, right? That’s probably pointless, or something. I don’t think anything will come of this, but I’d be surprised if the story were simply glossed over, comparing it to Bezos’ or Gates’ scandals. Again, as Jason said, this is a completely different situation, right? I think the market will be digesting this for a while.
I also think people are changing their views and becoming more optimistic about non-AI infrastructure companies. They’re watching margins grow in other industries. I think the market started moving in that direction over the last year. I don’t know how long it’s going to last, but my optimistic prediction about AI has always been that it’s going to be good for the entire economy and benefit all these other businesses. You can see a lot of that. The S&P 500 is only down 4% from its all-time high, and it has a huge weighting in many of these technology companies. It’s less than QQQ, but still huge, and it’s only down 4%. That shows you how strong everything else was.
Yes, last week the equal-weight index reached a new all-time high. We’re seeing a lot of rotation. This isn’t just a massive sell-off where everyone sells market beta and everything falls. There’s been a big rotation from AI and tech stocks to other sectors of the market.
4. Never Sell the Singularity
This is a very interesting chart. I don’t think we’ve ever seen such a divergence, with the market itself, at the index level, oversold at the end of last week. But when you look at the composition, the number of companies trading in the overbought zone was well above 50%, again supporting the idea that there was a major rotation and not just a massive sell-off. So far, it’s been a rotational trade.
How do you feel about all this, José?
In general, I think people have forgotten how to be optimistic and believe in something. I can’t fathom how you can sell assets during the singularity. As far as I can tell, we’re literally in a singularity.
There was a report published this week, or maybe even last week, from METR, which does very high-quality AI assessments. Anthropic gave them access, and they said that the opportunities are growing about twice as fast as they were from March 2024 to October 2025. Progress is literally accelerating right now, and you can see it in the models. It’s evident in the costs. Anthropic’s recurring revenue still grew 10% month over month in October.
Look at the prices of some of these stocks, like Nvidia, Google, Broadcom, and Oracle. They’re all down year over year, even before Claude Opus 4.5 came out, which really gave all this a boost. To me, it’s just crazy. Since Claude 4.5 was released last November, Nvidia has grown by 80% this year on a base of $200 billion. Hyperscaler capital expenditures have increased by 75% and are forecast to increase again next year. The rental price for an H100 chip in 2022 increased by 50%. Spot computing prices have increased significantly, and yet these companies are down year over year.
I truly believe all this short-term stuff is a distraction, and you should just own assets that benefit from it. There are many ways to express that, and you should probably have a mixture of them. There are assets that benefit from higher capital expenditures, like Nvidia and semiconductor manufacturers. There are assets that benefit if capital spending falls or if we don’t get as much new capacity next year as the hyperscalers may be expecting. And then there’s everyone else: every other company that benefits from intelligence. At the end of the day, the biggest beneficiaries are consumers and businesses that benefit from the acceleration of intelligence.
5. AI Stocks at COVID Low Valuations
So, yeah, I’m buying. I had a lot of cash. I also sold some crypto, as I said, a month or so ago, and now I’m mostly putting it into indices because I already have a lot of individual names. Over a two- to three-year time horizon, or even 12 months to three years, it’s hard for me to be pessimistic about what’s happening in AI. I truly believe there’s never been anything as significant as what’s happening right now, and I think we’re in a singularity. I wouldn’t sell. I just try to stay optimistic about the long-term trend and not let myself get crazy about short-term things.
When you look at this latest sell-off, it’s striking how cheap valuations have become, depending on who you are and where you sit, isn’t it? If you look at the Nasdaq, for example, we’re now at the level of the forward P/E ratio. We’ve fallen back to pre-COVID levels, haven’t we? These are the levels from March 2020. Look at Nvidia: it has a forward P/E of 12.
Yes, that’s right: 12. That’s not at all like what we saw during past bubbles. Ceteris mentioned a graph of what it looked like in 2001, remember? This actually takes us back to the time of Netscape’s release and what that bubble looked like after 1998 and the real crash. This is where we are now, after ChatGPT came out.
Comparisons, especially regarding multiples, have never made sense to me. The only thing I’ll say about using forward rates and P/E estimates is that they’re based on analyst expectations. These are the moments when it’s worth having your own process and figuring out whether you agree with the consensus, because often these estimates can get cheaper and keep getting cheaper. EPS estimates, these forward-looking fundamental estimates from analysts, usually lag far behind price movements. It’s the same idea that price shapes the narrative: price dictates forecasts.
Given the state of the indices and historical trends, I think we may still fall a little, but in 12 to 24 months the situation will be much better. Assets will be more expensive and indices will be higher. Valuation tools like P/E or price-to-sales ratios are bad for timing, but great for long-term predictions. Forecast estimates have been consistently underestimated over the past three years. For most of them, the estimates were consistently far below reality.
I think Nvidia’s stock price already takes into account that 2027 will be the peak year, and that profits will decline every year after that.
Yes, I’ve also seen a chart for the Magnificent Seven where capital expenditures seem to yield almost no return. I don’t think anyone is optimistic enough about AI.
You can also just hold the S&P 500 if you want. I prepared this with Claude to show that when these companies invest in capital expenditures—which is about $800 billion this year—it all gets amortized over the years. That’s the main argument of the bears, isn’t it? That depreciation will be very high and that the capital costs won’t pay off.
I tried to figure out what exactly the market is pricing in compared to current performance. The gray bar shows how much each company needs to grow its EBITDA just to stay in place—that is, to maintain profits at the current level, taking into account the increase in depreciation. For Microsoft and Google, it’s about 7% to 8%; for Amazon, it’s 12%; for Oracle, it’s 16%.
The purple bar shows the level of growth embedded in the current stock price. This requires some assumptions about the ultimate value, but it’s possible to deduce from the price the growth rate needed for the stock to be profitable. The difference between the purple and gray bars is how much the market believes in the return on investment in AI. As you can see, it’s minuscule: 5.6% for Microsoft and 3.3% for Alphabet.
If you look at their current growth, it’s much higher. Current EBITDA growth is much higher than what the market expects in the future. I think they’ll continue to grow quite rapidly and surprise people. So, I don’t know.
For me, all the major AI assets look pretty good. I didn’t really delve into semiconductors. I think this is a riskier area to play in, other than Nvidia and TSMC. TSMC, by the way, had an incredible month. That seems like a 60% increase in revenue.
6. Meta's AI Bet Is Priced at Zero
I think it looks pretty good. And, as I said, you can own the index, because I think the benefits of cheap intelligence will be reflected in every S&P business as AI starts to permeate everywhere and people start to use it.
Are there any specific hyperscalers that interest you more? Meta is a good example.
Yes. They often get lumped together, don’t they? It’s about access to computing power. We’ve been talking for the last few weeks about there being a premium for working compute—that is, installed capacity, not potential future capacity.
But look at a company like Meta. They use and integrate these capabilities, so advertising revenue, for example, is growing. The price of advertising itself on Meta has increased. They’ve become much more effective at targeted advertising, which is obviously a key part of their existing business. They use these capabilities to improve the fundamentals of the business, which sometimes seem like, or are considered by people to be, an add-on, even though they’re core to the company itself.
I think Meta is a great example of how they’re using this to significantly improve the business fundamentals they’ve been working on for the last decade.
Yes, 100%. Meta is my biggest individual stock, but the position isn’t huge because I just don’t have the time in the public markets to have a firm belief in any of this. So I want to keep some reserve. It’s not huge, at least compared with how focused I was in crypto before, and on the things I really understand deeply.
Meta is really interesting because, in my opinion, its traditional business has been the least affected by AI. In fact, I think it’s even benefited from it for the reasons you mentioned. And their AI business is now valued at almost zero. I did a basic sum-of-the-parts analysis for each of the companies. Meta is a business where the legacy component definitely makes up the majority of the company’s value.
It’s clear that Meta also doesn’t have the cloud business they were talking about creating. If they did, I think it would be extremely positive for the stock. They have an extremely large amount of computing power. So you’re essentially paying nothing for the option that they’ll be able to implement AI with Muse. I think they’re incredibly well positioned to do that, with the WhatsApp, Facebook, and Instagram user bases to adopt it across businesses and consumers.
I’m also optimistic about their team. For example, I think Alexander Wang—I’ve watched a bunch of his interviews—and Mark seem to have really great chemistry. Mark seems more human when you talk to him. I just think they have a pretty good chance of succeeding in AI.
Overall, the way I see these Magnificent Seven companies is that you have a traditional business that’s being impacted by AI to varying degrees. You have a cloud business, which essentially sells a “token factory.” Then you have the sale of intelligence, which is a very low figure for most of them. Google claims it’s tens of billions of dollars a year, but I have a hard time seeing that anywhere else.
7. How the Anthropic IPO Could Trade
And then I think the fourth part is just a bunch of options for different things. Google has Waymo and processors; Nvidia has Trainium [?]. For Meta, I think it’s the company that’s least valued in terms of whether any of its AI bets can pay off. So I find it very, very interesting.
Another big piece of news from the AI world: we’re finally getting clarity on what Anthropic’s IPO will look like. They’re trying to raise somewhere around $69 billion or $70 billion—approximately $70 billion—and the price will be set by the end of next week. We have some pre-IPO indicators that let you trade Anthropic shares in advance. That adds about a 20–25% premium to the price, depending on the situation. It’s definitely been volatile, but it’s interesting to hear your thoughts. Do you think there’ll be hype? There’s been talk that demand among institutional investors for the Anthropic IPO is falling sharply. Even with the recent volatility and sharp swings, especially in AI and tech stocks, it seems they’ll continue to move forward. How do you think this will all unfold?
I don’t have a clear opinion on the direction, but strategically it’s in their best interest to be first. If the bidding fails, it’s better to enter the market before OpenAI. If everything goes well, the shares will simply rise. You may not raise as much money, but you can always do it again later if the markets are on the rise. So I think it makes perfect sense to enter the market now. Although I personally don’t have a firm opinion on where things will go next.
Yes, I agree with you on that. From their perspective, it makes sense. Why take the risk and wait another six months? You don’t know what the situation will look like in six months, especially if OpenAI enters the market first and shows poor results.
Many are asking, “Why don’t they just postpone the exit?” And I say, the IPO isn’t for you. It’s not for retail investors. It’s for the company itself. So, in my opinion, the most sensible thing to do is complete it by the end of the year, as they plan.
What exactly? It seems like a tweet just popped up. They’re launching an IPO roadshow, and the share price is $70–78. I don’t know how the bidding will go. They don’t have what SpaceX had, right? There’s no automatic inclusion in indexes, at least I think so. There’s no built-in demand base that would immediately absorb part of the supply.
I really don’t know how it’s going to go. The market is in bad shape right now, so it’s hard to imagine that the arrival of this company will make things much better. It’s just another name, yet another thing to think about. This is similar to what Paul Tudor Jones was talking about regarding the situation through the end of the year and, at least, the first half of 2027. There are too many issues in the market and too much supply.
This has been a problem in crypto for a long time, right? At the moment, there’s simply too much supply of assets to buy. It’s hard to expect everything to show good results. I don’t know. I’m not going to buy this. I have no desire to buy it at the IPO or play any pre-market or pre-IPO games. I’ll just wait.
We’ll see. I don’t have a clear opinion on Anthropic. If things go badly, the market will be in a negative mood for a while. The fundamentals seem normal, judging by the documents they submitted in advance. This is a difficult market for IPOs. The valuations are excellent, but it’s difficult to absorb all of that right now because of the recent turmoil and the year-end revaluation of assets.
I think the market reaction will be very revealing, because everyone has been waiting for their release for a long time to see real demand. One risk you can build a narrative around is that Anthropic is probably going to get junk bond status, right? Its inability to enter the investment-grade bond market could be a hindrance, unlike the hyperscalers, who have rushed to the market with large bond sales over the past four to six weeks because of rising credit risks.
The lending market in 2027 won’t be as favorable as it was over the last six months, so companies are rushing to raise capital before that market is exhausted. In my opinion, the reaction to Anthropic in the first few weeks will indicate whether we’re in for another downturn in the broader market.
If we can handle this, we’ve already gotten through SpaceX, right? Now Anthropic. Then the unlocking of SpaceX shares. There’s still a lot to come. The next year or two will be difficult, and this is part of that “wall of fear” we’ve been climbing all this time. But if we can overcome it over time, I think everything will be fine.
I actually think Anthropic’s valuation of around $1 trillion isn’t that bad. I’m not buying now; I’ll buy when the indices do. But I think that’s acceptable, right? There was a period, even half a year, when everyone was wondering whether this company would go public at a valuation of $1.5–2 trillion. I think this is a more manageable number.
8. Will Anthropic Wipe Out SPV Buyers?
There’s definitely a lot of demand, and many will want a share of it. It’ll be interesting to see how the shares trade after going public. But the trading during the first week doesn’t say much. Even with SpaceX, it was like a game of IPO mechanics. I wonder if my share in the SPV was real or if I was scammed. I’ll find out in the next six months. It’ll be interesting.
I don’t know if anyone has said this before, but someone would have to be crazy enough to rug all the SPV investors. Usually, those who sold shares in the SPV violated the shareholder agreement. I didn’t see the deal itself, although I tried, but Anthropic even made a statement about it: selling shares, derivatives, or anything like that is forbidden. If you do, they have the right to void the deal and cancel the allocations.
Many shareholder agreements contain similar clauses. So many SPVs were created, especially during the $150 billion and $330 billion rounds, that the total amount in SPVs is probably in the billions of dollars. If Anthropic just finds those who broke the rules and doesn’t let them exercise their shares, it’ll be complete chaos.
A chain reaction of lawsuits will begin. Of course, these people will sue Anthropic, and those who bought shares in the SPV from them will sue the sellers themselves.
And so on, layer by layer, with all these SPVs. The second-level SPVs that bought shares in the first-level SPVs will sue them, and there will be complete chaos.
But for Anthropic itself, that’s probably normal, right? You’ve significantly reduced the float because these lawsuits can drag on for years. That pushes the price up. If you’re really in EA and optimizing to create as much value as possible for the public good, or whatever you’re doing there, I think it’s quite rational to do this and fool all the SPV participants, buyers and sellers.
Have companies done this before? Have they thrown away the SPVs after an IPO?
Yes, undoubtedly. I mean, obviously, Anduril did this quite loudly. This is a relatively new phenomenon, though; it’s only recently become a problem. Anduril did it loudly, but that was a different matter because national security was involved. First, there was fraud, and second, there were national security issues, since the defense company wasn’t owned by US citizens. Anthropic could someday become a matter of national defense, too.
Yes. I just think it’s pretty rational from their perspective, and they said they were going to do it, right? They made a statement about this. If they do, it’ll be a complete mess. I think Anthropic is probably one of the companies with the most secondary interests through SPVs. If you manage to write off all those profits and losses for everyone who bought through an SPV and wants to sell, that’s a very optimistic signal for them.
So you get the opportunity to just write it off, take the shares back if they… I guess you don’t know, because you haven’t seen the deal.
That’s a pretty cool deal for those who sold the stock, right? You get this free credit up front. Then the stock goes up a lot, and now you just need to pay back the money you borrowed, while keeping the stock, which is worth 10 times what you sold it for.
Yes, for them it is, if that’s really the case. But I think what will happen is they’ll just take their shares.
All of this will be spelled out in the shareholder agreement, but I imagine their deal is pretty aggressive and airtight.
Yes, this sounds like a huge headache. We’ll see. I’ll let you know if I… I’m betting on the lawyers. Everyone says AI will take away lawyers’ jobs. Just not with this nonsense that’s going on.
9. The Mystery 10:30 Seller
Yes, yes. Low free float, high FDV, as we often say. This didn’t originate in crypto. It’s a classic IPO scheme. These are really huge FDVs. It’s not like EigenLayer at $5 billion. We’re talking about $1.5 trillion, you know?
It looks like the salesman came back again at 10:30. Millions in revenue, but yes. What kind of salesman is this at 10:30? He keeps coming back. Actually, who are these people? I’m curious. Who’s selling? It’s always at the same time. Why do they choose this time? Jason, do you have any thoughts on this? I’m really curious.
Dude, I just noticed it happening every day, literally minute by minute. As soon as 10:30 hits, I look up and everything starts to fall. Maybe it’s some billionaire who doesn’t like waking up early and sits down at the computer at 10:30 instead of 9:30.
Apparently, the spreads are too wide right after the market opens, so they wait an hour for the order book to fill up, then come in and start selling.
That’s right, that’s right. I don’t know who it is. I don’t know why people choose a certain time of day, but if you’re constantly in front of a monitor, you notice things like this and can trade based on them to some extent. It seems so obvious: “I’ll sell at 10:30, I’ll buy at 11.” If you’d been doing this for the last couple of weeks, you would’ve made a good amount of money.
But would you get ahead of him by opening a short? What if one day he doesn’t show up and you’re left with nothing?
10. ZEC in the Wall Street Journal
I don’t know who it is or why. It’s just that almost every day—80% of the time—this guy shows up in the order book and pressures the market, wiping out everyone who opened a long position at the open.
I see we have an open Zcash chart. Grayscale bought a whole page in The Wall Street Journal. I don’t even know what to think about this. I won’t say I don’t like it. I’m probably neutral. I think you’re probably thinking, “Oh man, this is a classic sell signal,” right? But they spent, I don’t know, $3 million marketing this project.
And yes, there are a lot of nuances with Zcash. Some are pessimistic about Zcash by the end of the year because of all the unrealized profit, which I think is fair. Sometimes you can try to play on unrealized profit, as far as I’m concerned. You see how it works both ways: either the price keeps going up and you think, “Oh no, I missed it.”
That brings us back to what we were talking about at the beginning. The situation is very unclear through the end of the year. I think it was really disappointing. In June, July, and August, a lot of good on-chain metrics appeared, everything was improving, and you saw people coming back. But it was as fleeting as with Robinhood and even some things on Solana. If you look at all these graphs, they look the same. They’re very discouraging: no rebound, just a straight line down, then consolidation, then a straight line down again.
11. Why Stocks On-Chain Stalled
It’s a little disappointing. What did you expect? I think the difference was that “this time it’s different” because we’re moving stocks to the blockchain, and they’re showing good results and creating new opportunities. The problem is that it was all based on commissions, which, in fact, provided the income in the form of on-chain fees.
So there’s a discrepancy between something bullish and the stability of the mechanism. That’s actually part of the thesis about options, particularly covered-call option vaults and the possibility of earning income there. We discussed this with Jake. It’s not new, but I don’t think it’s discussed enough, because these vaults aren’t operating at scale yet. I’m waiting for them to appear.
The point is that you can earn profits on non-dividend-paying assets by placing them in on-chain vaults where operators sell call options on those assets. Usually you don’t get called away and you earn income, but sometimes it happens and you’re forced to sell the asset at a slightly higher price. That determines the overall return profile.
I think this is a strong argument for moving shares to the blockchain, though there are nuances. First, you can sell call options on individual stocks, whereas brokerage platforms usually limit you to larger volumes. You also need to consider how to manage this process. It’s a completely different matter: you need to sell options at a profitable price, analyzing the volatility profile and its curve to maximize income.
Another aspect is tokenization and using the “looping” mechanism. Looping has always been one of crypto’s basic capabilities: you can tokenize an asset, make it composable, and use it as collateral. So I believe you can earn a better return this way than even by working effectively through a broker, because using an asset as collateral is a unique property of crypto.
I think this will be a strong incentive to move shares to the blockchain, although the previous iteration was based on a 3% AMM fee that was used to buy back shares or pay income to owners.
12. Can Elysium Avoid Robinhood's Fate?
Yes. This can work and be flexible, but of course it also responds reflexively to market drops when trading volumes decrease.
I think the main problem with on-chain solutions right now is that, as we see with the Elysium example, you can create unique things like complex options strategies because you have direct access to HyperCore. That gives you access to all these Hyperliquid order books, stocks, and so on, which is really cool.
But on day one, it already seems like there are a million launchpads, right? That’s because it’s too easy and doesn’t require any effort. If you make a regular launchpad, it doesn’t take much effort; you can quickly make a lot of money and then just wash it off. Look at the Ponds chart, and all those Robinhood-style charts. This is essentially a transfer of shares to the network. We haven’t created anything truly valuable yet. We made memecoins with them, right?
No one has created anything truly useful because it takes time, effort, and intelligence to come up with something interesting. Instead, you can just get involved in a launchpad war. For Elysium, yes, there will be launchpads, because every network needs them, but that’s not why their L2 is interesting. Their L2 is interesting because of what you can do with Hyperliquid. If people go there just to trade on launchpads and all that, then it’s just Robinhood 2.0.
I think they have a good chance. A lot of interesting developers are coming to them. We published a very good report on Elysium from AD yesterday. If you trade on the order book on Elysium or Hyperliquid, you get much more detailed data, right? You can see how the order book is actually formed.
You can do a lot more with a network like this, where you have access to all this liquidity. You can think about pre-IPO, options, and so on. There’s so much that can be done here. Hyperliquid is good at perpetuals, but its spot volume is actually pretty weak. Now you can create a spot platform for many things. We’ll see how it goes. This is just the first day of their network.
But watching all these launchpads, it’s like, “Just not again.” This is simply the easiest way for many people.
Well, yes. Sometimes I think we talk about it a little abstractly, like online activity. Regarding Elysium, what are you looking at to keep your optimistic forecast in place? What could refute it?
If this becomes the next blockchain for meme coins, the price dynamics will probably be similar to what happened with the Robinhood network, as you just described. If it looks like Robinhood and smells like Robinhood, then it’s probably a lot like the Robinhood network.
That’s right. What exactly are you looking for to confirm that things are different this time?
What you said: seeing options vaults develop using Hyperliquid in the backend would be a very positive signal. Other things that use HyperCore in a way that can’t be implemented on another platform. Launchpads can be made anywhere, so they’re not what makes the project unique.
This should really expand Hyperliquid into a full-fledged financial ecosystem, beyond what they could do with HyperEVM. You want to see how this actually extends the capabilities of HyperCore, because Hyperliquid isn’t about meme coins. You need to invest a lot of money to deploy a project there, which is why there are no meme coins.
The ceiling for meme coins here also isn’t as high as for something like Robinhood, where the prospect of getting listed attracts users. Those are very attractive factors for people who want to speculate that way.
Well, I guess the Hyperliquid spot market lets a meme coin buy a spot ticker or something similar at auction and then get into HYPE. But yes, I agree that the ceiling isn’t nearly as high.
As I said, you shouldn’t dwell on this too much. I’ll keep an eye out over the next 1–3 months after the official mainnet launch for signs of what you’d expect to see in Elysium, given its prime location next to HyperCore. Do you see people using it? If teams don’t build things that use it, or those things don’t gain popularity, then you could go back to seeing it as just another LST and regular L2-type onchain nonsense.
With that in mind, I don’t think 1–3 months is the right timeframe. People can build something over the next year, and we might see a really successful application in about 6 months. But you want to see people moving in that direction and stopping these launchpad games. The launchpads themselves are normal. You just want something more. Do we need another launchpad token? No, definitely not.
I think there are some interesting things. For example, Ascend doesn’t have its own token. It simply uses Kinetic as its—
Yes, this chain is unique, as is the number of components that contribute to the accumulation of value. But there are already a few others, such as Mantra, which didn’t have—or perhaps now has—its own token. I don’t know. I just stay away from not trying to use L2 in the way it’s best suited for.
13. Is ETH's Store of Value a Hail Mary?
Of course. But yeah, check out this report, it’s really, really good. You’ll learn why everyone is asking this question: why do we need another L2, and you’ll understand why this is actually a very unique product and not just an attempt to play the same game again, right? We’ve been discussing this topic of a store of value, cryptocurrencies, and all that. I don’t know if you saw Vitalik say that ETH is a store of value. Now everyone in Ethereum is trying to promote it that way. I think it reflects the times we live in.
I’m definitely early, aren’t I?
I think so. I don’t know why it should be any different. It seems like everyone in Ethereum is actively trying to meme this right now. It’s strange that Vitalik is coming to all these realizations after it’s been obvious for so long, and many people consider that optimistic. But you can also see it as a last desperate step, a last hope. It’s like all we have left.
So I’m very neutral about it. There are arguments, honestly, about quantum technology in Bitcoin and everything else. Ethereum still remains a very diverse ecosystem. There’s a chance this narrative for ETH could become more popular next year, but there’s another side to it. ETH is either very undervalued or overvalued, and it will go one way or the other. I think the time to choose one of those directions is running out, and that will become quite clear over the next year.
I don’t have a clear opinion on this right now. I find it hard to imagine that the capital needed to move the market would decide this is a great hedge against the quantum threat to Bitcoin, since the market seems to have chosen other assets for that.
Yes, the probability was much higher without Zcash and similar assets.
Yes. I think things will continue to move in the same direction, and ETH will trade the same way as the rest of the market. Whether it outperforms or lags behind the market, I think that’s exactly what you’re talking about.
Yes, but I think there’s a pretty big “if.” If ETH gets to $2,500 and BTC goes back to, you know, $85,000–$90,000, I don’t see ETH at $1,000.
No, but it may just keep vacillating.
I just think it won’t be able to waver for too long. The real direction should be determined over the next few years. It can’t just hang around without knowing where it’s going. People will eventually either join or move on.
It’s still pretty well capitalized, right? Assets worth $250 billion.
Yes. Madness.
So Tom Lee has 5% of this bad boy. Maybe Tom Lee can make it work.
José, has anything changed in your forecast that might make you less optimistic about the singularity, other than what we discussed at our in-person meeting?
I think there are many factors that would make me more skeptical about individual stocks, semiconductors, or large technology companies. But I use AI every day, almost all day. Seeing how high-quality it is and how it’s penetrating the economy, it’s hard for me to be pessimistic about this trend as a whole.
I’ll just check in with you every now and then, for comparison.
I’m not a short-term trader and spend most of my time in private markets. In public assets, I just want to hold what I’ve believed in for many years. I think this is the most important trend of all for me. So I’m just trying—
Is there anything that would bring you back to crypto? Not that you’ve left it, but something that would get you back into it. Are you keeping an eye on crypto P/E ratios that are starting to look attractive?
Honestly, not really. I still hold the 4 or 5 assets I’ve held for quite some time. I sold a little bit, and I even reduced my stake in one of them a couple of months ago, I think. I sold pretty well on HYPE, actually, because I dumped when the news was too optimistic.
Was that you? Devil?
But I mostly keep everything else. I don’t know what would make me actively invest in crypto again. Perhaps something else like Hyperliquid, preferably in the field of AI. Something with a real market product and returns most of the fees to the token—that would be cool.
Yes, although with AI that’s hardly possible.
It is possible.
No. Returning most of the fees to the token—I think AI in general requires a lot of capital investment, and you’ll be competing with players that have huge costs. I don’t think that dynamic is possible.
I think it’s possible, but it’s just not—
In what form?
Inference, probably. Inference is the thing that’s most—
14. MegaETH Pivots to AI Inference
I don’t think it’s happening right now, but I think it’s an area where theoretically it could happen. You could have onchain inference, which—
MegaETH inference just reduces the margin. Okay, I just saw this announcement: MegaETH is rebranding to Mega and allocating $30 million to create a marketplace for AI inference providers based on MegaETH. There you have it, José. Maybe that’s your conclusion. Maybe this is your strategy.
Well, you see. The skills that make a company a good inference provider are very similar to the optimizations that were done at MegaETH.
Yes, that’s true. Tom had a good point about it.
That seems like a great turnaround for them. I think the reversals in crypto today are good. I don’t want to get stuck in one place. I think something didn’t work out for MegaETH with what they were planning, so this is a good move. In my opinion, this is a smart turn.
Yes. We’ll probably see many more like this. Guys, how do you position yourselves? I think we’ve discussed this before, so it’s probably very similar. You’re still mostly in crypto, or at least you think so.
I’m pretty diversified into a bunch of different things now.
We’re going to glory with HYPE, brother. You think I’m joking. I still think HYPE will be okay, though it might fall. There are just so few truly good businesses in crypto, and perhaps The Graph is one of those that looks better than the others.
Really? I didn’t know you were being ironic.
No, I was just looking at it. Apparently it’s pretty cheap now, although probably not very cheap. The fully diluted capitalization is still $500 million.
This is madness.
Yes. Okay, guys, I think we’re out of—
Yes, we’re out of ammo.
Well, as always, it was nice to chat.
Perfect.
We’ll be back, and hopefully nothing too crazy happens right after the video comes out, because that’s been our excuse lately.
Yes, it may turn out that Jensen worked for the CCP for a while.
Yes, imagine. Zero. Maybe this would make developers look more optimistically at the development of data centers.
He said, “Okay, we need to catch up.”
Yes. You assume they are at least somewhat rational.
Yes.
Okay, guys. See you next time.
Bye, guys.