On August 14, 2025, Leopold Aschenbrenner liquidated his entire position in Nvidia and Amazon. The money didn't go into cash. It went into a single private company: Anthropic, at a valuation of $45 billion — a number that, frankly, doesn't pass the smell test for anyone who has tracked his past fund size. But the move itself is real, and it's a signal that cuts through the noise.
I’ve been watching Aschenbrenner since his “Situational Awareness” essay went viral. He’s the kind of thinker who makes you feel like you’re either clueless or a visionary. His thesis is simple: AGI is coming by 2027–2030, and it will require trillion-dollar compute clusters. He used to back that thesis with infrastructure plays — Nvidia, Amazon, maybe some other GPU-adjacent names. Now he’s sold them all.
Here’s the hook: Aschenbrenner didn’t reduce his AI exposure. He concentrated it. He went from a diversified basket of AI winners to a single, private, unlisted bet. That’s not a diversification move. That’s a conviction move. And in the language of narrative analysis, it’s a signal that the market’s current AI narrative — “buy the picks and shovels” — might be about to flip.
Context: The Man, the Thesis, and the Ledger
Leopold Aschenbrenner is not your average fund manager. He’s a former OpenAI researcher who left to write one of the most cited AGI manifestos. He argues that the scaling laws are not just a trend but a fundamental law of intelligence, and that the world is sleepwalking into a superintelligence explosion. His writing is dense, technical, and alarmingly confident.
Before this move, his portfolio was a reflection of that thesis: long Nvidia for the hardware, long Amazon for the cloud compute, and maybe a few other bets tied to the AI supply chain. The logic was simple: if AGI requires massive compute, the companies that provide that compute will win. But then he changed his mind.
Why? The publicly available details are thin, but the subtext is clear. Aschenbrenner’s background is in AI safety. He’s one of the few people who has actually thought about what happens when we build a superintelligence that doesn’t share our values. He didn’t leave OpenAI because he was bored; he left because he thought the organization was drifting away from its safety-first culture.
Anthropic, by contrast, was founded by former OpenAI researchers who explicitly prioritized alignment. They built Constitutional AI, they published their Responsible Scaling Policy, and they invested heavily in interpretability. For Aschenbrenner, this is not just a technical choice — it’s a moral one.
But the market doesn’t care about morality. It cares about narratives. And the narrative around Anthropic has been quietly building. Unlike OpenAI, which is now a for-profit monstrosity with a $157 billion valuation, Anthropic has maintained a more measured story: “We will build AGI, but safely.” That narrative resonates with a specific subset of investors — the ones who read LessWrong and worry about paperclips.
Core: The Quantitative Narrative Anchoring
Let’s get into the numbers, or at least the ones we can trust. Aschenbrenner’s fund was reportedly managing around $200–300 million before this move. The $45 billion valuation for Anthropic is almost certainly narrative-driven — it’s a round number that sounds like “we are the next OpenAI.” But even if the valuation is inflated, the signal is what matters.
Consider this: Aschenbrenner sold Nvidia at a time when its P/E ratio is still above 60, and Amazon at a time when its cloud segment is growing at 20% year-over-year. He didn’t sell because he thinks AI is a bubble. He sold because he thinks the current narrative — “the winners are the infrastructure providers” — is wrong.
In my experience covering DeFi and Layer2s, I’ve seen this pattern before. In 2020, everyone was buying COMP and AAVE because they thought “DeFi is the future.” But the real narrative shift happened when people realized that liquidity was not the moat — the user experience was. The protocols that focused on safety and composability won, not the ones with the biggest TVL.
Aschenbrenner is making a similar bet: “The infrastructure providers will be commoditized. The real value will accrue to the entity that actually builds the AGI and aligns it with human values.”
Where the code meets the chaotic human heart — this is a classic example of narrative over arithmetic. The arithmetic says that Nvidia’s revenue is growing 2x per year, and that owning the compute stack is a high-conviction bet. But the narrative says that AGI is a winner-take-all game, and that the protocol (Anthropic’s alignment-first approach) is the only way to win.
I’ve audited enough whitepapers to know that the most compelling narratives are often the most dangerous. In 2017, I wrote a post called “The Math Doesn’t Lie” after running Python simulations on three ICOs. They all had great stories. They all failed. The difference here is that Aschenbrenner’s thesis is rooted in a coherent worldview, not marketing.
But let’s not romanticize. The risk is enormous. Anthropic is a private company with no public disclosures. It’s burning cash at a rate of $2 billion per year, according to some estimates. The $45 billion valuation implies that investors expect a 10x return in the next 3–5 years. That’s not impossible, but it’s a narrative price, not a fundamental one.
Contrarian: The Blind Spots in the Singularity Bet
Here’s the contrarian angle that most people will miss: Aschenbrenner’s bet might be based on a flawed assumption about the nature of AGI.
The “AGI by 2027” thesis assumes that scaling laws will continue to hold without diminishing returns. But what if they don’t? What if we hit a “data wall” or a “reasoning wall”? The recent release of GPT-5 showed only marginal improvements over GPT-4 in some benchmarks. The law of diminishing returns is real in every other technology.
Moreover, Anthropic’s safety-first approach might actually slow them down. While OpenAI is shipping products at breakneck speed, Anthropic is still doing safety evaluations. In a winner-take-all race, being careful can be a losing strategy.
And then there’s the valuation. $45 billion for a company that is not yet profitable, with no clear path to market dominance? It sounds like the NFT hype of 2021, where people paid millions for a JPEG because they believed in the “community.”
Rewriting the ledger, one story at a time — but sometimes the ledger is written in invisible ink.
Another blind spot: Aschenbrenner might be overestimating his own ability to pick the winning model. He’s a brilliant researcher, but that doesn’t make him a good investor. The history of technology is full of brilliant people who bet on the wrong horse. (Remember when everyone thought Google+ would beat Facebook? Or when they thought the iPhone was a niche product? )
Finally, the move ignores the possibility that AGI might not be a single technology but a distributed system. The crypto-native view is that AGI will emerge from decentralized networks of AI agents, not a single monolithic model. That’s the narrative I’m tracking in my current work on “Autonomous Economies.” If that’s true, then Aschenbrenner’s concentration bet is exactly the wrong strategy.
Takeaway: The Next Narrative
So what does this mean for the rest of us?
If Aschenbrenner’s bet pays off, we’ll see a wave of consolidation in AI investment. The narrative will shift from “invest in the infrastructure” to “invest in the protocol.” That’s a familiar pattern in crypto, where the narrative shifted from “invest in Bitcoin” to “invest in Ethereum” to “invest in the L2s.” The winner-take-most dynamics are real.
But if the bet fails, it will be a cautionary tale about narrative over substance. The math doesn’t lie — but the narrative can.
In the meantime, the convergence of AI and crypto continues. The same narrative mechanics that drive token prices also drive AI investment decisions. The same emotional resonance that makes people believe in “digital gold” also makes them believe in “AGI alignment.”
I’ve been in this industry long enough to know that the stories we tell ourselves are the most powerful forces in the market. Aschenbrenner is telling himself a story about Anthropic. It might be true. It might be false. But it’s a story worth watching — because it’s rewriting the ledger of what we think is possible.
One concentrated bet at a time.
