Why OpenAI Is The Next Amazon - GoodAlexander
GoodAlexander expects SpaceX’s IPO to trade well while draining liquidity from almost every adjacent speculative asset. The next year may bring roughly $200 billion of issuance across three coveted companies, versus $130–140 billion raised by crypto VC over five years. SpaceX could represent a roughly $25 billion slug, with about 30% allocated to retail versus the usual 5–10%. His base case: “SpaceX itself trades well, every other speculative asset trades down.”
The SpaceX valuation rests less on current growth than on Elon Musk’s ability to make investors “dream the dream” of AI data centers in orbit. Musk has filed to launch one million satellites and, in GoodAlexander’s account, proposes putting compute equivalent to 20% of US power supply in space, where cooling and local resistance are less constraining. The thesis may take years to falsify, which is precisely why “the sun is really big” works as an equity narrative.
His preferred SpaceX trade is short legacy telecom, not short SpaceX or Tesla. Satellite-heavy telecom baskets had risen roughly 120%, while some slow growers were up around 180%; if Musk’s architecture works, incumbents lose leverage across connectivity and compute. He refuses to short Tesla absent Chinese EV access to the US: “I try not to short things that could invent the cure for cancer.”
OpenAI is GoodAlexander’s favored AI IPO because he sees the same monetization blind spot that investors once had with Amazon. Amazon’s retail operation was once marked at zero, yet 300 million active shoppers supported a $69 billion-a-year advertising run rate by Q1 2026; OpenAI, he says, has 900 million ChatGPT DAUs. A six-week ad pilot reportedly generated $100 million against a $2.4 billion revenue target, making ads the overlooked option embedded in its consumer distribution.
Anthropic’s Fable model may be technically exceptional while its reported economics remain less attractive than they look. GoodAlexander says Anthropic highlights a non-GAAP operating-profit measure despite remaining deeply free-cash-flow negative, and treats model-training capex as though it were temporary even though stopping investment would destroy future competitiveness. Thread Guy says Fable finds many bugs in GPT-5.5 code that Opus 4.8 missed; GoodAlexander agrees its diagnoses are generally right.
The deeper enterprise-AI risk is that model costs rise while generated software cannibalizes the market it is meant to serve. Two Fable agents may cost around $100 an hour—effectively another full-time engineer—while customers such as Uber spend heavily without producing an obviously better app. Consumer commerce looks greener because models understand intent, while frontier cryptography, biology and mathematics remain the strongest cases where AI can enlarge rather than replace an end market.
The political-market overlay creates a liquidity warning and a possible threat to Palantir. GoodAlexander speculates that official intervention may have supported S&P futures, while stablecoin supply remains stalled near $180 billion. He treats the stablecoin stagnation as a warning for crypto and non-AI risk assets, and sees Palantir as exposed Trump beta trading poorly. Anthropic’s “forward-deployed” work with Goldman Sachs also tests Palantir’s core claim that LLMs require its ontology: “What if Karp’s wrong?”
1. SpaceX is a liquidity event before it is a fundamental bet
GoodAlexander’s starting point remains market plumbing: crypto VC raised roughly $130–140 billion over five years, while the coming year could contain about $200 billion of equity issuance from SpaceX, OpenAI and Anthropic. “The market cannot go up” cleanly through that much supply from three of the world’s hottest companies.
SpaceX is unusually consequential because roughly 30% of the offering may go to retail, versus a normal 5–10%. GoodAlexander describes roughly a $25 billion slug coming out, capable of “suck[ing] the liquidity out of every speculative asset,” irrespective of whether the company ultimately executes.
The setup was already overheated: GoodAlexander points to an S&P telecom ETF containing satellite exposure that rose approximately 120% in a year. As a former telecom analyst, he regards these as historically dull companies whose operating performance did not justify a moonshot.
Thread Guy’s pushback—worth keeping—is that recent Broadcom and Oracle results showed “beating” is insufficient when expectations are extreme. GoodAlexander agrees that a SpaceX decline would be “horrendously bearish,” but thinks every interested bank, issuer and AI company is incentivized to ensure the flagship offering succeeds.
2. Orbital compute is the narrative underwriting the multiple
SpaceX may be growing around 30%—perhaps nearer 15% before recent revenue deals—while Anthropic and OpenAI are described as growing 500–800% and roughly 500%, respectively, at perhaps half SpaceX’s valuation. That mismatch forces the question GoodAlexander asked after watching Musk with Jamie Dimon: “What do you got, Elon? Like, what’s the pump?”
The answer is the sun. Musk has filed for one million satellites and, in GoodAlexander’s retelling, wants to launch compute equivalent to 20% of US power supply into orbit: a dense pack of satellites “cooled by space,” avoiding terrestrial coolant logistics, power-price inflation and communities that do not want data centers nearby.
GoodAlexander puts the proposed prize at $300–600 billion annually and sees the Anthropic and Colossus deals as foreshadowing the pitch: Musk built terrestrial AI data centers, rented them to major users and now says he can repeat the model in space. One unnamed top investor captured the distinction: “I don’t know about SpaceX as a company, but as a stock, it’s one hell of a stock.”
The story’s advantage is its long verification window. GoodAlexander says Musk can spend nine months “running around saying how big the sun is,” while Thread Guy argues that failed reusable-rocketry economics or inadequate orbital structural integrity would be the kinds of evidence that could break the story. Both see near-term falsification as difficult.
3. The cleaner trade is against legacy telecom
GoodAlexander is short telecom because no listed “space beta” can compete with Musk’s version of the story. If SpaceX can beam connectivity and AI compute from orbit, incumbents such as T-Mobile face a supplier with “everyone by the balls”; slow revenue growth and shareholder-destructive management leave little fundamental cushion.
Thread Guy proposed the crypto analogy of a developer launching a second token: short Tesla when SpaceX arrives. GoodAlexander initially saw the appeal, then rejected it—Tesla remains one of the only liquid public expressions of robotics and autonomous driving, while telecom offers aging management teams, roughly 6% growth and stocks that had run as much as 180%.
His short-selling scar is Plug Power. After the CEO left, he tripled a short he believed must finally go to zero; the market welcomed the departure and the stock rose roughly 600% in a straight line. The lesson from his “five whys” review: “If this has been a scam for 20 years, what stops it from being a scam for 23 years?”
4. OpenAI has Amazon’s once-mispriced advertising option
The market frames Anthropic as the safer enterprise company, promising free-cash-flow positivity in 2027, while OpenAI is the frightening cash burner that needs an IPO before it runs out of money. GoodAlexander prefers OpenAI precisely because that consensus discounts the strategic value of its consumer audience.
His analogy comes from trading Amazon when investors owned it for AWS and assigned the retail business no value. Amazon later turned 300 million active shoppers into a $69 billion annual advertising run rate as of Q1 2026; OpenAI, he says, already reaches 900 million ChatGPT DAUs. “It’s really hard to get 900 million users.”
A six-week OpenAI advertising pilot reportedly printed $100 million amid heavy oversubscription, while the stated target is only $2.4 billion. GoodAlexander also praises OpenAI’s GPT-5.5 Instant search experience: unlike traditional search, it can consume ten pages simultaneously and synthesize them without forcing the user through ten sites and their ads.
The unresolved risk is trust. Anthropic can promise an ad-free product, while OpenAI must show that recommendations remain useful rather than becoming disguised shilling; GoodAlexander’s favorable specimen is Codex automatically choosing a Cloudflare tunnel for security—a commercially valuable recommendation that still feels justified by the task.
5. Anthropic’s product is stronger than its accounting story
GoodAlexander calls Anthropic’s profitability presentation “misleading,” because its press releases emphasize a non-GAAP operating-profit measure while the company remains deeply free-cash-flow negative. Model-training capex is not analogous to optional Coca-Cola capex: stop spending, and the next model upgrade can destroy today’s unit economics.
He therefore treats that investment as structural R&D, not a one-off adjustment. Anthropic still deserves credit for structurally better margins than OpenAI because most of its revenue is enterprise, but its claim to current profitability should not be confused with actual cash generation.
On the product, GoodAlexander reverses his initial annoyance with Fable. It refused a requested wallet-security analysis, yet once prompted correctly it proactively found issues such as a possible XSS exploit. Thread Guy says Fable found many more bugs in GPT-5.5 code than Opus 4.8 did, and GoodAlexander says GPT-5.5 generally agreed with its diagnoses.
The cost is severe: a typical engineer running two agents may spend about $100 an hour, or several hundred thousand dollars annually—roughly the price of another engineer. GoodAlexander’s honest tension is that Fable is “the alpha model” for deployment even though the economics invite cheaper Chinese replication and tighter corporate usage caps.
6. Enterprise AI may cannibalize itself while frontier science expands demand
GoodAlexander’s enterprise objection is causal: increasingly expensive models make software easier to reproduce, which lowers demand for packaged software and games. If everyone can customize an application, the models are “cannibaliz[ing] their own end market” even as vendors celebrate how much code AI generated.
Thread Guy’s Uber example becomes the reality check. The company can spend an enormous sum on model credits, and Salesforce can say AI wrote 45% of its app, but GoodAlexander asks whether either product is detectably better than two years ago: “I can’t tell, man.”
The better case is work that previously required scarce specialists. Orchard, Zcash’s complex cryptography system released in 2022, allegedly carried bugs for years because so few people could understand it; GoodAlexander says a frontier model can inspect the implementation, identify the exact flaw and potentially improve the underlying algorithm.
That logic extends to biotechnology and mathematics, where AI might enlarge the discovery frontier. It does not make public biotech an easy trade: Ginkgo Bioworks, ticker DNA, is his cautionary example—down roughly 99%—and the best private longevity assets are often bought early by ultra-rich investors who value health more than returns.
7. Political support may be real without being easy to monetize
GoodAlexander labels his own claim controversial: he speculates that US equity and oil futures have seen intervention, possibly involving Scott Bessent. His evidence is behavioral—during geopolitical escalation, unusually large buyers repeatedly appeared around a 0.5% S&P decline, preventing the gaps and cascades familiar from Ukraine in 2022 or COVID.
His inference remains explicitly a conspiracy theory, not proven fact: Bessent trained under Stanley Druckenmiller, Trump is intensely focused on the market, and the price action looked unlike prior risk-off episodes. GoodAlexander expects later administrations may investigate “what the hell is going on.”
Even obvious political information is not automatically profitable. He cites Hegseth buying defense stocks that subsequently collapsed, while Thread Guy compares the game to chasing CZ-endorsed memecoins: each successive endorsement may create a weaker, shorter-lived pump.
The harder warning is stablecoin supply, stuck near $180 billion when policy advocates expected hundreds of billions. Thread Guy recalls a possible $10 trillion Scott Bessent plan, but GoodAlexander’s stated concern is that USDT and USDC have flatlined and the Clarity Act has not passed. If supply does not resume growth, he sees that as a “canary in the coal mine” for crypto and non-AI risk assets.
8. Anthropic could attack Palantir’s ontology moat
Anthropic’s Goldman Sachs language caught GoodAlexander’s attention because it describes placing “forward-deployed” engineers inside banks—Palantir’s own terminology. After Alex Karp publicly attacked Dario Amodei’s employment warnings, GoodAlexander sees motivation for Anthropic to challenge a company whose deployments it may already help power.
Palantir’s ontology originated in a real intelligence problem: agencies could not share information, and even shared descriptions—“a guy in a white outfit with a Jeep in Afghanistan”—matched too many people. Its software attached identities and relationships to data, then transplanted that structured layer into commercial AI deployments.
Karp’s current claim is that LLMs do not work reliably out of the box; they need Palantir’s ontology. GoodAlexander’s open question is whether stronger models, vector databases and automated tagging make that layer unnecessary: “What if you just need a Snowflake database” that Fable or Mythos can organize in a day? If so, he imagines an 80% stock decline.
Having worked at Palantir for two years, GoodAlexander still calls Karp a fantastic “CEO whisperer”—interesting enough to captivate executives afraid of being boring, and skilled at combining “spike talents” who excel at one thing. That same combative charisma becomes dangerous if it starts a war with the company’s most important AI supplier.
9. SpaceX needs the dream; OpenAI needs proof that users like the ads
Into the listing, GoodAlexander watches whether retail can understand, articulate and amplify the orbital-compute story. At roughly 90 times sales, the offering needs Musk, Jamie Dimon and sell-side research to sustain a credible dream; Thread Guy says evidence against reusable rockets or space data centers would be the clearest way to break it.
His base case is near-term strength but poor long-term value: execution risk is already priced in, current growth is too slow, and performance becomes path-dependent on xAI. If subsequent Composer models combine state-of-the-art quality with speed and privileged compute, he would hold; if xAI “is not a thing,” SpaceX could drift lower.
He remains broadly bullish on both AI labs because Fable and Codex are “some of the best tech products I’ve ever seen,” growing at perhaps 8× despite 30–40-times-sales valuations. His stronger preference remains OpenAI, contingent on repeatable ad pilots and evidence that advertising does not erode DAUs or user trust.
Google is the relative fade: it is the consensus buy-side long because TPUs, cash flow and cloud backlog look obvious, yet the stock is not trading well and Thread Guy says the backlog “looks like fraud” and Gemini is poor for coding. Thread Guy closes with an $80 billion Google equity raise—larger, he says, than Anthropic’s last round, OpenAI’s last round and the SpaceX IPO raise—as another major source of equity supply.
Full transcript
Boom. Mr. GoodAlexander, welcome back to the show. It's been a while, man. You got a new background. How have you been, dude?
I'm good. I'm at home in Omaha, in a basement, so, yeah, it's a different setting. How are you doing?
I hope you're enjoying home, man. I'm doing great, dude. You were on with David a couple of months ago. We were talking chip, which was a really fun one, and we got to get a little bit of SpaceX action on that stream, although it obviously wasn't the focus.
Now we're a day or 2 out, and a lot has happened since you came on. I think the main thing to start with is that there's been a lot of new information on the IPO circuit. Since you first came on, we didn't really know that much about SpaceX other than the fact that it was coming. We now know more information on their filing and financials. We know about this $12 billion annual revenue deal with Google, and we know about the Anthropic partnerships.
We also know OpenAI and Anthropic are coming—confirmed. I think we were just speculating back then. To start, how are you thinking about what is going to happen with the SpaceX IPO and the impact it's going to have on the market? Has your perspective changed at all since our last stream?
Yeah, I think now that we know all the numbers, my previous comment on the stream was that this is a really big liquidity event. To put it into perspective, over the last 5 years, crypto VC—which we know has generated a lot of pain for various charts—has raised about $130–140 billion. In the next year, we have about $200 billion of issuance.
The SpaceX allocation specifically has a much higher allocation to retail investors than would be typical. Normally, it's around 5–10% for retail. For the SpaceX allocation, it's about 30%. We're looking at a $25 billion slug coming out, and from a market perspective, you have to pay attention to it because it's going to suck the liquidity out of every speculative asset, regardless of how these things turn out.
I think what's interesting in markets right now is that previously we were in a kind of overheated environment. A really good example of this is the S&P Telecom ETF. It's a telecom ETF, but it has satellites in it, so it has some satellite companies in it.
Yeah.
This thing ran 120% in a year. I used to trade telcos. These are the most boring companies in the world, and you're like, there's no reason why these things should moon. We went into this with a lot of speculative excess, and I think some of that excess has been taken out of the market in the past couple of weeks.
The way I look at the raw picture is that the crypto market cannot go up, and risk assets cannot go up, with $200 billion of equity issuance in 3 of the hottest companies in the world. It's just really hard. What I think of the equity issuance is a very separate thing.
I watched the Elon Musk presentation with Jamie Dimon.
Jamie Dimon. I saw that, yeah.
You watch that interview and you're like, okay, what's actually the play? How is this thing growing at 30% a year? Before all these revenue-pump deals came out, it was maybe growing 15% a year, which is not an acceptable growth rate.
Anthropic is going at least 500–800% a year. OpenAI is likely doing 500% a year. These are coming out at roughly half the valuation of SpaceX, which is growing at one-tenth the rate. You're like, okay, what do you have, Elon? What's the pitch?
He sort of converged on the sun narrative.
He has converged on the sun. That was the focus of the Jamie Dimon interview: the sun is massive.
Yeah. The sun meme is actually funny because Nick Land is the famous accelerationist who wrote the CCRU stuff, and he was hanging out with Elon Musk. There's a picture of this ghostlike being with Elon, hanging out by a fire pit. He went on a podcast afterward and said, “Elon told me the sun is really big.”
I'm like, “Fuck, man. Where is he going with this?” Now we know where he's going with it, because he basically filed to launch 1 million satellites. The thesis is that you can have data centers in space.
The structural argument is that you have this pack of satellites cooled by space. In the real economy, you have all this coolant and all these logistics. His assertion is that he's going to launch 20% of the U.S. power supply into space and have data centers in space.
The Anthropic and Colossus deals are important because they foreshadow what he’s pitching: he was already the visionary guy who got the Colossus data centers, and now everyone’s renting them from him and paying him billions of dollars; he’s going to do it again, this time in space. What you're paying this insane multiple for is the idea that you could make $300–600 billion a year from basically deflation, because the alternative is building out massive, unprecedented amounts of data centers in the United States. People don't want these data centers by their houses, and it drives up power prices.
That's his thesis. When you're buying into this thing, you're buying into a story. A top investor who I won't name told me, “The way I think about SpaceX is, I don't know about SpaceX as a company, but as a stock, it's one hell of a stock.”
Oh my God, that is a sick line. I love that. How do you, GoodAlexander, think of SpaceX as a stock? Also, I feel like the sun thing is new. This is a new narrative. Tell me this first: what do you think of SpaceX as a stock, and roughly, do you have a plan for what you're going to do when the IPO is live?
Yeah, I'm short every telecom stock right now.
And explain that.
Everyone's long every space beta into the SpaceX IPO, and you cannot have better space beta than what Elon is going to deliver. It's impossible. Essentially, the thesis is that all the space trades go into hardcore, full reverso mode. That's my very near-term view.
It's just too lucrative. It swallows everything. How can Aster go up when you have a Hyperliquid-type thing?
The amount that the bankers are getting paid is also important. The reason why Jamie Dimon is interviewing Elon Musk is not because he likes Elon Musk. It's because JPMorgan's investment banking division is minting money on this IPO. That's why he's fast-tracking it into the Nasdaq.
I think it's something like a 3-month fast track. I don't know the exact terms of it. The S&P refused to do it. They tried to pressure the S&P to do it, and then—
The theory that I have is that Elon is the greatest equity salesman of our time. You don't want to bet against his ability to completely dominate the space narrative. The thing about the space narrative is that it's completely unfalsifiable, because it's going to take him quite a lot of time to build the products.
The other thing that I was originally bearish on was the Cursor deal. I was like, “Dude, what is this? Why are you paying $60 billion for Cursor?” The thing that changed my mind, or got me more excited, is that I use a lot of AI coding tools.
They cooked something with Colossus called Composer 2.5 Fast, and they've started training new coding models with all of Cursor's proprietary data. I thought, okay, this is a meme. This is something you're doing because Grok basically failed and you're just trying to play catch-up.
Then I used it, and I'm like, this is a pretty good model. It's nuanced. Claude is the great designer, engineer, and mathematician. Codex is the reliable tank that just gets shit done.
Composer 2.5 Fast is just ridiculously fast. You can analyze a cryptocurrency, a new crypto protocol, or a Bittensor amendment at the speed of thought. You can literally ask, “What exactly is this subnet? What does it do in the codebase? Create a chart for me that shows me visually what this does,” and it'll do it.
With Claude or GPT, you're like, “Make me a chart,” and then you go get a coffee. Ten minutes later, the chart is done, and it totally ruins your flow. It destroys it.
That really changed some of my thinking. Before, I was like, okay, this is the end, because it's a 33% compounded revenue growth thing trading at 99 times sales. Good luck at a $1.7 trillion valuation, with all these related-party deals between xAI and SpaceX, jamming xAI into SpaceX at the last minute.
At first I was like, what are you doing? Now I'm like, okay, what you're doing is telling a story about AI data centers in space. That's the story, and it's a good story. Elon can sell it.
Do I think the stock is going to puke when it IPOs? No, I don't. Do I think it's going to completely suck the liquidity out of everything that's possibly or tangentially related to the story he's telling? Yes, I do.
That's sort of my opinion on SpaceX.
And why short telecom specifically?
I know the stocks the best. I used to be a telecom analyst.
I've never heard anyone say that.
Yeah. I was on a TMTC team, so I covered Verizon, T-Mobile, and SoftBank, and I traded telco stocks. I just know that these management teams are clowns. They're shareholder-destructive. You can't buy them—I’m confident they're not going to do anything good with the influx of capital. That's sort of the thesis.
I started ramping back up on them because I started building this position, just to make sure that T-Mobile isn't growing at 100% a year and that I'm not missing something. It doesn't seem like I'm missing anything. If anything, let's say SpaceX is right and they launch 1 million satellites. Elon suddenly has everyone by the balls. If SpaceX is right, he has everyone by the balls—from internet to AI. Let's say there's AI compute on your phone and Elon can beam it from space. What are you going to do as T-Mobile? You're wrecked. You pay whatever price.
That's sort of how I was thinking about it. It's very much a trade because I wanted to have this on in some fashion. I also have a ton of long telecom exposure because SK Telecom is one of the big early investors in Anthropic, and SoftBank is a big investor. A lot of people don't even realize that SoftBank has a telecom business in Japan, and they have a monster stake in both Arm and OpenAI. My other IPO exposures happen to be long telecom stocks.
So I was already asked to short them because I wanted to get rid of this risk. I don't want to be long telecom stocks because I just don't have a view on the telecom sector. But then I started looking into it and thought, "Wait a second. Why are these things up triple digits year on year?" And you're like, "Okay, all right, got it." That is a perspective I did not think about at all.
I told the stream you're the IPO specialist now. I think that's going to be your assumed role. I did not think about the telecom thing at all, but it makes a lot of sense. You also can't compete. I like the way you thought about it, where a lot of these narratives are far out, so you kind of have to take him at his word until it gets built or it doesn't. You spend so much time in anticipation of it, and the price action happens anyway, whether he builds it or not. It's sort of irrefutable.
Dude, Tesla—I’ve followed Tesla my entire career, and I've just seen everyone die shorting this thing. There are generations of people saying, "Elon's not going to deliver the cars. Self-driving isn't going to be a thing for the next 10 years." And you're absolutely right, but the stock is up 300% and now you're out of a job. You just can't fade it.
The thing about Elon is that he's going to be right eventually. It might be 10 years later, but in the meantime retail is going to keep buying his stock because he's the best equity promoter of our generation.
What do you think about the prospect of that? The first thing I thought about when I saw the SpaceX IPO was short Tesla, because we're crypto guys and the 2 stories we know very well are low float, high FDV, and a dev launching a second token. I'm like, "The dev launched a second token. I'm short Tesla."
But then immediately Tesla starts hitting red candles, and then the leaked merger rumors start: SpaceX acquiring Tesla, merger, merger, merger. Everyone's terrified to short this thing because at any moment you could get blown out. Elon says, "Fuck it, I'll break some whatever. I'll make it happen," merges Tesla and SpaceX, and then you get blown out. How do you think about that prospect?
I don't like shorting it. The only way I would ever short Tesla would be if Trump did a deal with the Chinese and we got Chinese electric vehicles in the U.S., which by most standards are market leaders in terms of cost and consumer experience. Then I would short Tesla.
I don't want to short Tesla. I originally thought it would be a good idea because of the dev's second token, but then I thought, "Why would I short a robot?" Why would I short the only liquid expression of robotics and self-driving in the public markets, other than Google, when I could short some 70-year-old telecom company that's growing revenues at 6% a year and whose stock is up 180%?
Generally, when I short things, I try not to short things that could invent the cure for cancer. That's one of my heuristics. I learned it the hard way because my first big trading blowup was shorting Plug Power. It's an amazing stock because it shouldn't exist. It's like an altcoin that's pumped 800% multiple times on hydrogen fuel that it never delivers. It never actually delivers anything. It just says that it's going to, but it always pumps.
I thought, "This is going to zero." Then the CEO left and I tripled down. I thought, "All right, he left. It's a fraud. It's completely over." The market was like, "Well, if the CEO left, we actually didn't like this CEO, and a regulator didn't like him either." I didn't realize that, and the stock went up 600%.
Damn.
And in a straight line. That was my first trading blowup. I did this ritual called the 5 W's, where you're asking, "Why did this happen?" What it really boiled down to was that I wasn't humble enough to say, "Okay, if this has been a scam for 20 years, what stops it from being a scam for 23 years?" Nothing. They have a track record of pumping 300%.
Generally, with shorts in crypto right now, there are no good shorts because everything's market cap is so destroyed. In equities, there are actually a lot of good shorts, and a lot of them are these dinosaur companies that have gone on generational runs.
Even Intel—when Intel was going limit-up, David Gills and I were in Intel because we were memeing early on. We were pretty early on Intel.
I remember David Gills being very early on Intel.
Yeah. We were joking, "Incel inside." It was just a meme, and then it started pumping. Trump took a stake. It didn't really go for quite some time, and then when it did go, you really zoned in on the quarterly revenue and realized the stock was up 200% while quarterly revenue was up 9%.
At least with the memory stocks, you're like, "Okay, you're growing at triple digits. This is insane." The compounder bros can make a straight-faced argument that if it grows 200% a year every year for the next 10 years, maybe it's reasonable. But with some of these stocks, you're growing revenue at 8% a year. It's not like you're doubling or something.
Those are the ones where I'm a lot more comfortable betting against them. With Tesla, you're like, "Dude, I don't want to bet against AI." Come on.
This is something I kind of stole from Flood, but we came to the thesis on longing oil on our own. If you're long oil, you're actively long an asset that the entire world is incentivized to see go down. Good luck.
It's similar to asking, "Why would I short a stock that could cure cancer or solve artificial general intelligence at any moment?" You're just going to get wiped out.
Okay, the dust settles on the SpaceX IPO. Then you're immediately thinking about the ones on the horizon. I don't know if this is a terrible analogy, but we saw Broadcom earnings about a week ago. And, by the way, if we've topped for the summer, that was the first sign.
The fact that you're looking at Broadcom earnings, by the way, means we've just—what have we done?
I know. Watch this. It's going to get worse. What I'm about to say is going to get worse.
Broadcom's earnings happen. They double-beat, but they don't beat by enough. It goes down 20% and drags everything down with it. Then today, the same thing happens. Oracle beats, but it doesn't beat by enough and drags everything down with it.
Is there a scenario in which SpaceX beats or launches green, but it doesn't launch green enough or pump hard enough, and then everybody panics? "Oh, God, there's not enough liquidity." It's sucked from all these other sectors. Space is down. Telecom is down. Maybe semis are down a little bit on the speculative hop.
Then you're immediately looking at OpenAI and Anthropic thinking, "We don't have the liquidity for this." How do you think about the dust settling after SpaceX, and how would SpaceX's performance affect the way you think about Anthropic and OpenAI?
Yeah, if SpaceX actually tanks, that would be horrendously bearish. I believe that's probably why people are doing such large deals with SpaceX, because Elon probably called them and said, "Hey, if my stock trades down, your IPO is dead."
So, it’s like everyone’s incentivized to make this thing go up. The great thing about the sun story is that it’ll just be Elon running around saying how big the sun is for 9 months. That’s an unfalsifiable bull case, which is his specialty. And it has some solid physics arguments, right? Because Elon’s an autist—he’s figured this out. He’s like, “Okay, actually, structurally, there’s no reason why you wouldn’t build a data center in space.”
I’m not bearish. I don’t think these other companies are going to collapse. It’s more like that is going to suck up the liquidity from so much other stuff. I think that’s the cleaner expression.
Regarding the other 2 IPOs, they’re very different stories, and they have very different risk profiles. Investors—and the current market—view Anthropic as this de-risked enterprise bet that reported this made-up profitability number. But when you dig into it, it’s deeply free-cash-flow negative. Nonetheless, they say they’re going to become free-cash-flow positive in 2027, which is a typical enterprise IPO story where you’re like, “We’re burning money right now, but if we have our net retention number above blah, it’s all enterprise.”
OpenAI is the scary one, right? You’re raising—you know, you’re going to run out of money if you don’t do this IPO because you’re burning so much money.
Yeah. And I think, funnily enough, OpenAI is the one I’m most optimistic on. In my past life, I was a founder, and I used to trade Amazon stock really actively. Back when I was on the buy side, essentially what happened was everyone was in Amazon because of AWS, and they all marked the retail business at zero.
I was like, “Dude, in China right now, the retail businesses of these e-commerce players are crushing it because they have ads, and right now Amazon doesn’t have ads.” As of Q1 2026, there are 300 million active shoppers on Amazon, and it’s a $69 billion-a-year run-rate advertising business. This business that they thought was worth zero a decade ago is actually just this money-printing machine because it has an advertising engine. OpenAI has 3 times the number of users as Amazon.
It has 900 million ChatGPT DAUs. The current market narrative and consensus is that Sam Altman’s an idiot because he has 900 million WAUs and he’s losing money on those WAUs. It’s literally the exact same thing that caused everyone to be sidelined in Amazon for a solid 10-year run, just because they didn’t understand that it’s really hard to get 900 million users. Appealing to normies is not easy, and Altman’s done a good job.
The contrarian angle with OpenAI is that they had this 6-week pilot for OpenAI ads that printed $100 million, with massive oversubscription. They’re targeting $2.4 billion of revenue, which is very small relative to where I think this thing could go.
I personally think that if OpenAI really focuses—I mean, Bing, for example, has about a 5% market share in search, while Google is a complete monopoly. OpenAI’s GPT-5.5 Instant search experience, if you use it, is really, really good. The reason it’s so good is that it can consume 10 pages at once and tell you what’s on those 10 pages without you having to open 10 different sites and click through everyone’s ads.
And so, basically, the market consensus right now is that Anthropic is the killer. It’s the winner. The Fable model—or Mythos, as it’s called—is excellent. Their story is that they’re going to become free-cash-flow break-even, while Sam Altman is an idiot and is going to go bankrupt along with Oracle. That’s the story.
Yeah.
And so, yeah, I actually think the story is probably—
A little different than what people think. I’m a big OpenAI believer. That’s my IPO preference. Okay, give me one second to flip these headphone batteries. That was a sick fucking take. I want to ask you about Anthropic and everything that’s happened there, and then we can go back to OpenAI. Can you first touch briefly on the fake numbers comment, what that means, the implications of that, and how that plays out?
Well, okay. When I say fake numbers, if you read Anthropic press releases—and even if you have AI models interpret Anthropic press releases—you would probably come to the conclusion that they’re free-cash-flow positive, when in fact they’re disclosing a non-GAAP operating profit metric that is made up.
It’s common in enterprise businesses where they’re talking about—and it’s an important number because the story they want to tell is that they can serve these models profitably on a unit-economic basis. But that doesn’t mean their capex is not the problem.
One way to look at it from an equity analyst perspective is: do you have to spend the money in capex or not? Can you cut your capex or not? If you’re Coca-Cola and you cut your capex, people are still going to drink Coca-Cola. You can cut your capex and no one gives a fuck. If someone launches a new product, it doesn’t matter.
If you’re Anthropic, you cannot cut your capital expenditures and expect your unit-economic profitability to last, because you will get wrecked in the next model upgrade. The reason I say it’s misleading is because they’re characterizing their capital expenditures as a one-off, when really I would say they’re structural to their core business in such a way that they’re effectively an R&D expense.
In my opinion, that’s why it’s misleading to say that they’re unit-economic positive. If they stop training models, those unit economics would go off a cliff. I just think it’s bullshit. That said, they have structurally better margins than OpenAI because almost all of their revenue is from enterprise, whereas OpenAI is a different story.
Got it. And so I guess, as a follow-up on Anthropic, we’re 24 hours out from the Fable launch, and it is—I mean, I don’t know—the most controversial model since the original Gemini woke launch. What is your take on the whole situation, how good it is, and this “Fuckthropic, they’re dangerous” narrative as well?
The model is excellent. I think a lot of the takes that it’s not good are probably misinformed.
Originally, I was getting upset because it was refusing to do some security analysis on a feature I had just shipped. I was like, “Okay, I have this wallet integration. I want to check it.” And it was like, “No, I won’t do that.” You’re like, “Fuck, man. What am I paying this thing for?” But then—
But then you learn how to prompt it, and you actually learn, for example, that it’s allergic to making security mistakes. It will actively find security mistakes in the repo as it’s working and be like, “By the way, there’s one thing that could be an XSS exploit. I’m going to fix it. You should fix that.”
As you work with it, you’re like, “This thing is really good.”
Additionally, I write a lot of code internally, and one of the metrics I like is: how many bugs did Opus 4.8 find in GPT-5.5’s code? Not many. How many bugs does Fable find in 5.5’s code? A lot. It’s basically the alpha model in terms of deployment.
And then when you go and report those bugs to 5.5 and you’re like, “Okay, do you agree with this assessment?” it’s always like, “Yes.” It’s basically right.
Right. So it’s kind of like Fable is the alpha model in terms of deployment, and I think that’s just real. I actually don’t like Anthropic. I don’t—
I remember.
But at the end of the day, you have to respect the fact that they have a really solid model. You can’t pretend it doesn’t matter that I don’t like them. At the end of the day, the model is the model.
So, you understand the economics probably a lot deeper than I do, but it feels like the expenses here are racking up, and it feels like we’re exiting the subsidy era on compute for consumers. I’ve seen all these headlines that Microsoft is no longer really looking to use Anthropic because they’re spending way too much money. The Uber CEO went on Invest Like the Best and was flexing this gargantuan number that the company had spent on Anthropic credits, saying they’re basically done using it or that all engineers are capped.
Tup King from our team made this tweet today. Chimothco tweeted something I thought was pretty funny: “I look forward to our Chinese brothers liberating the knowledge from within Fable 5 and selling it to me at 5% of the cost and 2 times the speed.”
How much of a problem is this going to be for these IPOs and for the usage of these models going forward, if it’s just this forever cat-and-mouse game of an American lab shipping an incredible model and China distilling it—or however they do it—and open-sourcing it at 5% of the cost or 5% of the speed? How big of a problem is the end of the subsidy era here?
So this is why I like OpenAI more than Anthropic. Got it. The basic premise is that—
Right now, if you wanted to run Claude in a single agent—let’s say a typical engineer runs 2 agents and can pay attention to both—that costs about $100 an hour. That’s expensive. You’re dropping a couple hundred grand a year. You’re basically paying for a full-time engineer to use this model.
And then the question is: What’s the ROI on doing that? For most applications, the ROI is tied to the actual ROI of software, because what are you doing with that stuff? You’re building software. How much software demand is there going to be on a forward-looking basis if everybody can customize their own software? It’s going to be lower, right?
If I can build my own software, I don’t need to pay for yours. If I can build my own video game, I don’t need to pay for yours. So what you have in enterprise is this nasty situation where the models keep getting more expensive, and the value of delivering those models keeps going down because the models essentially cannibalize their own end market.
Yeah.
Whereas in consumer, the thing that I like is that Google is already this monopolistic behemoth that just prints money, and everybody talks to ChatGPT all day about what they want to buy, their feelings, and everything.
Consumer is a much more greenfield market. I already know from running ads that people don’t think about product SKUs. When people are looking for a coffee mug, a lot of the time they’re looking for a cute Christmas coffee mug. They’re not looking for this brand’s coffee mug with this decoration. They want someone to help them think through their purchase decision, and large language models are really good at that.
They’ll say, “These are the top 10 Christmas coffee mugs, and this is my preference.” People like that. I’m very bullish on it, which is contrarian because right now everyone is saying, “Enterprise is the only thing.”
We’ve used Uber, right? I use Uber a lot. You probably do too. Is the app actually better than it was 2 years ago?
I can’t tell, man.
I can’t tell you a single difference. It’s more expensive, probably, right?
Right?
So it’s like, okay, they just burned a bunch of money on Anthropic. It’s the same thing with Salesforce. They’re like, “45% of our app is written by AI.” Cool. Is the app actually better?
You know, maybe there are some new features. That’s the thing I don’t like about the enterprise AI side: it’s just replacing a lot of software.
Now, I think the exciting stuff is biology. Dario Amodei is a biologist, and the thing I’ll say about Fable that’s exciting—especially in cryptography—is that we saw the Zcash hack.
Zcash is a really good example. You have a bunch of Israeli geniuses employed by Zcash, and they’re the only people who actually understand Orchard. It’s a very complex cryptography system. It came out in 2022, and it had bugs in the code sitting there for years that nobody was able to find because it’s genuinely hard to understand. It’s also very expensive to hire people who can understand it.
Now, that’s the really exciting thing about Anthropic: it can understand it very well, and it can find the specific problems with it. Then it can actually improve those algorithms. Especially in things like privacy protocols, really complex technology, or emerging areas like biotech and mathematics research, you can be more bullish because maybe the end markets can grow a lot. These systems can really move the needle on innovation, discovery, and better privacy technology.
But that’s going to be a very small percentage of sales relative to enterprise buyers like Uber. In Uber’s case, the app isn’t better than it was last year, but they’re spending $20 billion. You’re like, okay, the ROI probably isn’t there.
Do you think we’re about to have this explosion of drug discovery? Are you aping into biotech stocks, pharmaceutical stocks, and things of that nature?
You know, it’s funny, because there’s a good expression that I think applies to me: being early is indistinguishable from being wrong.
I wrote, a couple of years ago, something called The Blueprint Manifesto. My thesis was that the scary thing about AI is that if you had a company or an AI algorithm that pulled ahead in biotech research and could deliver vaccines or things that you couldn’t get elsewhere, that would be incredibly powerful.
The only thing you can’t buy with money is time. Imagine if there were a company able to say, “You can be young. We can keep you young if you work for us.” You’d be operating on a different level of monetary value.
For me, the scary version of the permanent underclass is more of a Gattaca vision.
Gattaca is an old movie. You should watch it.
I’ve never seen it.
Classic.
Yeah. It’s basically a society where half the population is genetically engineered. There are 2 brothers, one of whom is genetically engineered and the other is not.
There are these awesome scenes where they’re swimming in the water, and the non-genetically engineered guy beats the genetically engineered one. It’s a good movie. It’s from the ’80s, right?
So people have been thinking about this for a really long time. I do think it’s a very powerful narrative. A lot of crypto people are very active in this. Fred Ehrsam pivoted out of crypto into—
Bryan is doing this, isn’t he? Bryan Johnson?
Brian Armstrong—excuse me, Coinbase.
He’s running a big fundraise. A lot of early crypto people, like Dovey Wan, are super longevity nuts. Vitalik is hanging out with the human genetic engineering guy’s ex-girlfriend. I don’t know what’s up with that. She was posting photos with him.
Yeah, there’s an aggressive overlap between crypto and longevity. What do I think of it? I’ll give you my take.
If you look at Bryan Johnson’s biggest investment, it’s Ginkgo Bioworks. Its ticker is DNA.
Down 99%. It’s a down-only chart. He publicly—
Bioworks Holdings.
He publicly shilled it. It’s in ARK Invest—
And it’s—
A $500 million market cap.
Down 10x, right? It used to be massive, and now it collapsed. What you find when you talk to ultra-high-net-worth individuals who are investing in biotech is that you think longevity would be a good investment, but historically it’s been a very bad investment.
The reason is interesting. Let’s say there’s a new startup in Panama that can do something for ocular degeneration. Eric Schmidt will already have bought that company before you even have any interest in investing in it.
All the good stuff gets acquired. The really good stuff gets invested in at a price-invariant multiple because the ultra-rich care about their health more than they care about their returns. When you’re competing with people who have no price target—who don’t care what they pay for things—and you’re a return-focused investor, you lose money.
You’re losing. Yeah.
That’s sort of the problem with longevity investments and the reason why longevity funds haven’t done well in the past.
That said, Demis Hassabis, the Google guy, has basically raised billions of dollars. I would take that seriously if you can get access to it. I’d rather be exposed to the stuff coming out of labs like Google or Anthropic. They’re going to do something in biology because that’s Dario’s background.
I’m less certain about longevity as an investment. I think Bryan Johnson is a cautionary tale. If longevity investing worked well, why is he selling olive oil? He’s doing it because it didn’t work well.
It’s a really tough game, and it’s an insider market. My bias with ARKG and biotech stocks is to stay far away from them. I’m not a domain expert, though I know people who are.
There’s a really excellent biotech portfolio manager in Puerto Rico. I had a night out with him, and there were a bunch of crypto guys who were into longevity, peptides, and biotech. They started rattling off their investments, and he was like, “That one went through a clinical trial that was statistically invalidated. And that one…”
The level of detail that really good biotech investors operate at is truly frightening.
You don’t want to be in the trenches with them by default. I stay away from that personally as an investment. I would change my mind, I think, if I had an order of magnitude more money.
Yeah, that was an honestly well-thought-out take. Thank you for that one. I’ve got to get one more rant out of you. Last time you were here, I told you about how we were covering geopolitics and war for the first time, and we were talking about this headline-terrorism concept, where Trump is clearly in the books and clearly timing headlines around futures open, market close, and things like that.
Since then, we’ve had the Intel run. We’ve had “Go out and buy a Dell.” We’ve had IBM with the quantum investment and the general pumping of IBM. We’ve had Jensen, Marvell. We had Jensen, QCOM—Qualcomm—which didn’t go very well. We’ve basically gone from Trump timing strikes around market close to Trump and Jensen being like prime Shaq and Kobe, selling stocks and picking winners.
I feel like it’s devolved another order of magnitude, or maybe multiple step functions, further into this. It’s gone from “That was bizarre” to “This is the new norm, and this is how we play.” Where are you with playing this game, and with the evolution of this game that seems to have happened?
I’ll make a controversial take that we can revisit in the future. I believe there’s been intervention in US equities and US equity futures, and I think future administrations will look back at some of the actions taken in the oil futures market as well as the S&P futures market and ask, “What the hell is going on?” I think Scott Bessent is actively involved. That’s one of my takes.
There’s some very abnormal price action in the market, and Trump is myopically focused on the stock market in a way that, given the portfolio of his behavior, it wouldn’t shock me if he called Bessent and said, “I’m going into a run. Make sure the S&P goes up.” That’s something Trump would say. You’ve seen him for 8 years; we know what he’s like.
I do think the market is controlled by insiders. I don’t think it’s obvious that it’s monetizable. The canonical example is Hegseth buying a large position in defense stocks that then collapsed after he bought them. Is there egregious insider trading? Yes. Are they TACO-ing equities? Yes. Is it worth trying to predict what they’re going to TACO next? Potentially.
A lot of people are building Polymarket insider-detection systems because they’re trying to identify who in the Trump administration knows when certain things are happening. Insider-trading wallet tracking is popular. I know a lot of people who’ve tried doing that.
I feel like I’m being too bearish on money-making activities. Yes, there’s definitely a way to profit off of this, and it absolutely is relevant in terms of the government and the political environment, because the average American looks at this and thinks, “These are rich guys playing dice with the global economy.” I think that’s politically unpopular.
I fully agree. Also, for what it’s worth, this feels like such a microcosm of what’s actually happening. I remember there was a 2-month period when we were doing BNB Smart Chain—basically an on-chain memecoin season—and the whole game was which coin CZ was going to engage with next. It would go up for 5 minutes, and if you hadn’t bought it before he showed it, you lost money.
That wasn’t exactly the case with Intel, but Intel pumped more than Dell, which pumped more than IBM, which will probably pump more than the next one, and the next one, and the next one. Let’s just assume I like the take, because I agree with you. I don’t know if I have as much proof, but Bessent and Trump are very clearly in the order books. Why do you think that’s happening? What did you see that made you think that’s happening? More importantly, what are the implications? What does it mean for the future of capital markets, and how should you think about trading, especially as a retail investor in the US?
The market—there was a meme that I think Base and a16z started: anything bad happens, and the market is down 0.3%.
If you trade war markets, or if you’ve traded them for a while—for example, Ukraine in 2022, or even COVID—you have this really gappy market. There isn’t usually someone just showing up and buying the dip in massive size, especially if there’s geopolitical escalation. It’s not a thing.
That’s exactly what happened. I thought, “Okay, there is someone buying unlimited amounts of S&P futures down 0.5%, such that we literally can’t go down on any bad news.” That person is probably a Soros-trained trader who was trained by Stanley Druckenmiller, who said, “I like it when markets go up on good news.”
So it’s probably literally Scott Bessent, who was trained by Stanley Druckenmiller, who told him exactly the phrase he would need. I’m looking at this price action and thinking, “This is nothing like the price action in any other risk-off geopolitical event I’ve seen.”
Scott Bessent was trained by a guy who said that the market loves reading into bad news and going up, and Trump has basically no respect for the rule of law. You combine those things, and you ask, “What’s the easiest explanation for the market not dropping?” The easiest explanation is the obvious one.
That’s sort of a conspiracy theory. I think we’ll see, in a couple of years, investigations into this period that will be more concrete.
In terms of what it means for markets, I think people probably overestimate it. I have a relevant take: I think there are some stocks that embody the legitimacy of this administration. One of those stocks is Palantir.
I think Peter Thiel fleeing to Argentina, JD Vance being on the wrong side of things, and Democrats generally doing quite well means there’s probably a set of stocks that are Trump beta and maybe won’t do as well. Unfortunately, that hurts.
One of the things that scares me in the current crypto market is the odds specifically on the Clarity Act. USDT and USDC have basically flatlined in supply growth when they were supposed to be adding hundreds of billions.
Scott Bessent’s big plan was like $10 trillion, right?
You’re asking, “Why is it stuck at $180 billion?” Circle’s trading like shit, and you’re thinking, “Okay, you’re not able to get this Clarity Act passed, which is your hallmark legislation. Why isn’t stablecoin volume going up?”
Damn.
It’s not great. Howard Lutnick has this massive Tether equity stake through his son’s business, which is the auditor of Tether’s reserves. There are all these businesses at the intersection of Trump, capital markets, the state, and its spending. We know what they are: Circle, Palantir, Tether, Ripple.
There’s a series of entities that are based on this regime continuing, and I hope, for the sake of crypto, that they play the politics well. Ripple is actually really good at this. Chris Larsen is intensely obsessed with climate change, so they naturally play it well because they’ve been in bed with all these left-wing politicians for a long time.
There’s a lot of gap risk. Actually, Anthropic and Palantir are the most relevant things to this IPO discussion.
Okay.
A lot of people aren’t paying attention to the language Anthropic is using regarding its partnership with Goldman Sachs. They’re saying, “We’re dropping forward-deployed Anthropic engineers into these banks,” and that’s Palantir’s name for what they’re deploying into banks.
By the way, Karp basically called Dario implicitly “retarded.” Dario went out and said that high unemployment would result from models and that people need to be worried about it.
Oh, I saw this.
He said that high unemployment would result from models and that people need to be worried about it. Karp chirped up and was like, “Yeah, well, if you think this is how it’s going to happen, maybe you have a high IQ, but you’re actually retarded.” That’s what Karp said.
I think that probably angered Dario Amodei, and as it should. I think Anthropic is powering a lot of Palantir Ontology deployments.
I believe that if they came to market, one thing we know about Anthropic is that they have no fear of launching competitor products to their various customers. Figma is excited about using Anthropic, and then Anthropic is like, “Yeah, I made an internal Figma, and your stock’s down 80%. That sucks.”
That’s what Anthropic is like when they don’t give a shit. I would be very afraid of what Dario Amodei is like if he does give a shit. If he really wants to say, “Fuck Alex Karp,” because Alex Karp sort of told him to fuck off, you really don’t want to be on the wrong side of Goldman Sachs and Anthropic right now. You’re going to get crushed. If JD Vance was your cover, good luck, man. Honestly.
One of the scary takes is that one of the really good macro indicators is stablecoin supply. Tether supply and Circle supply are supposed to be up-only charts, and it’s not happening. They need to start going up again in order for us to rebound in a material way in cryptocurrency, or in a lot of risk assets that are not AI stocks. We need to see stablecoin supply start to run. If it doesn’t, that’s a very strong canary in the coal mine that things are not good.
From a trading perspective, where do you want to place bets right now? Do I really want to fade SpaceX? No, I don’t. Do I want to buy Anthropic? Probably not. It’s probably too expensive in consensus. OpenAI seems appealing to me. I’m there. I’m in SoftBank. I like the advertising thing because people are sleeping on it, and I think it could be huge.
Then you’re like, okay, in public markets there are all these expressions of the Trump-esque era. How does this era end? There are two ways it could end. It could be Marco Rubio and the anointed one continuing a Republican administration into the sunset, with Larry Ellison winning. Or there’s another era where Gavin Newsom and AOC just show up, regulate everyone to death, and investigate everything that happened over the last 4 years.
Unfortunately, I think the second thing is more likely to happen. The excess is going to be paid for. It’s pretty crappy for people like us because I think you and I, to some extent, have a lot of personal beta to things staying the way they are.
Whoa. I love the “If JD Vance is your cover, good luck” line. It’s also interesting that Palantir is probably the most— I mean, it’s the only stock that Trump tweeted the ticker for. He tweeted ticker PLTR, and it’s lower. They’re at par with where they were when he did that 2 or 3 months ago. There is something to be said there: the most tightly anointed Trump beta is trading horribly. Go ahead.
I think there’s a really interesting core Palantir story. I’m doing the work on it because I used to work at Palantir, although it’s been a long time since I’ve been there.
How long did you work there for?
I worked there for 2 years. I did capital markets stuff. The interesting thing—the thing I don’t have insight into—is the core claim. The core claim is fascinating. Alex Karp’s thesis is that LLMs don’t work out of the box.
In order to make LLMs work, you need to have an ontology. An ontology is a structured way to tag your data in an intelligent way. That aligns with their old intelligence product, because the old Palantir pitch was about the 9/11 Commission Report: How did 9/11 happen? How did we screw this up? Why weren’t we able to kill Osama bin Laden when we knew exactly who he was, where he was, and everything else?
The two conclusions were, first, that agencies can’t share information. Second, when they do share information and you say that there’s a guy in a white outfit with a Jeep in Afghanistan, it turns out that’s every guy in Afghanistan. You can’t just say, “Kill the guy in the white outfit with a Jeep and a beard,” because that’s every guy in Afghanistan. You’d start shooting random people, and you can’t do that.
Palantir’s position was that you need to have an intelligence method for tagging whose Jeep is whose. How do we tell who the right guy is? Literally, that was the beginning of the ontology. They evolved that pitch into AI. They’re saying, “If you take the ontology that we built to find terrorists, map it onto business data and intelligence data, and then feed that into large language models, the large language models work really well.” That’s their pitch.
Yes. The problem with that pitch is that all of the research on vector databases, and even the behavior of prompts, suggests something different. One of the more common things in prompting used to be that people had these crazy prompts where they told models to do 19 steps in order to make Opus do what they wanted. There were these prompt engineers.
Prompt engineering has been going down a lot because, as models get smarter, you don’t need to feed them an ontology to get their shit done. The big question to me is: What’s actually under the hood? That’s the tell. I think we’ll find out if Anthropic and Goldman Sachs are actually able to do something.
Palantir is an amazing business, by the way, and there’s a reason why they have a huge multiple. They have great people, and once you show up, integrate everyone’s data, and have a system to interact with it, it’s really hard to unplug. But if Anthropic is motivated and pissed off enough, and they’re working with Goldman Sachs, the ontology might not be necessary for vector database performance.
That’s the really spicy thing. What if Karp’s wrong? What if you don’t need an ontology? What if you just need a Snowflake database with some automated tagging that Mythos can figure out in a day? Then you’re screwed. Then your stock is down 80%.
That’s a really interesting place to spend time analytically: Do you actually need an ontology for large language models to work? I don’t have an ontology, and I use LLMs every day, and they work great. What if I did have an ontology? Would they work better? I don’t know. Maybe I’m missing out.
Have you ever met Alex Karp?
I have.
What is his presence like in person?
A good summary is that he’s a CEO whisperer. I think a lot of CEOs, deep down, are worried that they’re not interesting people. If you’re Jamie Dimon or at the top of the finance game, you’re not afraid of being rich; you’re afraid of being boring.
Alex Karp is a very interesting person—extremely interesting, almost painfully interesting. He’s read every book. He does qigong barefoot in his office. He’s extremely inappropriate. He gives these 1-hour speeches to everyone who shows up, and he berates people and makes fun of them.
I remember one time he took me aside and said, “I think you probably imagine that you’re good at sales.” I was like, “Well, I was just in a sales meeting, and I think I kind of crushed it, actually.” He said, “That sale was going to close regardless of your presence.”
He’s able to pick people apart. He’ll tell people things in a very Steve Jobs-esque way. One of the things he tells people is, “You’re a spike talent. You’re really, really good at one thing, but you’re not good at anything else.”
His job at the organization is to take all these mutants who are good at exactly one thing, jam them together, and create an enterprise-grade solution. He’s a fantastic CEO. He’s great with retail investors. I was basically a retail investor because I was so young when I started there. I thought, “This guy’s crazy. I love it. This is bullish.”
But that can turn out poorly if you end up starting a fight with your biggest AI supplier.
Yeah, Dan, that was a great one. Thank you for that. That was a sick excerpt. This is going to be a memorable stream that we’ll come back to, and some of these takes are either going to age great or age badly. It’s a sick time capsule.
I guess, to give you a wrap-up, we’re going to have to do it again—maybe for Anthropic and OpenAI. You could be our IPO whisperer. I hate wrap-up questions, but I believe trading will go live Friday. What do you think are going to be the most important things to pay attention to? Do you think we’re up or down in a week on SpaceX? What should people pay the most attention to over the course of the SpaceX IPO and then the next 2 that are on the horizon?
Well, I think the most important thing is going to be retail understanding the space story, articulating it, and amplifying it. Because if you're paying 90 times sales, you have to keep in mind that you're paying a much lower sales multiple for OpenAI and Anthropic. So someone going into the SpaceX IPO really has to dream the dream.
You know, they have to dream the dream. In other words, you need Jamie Dimon and the Goldman research team. It's a concerted effort between Elon, the Elon marketing machine, and the sell-side marketing machine to build a credible story around space. The thing that would blow us up would be someone saying, "Okay, the reusable rocket thing isn't going to work," and explaining why, or saying, "The structural integrity of data centers in space isn't good." But I'm bullish on the ability to overcome that, because you don't have to do something in real life.
Yes.
You just have to meme people. As a long-term hold, it's probably a terrible investment because you've got all the execution risk, and it's already priced into the stock. Yeah.
And it's growing too slowly. What you need to see to hold it is that the Composer, the next Composer model, is actually good. If they start getting state-of-the-art stuff on Composer, where it's fast and they also have unfair compute advantages, then you just hold it. So I think my assumption would be SpaceX itself trades well while every other speculative asset trades down. It's path-dependent in the near term based on xAI being a thing.
Yeah.
If it's not a thing, then it just starts drifting lower. In terms of OpenAI and Anthropic, I honestly don't know; I'm just kind of bullish on both of them. Fable is incredible. Codex and Fable together are some of the best tech products I've ever seen in my career. You don't want to bet against the best product you've ever used, right? You're like, "Okay, it's expensive. It trades at 30 times, 40 times sales." But it's growing at 8×; it's growing insanely fast.
I think the thing that would make me raging bull on OpenAI would just be more successful ad pilots. There was a famous Facebook line—Zuckerberg was running around saying, "We did a trial where we turned off ads on Facebook, and people used it less because people actually like the ads." That's insane.
I remember I had this friend—he's actually a billionaire now. His name is Garrett Lord. You should have him on the stream sometime. You should get Garrett on. He was running an education startup and got into the AI-labeling game. This guy's just a genius. I remember he was fully bought into Facebook, fully bought into Meta—or Facebook at the time—and I was like, "What are you thinking?" He was like, "Well, yeah, if people like the ads, then it can only go up." And you're like, "Yeah, I guess that's true."
That's the OpenAI thing. Anthropic has already attacked OpenAI. They've run advertising campaigns saying, "You won't ever see an ad in Claude." It's like, yeah, but also Opus will [__] me out for no reason, and it's an unpleasant user experience.
Super unpleasant. You need to see Sam Altman or OpenAI give clarity around the user experience with advertising in ChatGPT. If people feel like they lose trust or DAUs decelerate because they're like, "Dude, I was going to you for advice, and now you're shilling me a diaper product," then I'm not going to use ChatGPT anymore. But if they're like, "Okay, I'm delivering you products," one of the magical examples is Cloudflare. Codex will automatically spin up Cloudflare tunnels without asking you. It'll just be like, "You're about to do something insecure with an HTTP port. I'm going to spin up a Cloudflare tunnel," which is a publicly traded stock. And you're like, "Huh? I would have maybe done that differently."
But you spun up a Cloudflare tunnel. Now I'm thinking about Cloudflare. This is basically advertising, right?
Yeah.
And I'm not that mad. I'm like, "Yeah, I can see why you did that."
Justified, at least.
A lot of the time, the agents will tell you, "Dude, you shouldn't build this from scratch." If you tell Fable, "I want to build HubSpot from scratch," Fable will be like, "You should just buy HubSpot." The models are not designed to blow up enterprise sales processes. So, yeah, I think the thing that people are probably sleeping on is OpenAI advertising. I think that's going to be a huge story. It's very contrarian because everyone's in love with Anthropic right now.
And I think the fade of all these AI stocks is probably Google. Notice that we just spoke for nearly an hour and a half about tech stocks without talking about Google. It's the most consensus buyside long. Everyone's long, right? Everyone thinks TPUs are the obvious cash-flow-positive, publicly traded comp. Their cloud backlog looks like fraud; it looks like a hockey stick. Everyone's long, and it's not trading that well. It's because Gemini sucks.
It sucks.
Nobody uses Gemini for coding. It's just not a good model. So I think the other take would be that Google, relative to consensus—the fact that we didn't spend any time talking about it probably means it's a fade.
Wow. And on that note, I just got to thinking: USD notional, bro. It really feels like Google is the top of the entire equity market by raising 80 billion of equity, not debt. For context, 80 billion is larger than Anthropic's last round, OpenAI's last round, and SpaceX's IPO raise. So, yeah, Gemini does suck. And good, Alex, always a fucking pleasure. Until the next IPO, man.
Peace, brother.
We'll get you back on. Thanks, man, for your time. Have a good one. Ace.