In the twilight of August 2025, Leopold Aschenbrenner did something that would have been unimaginable to his former self. The author of Situational Awareness, the most rigorous manifesto for the scaling law orthodoxy, sold every share of Nvidia, Amazon, and every AI infrastructure stock he owned. He then took that capital—rumored to be around $45 billion, though the figure feels like a ghost in the machine—and placed it all into a single private company: Anthropic. This was not a trade. It was a governance signal. A declaration of faith in a specific technical and ethical path, executed with the conviction of a DAO founder allocating the entire treasury to a single protocol.
When I first read the news, I felt a familiar resonance. In 2020, during the MakerDAO governance wars, I watched as large holders concentrated their votes on a single collateral type, betting their entire reputation on the stability of a single asset. The community called it reckless. I called it the purest form of decentralized governance. Because when you strip away the noise, true conviction is not about diversification. It is about standing behind a single, irreducible belief. Aschenbrenner's move is the same, only now the asset is not a stablecoin—it is a frontier of intelligence.
To understand what this means for our space, we must first understand the man. Aschenbrenner is not a trader. He is a former researcher at OpenAI, a protégé of Ilya Sutskever, and the author of a 165-page treatise on the inevitability of AGI by 2027 to 2030. He argues that the scaling laws—the empirical observation that larger models with more data and compute produce emergent capabilities—are not just a trend but a law of nature. In his view, the path to AGI is a straight line, and the only variable is the number of FLOPS. He believes that trillion-dollar compute clusters are not a possibility but a necessity. His investment in Anthropic is not a bet on a company. It is a bet on a timeline.
But why Anthropic, and not OpenAI or Google? Here is where the technical thread becomes a governance thread. Anthropic's core differentiator is its commitment to alignment research—Constitutional AI, Responsible Scaling Policy, and mechanistic interpretability. These are not just buzzwords; they are architectural choices. Aschenbrenner, who started his career studying AI safety, saw in Anthropic a reflection of his own values: that the most powerful intelligence must also be the most constrained. In the crypto world, we call this "code is law." But in the AI world, it is "alignment is the protocol." He is essentially saying that the governance structure of the AGI itself—the rules that will guide its behavior—is more important than the raw performance of the model.
This is where the lesson for blockchain becomes visceral. We have spent years arguing about L1 vs. L2, proof-of-work vs. proof-of-stake, on-chain vs. off-chain governance. But the fundamental question is the same: Who writes the rules, and how do we trust them? Aschenbrenner's portfolio shift is a signal that he trusts Anthropic's governance of its own technology more than he trusts the market's valuation of infrastructure. He sold Nvidia, a company that sells the picks and shovels, because he realized that the real value is not in the hardware but in the protocol that runs on it. This is the same shift we saw in crypto when people moved from buying Bitcoin mining rigs to buying ETH. The infrastructure is derivative. The soul is in the application.
I recall a moment in 2021, when I was curating the Ethereal Archive, a small DAO of 120 members dedicated to preserving on-chain narratives. I rejected a collection of generative art that was technically flawless but algorithmically cloned from a dozen other projects. The artist was furious. He said, "But the code is perfect." I replied, "But the soul is missing." Curating the soul in a world of derivative clones. Aschenbrenner is doing the same: he is ignoring the derivative bets—the GPU stocks, the cloud providers, the AI ETFs—and focusing on the one entity that, in his view, has a soul. Anthropic.
Of course, the numbers are suspicious. The $45 billion figure, as noted in the analysis, is unverifiable. Aschenbrenner's fund, while substantial, was never known to manage such a sum. This is where the crypto parallel becomes sharp: in our world, we have on-chain verification. We can see the transaction, the wallet, the timestamp. In the private AI world, we have to trust a press release from a journalist who may have misheard a source. The opacity of private markets is a governance flaw. If Aschenbrenner had made this bet on-chain, we could verify the concentration, the cost basis, and the timing. We could build a reputation system around his conviction. Instead, we are left with a story. And stories are the only thing we have to curate.
But here is the contrarian angle: what if he is wrong?
The most dangerous blind spot in Aschenbrenner's thesis is the assumption that scaling laws will continue without resistance. There is growing evidence from within the AI community that the scaling curve is flattening. The cost of compute is increasing exponentially while the performance gains are diminishing. Aschenbrenner's response to this is to double down: bigger clusters, more data, longer training runs. But what if the real bottleneck is not compute but intelligence itself? What if the AGI timeline is not 2027 but 2047? In that case, his concentrated bet on a single private company becomes a catastrophic governance failure. The DAO that allocates its entire treasury to a single protocol dies when that protocol fails. Aschenbrenner's portfolio is now a single point of failure.
Moreover, the lack of transparency in Anthropic's governance is a risk that Aschenbrenner, as a former researcher, should understand. The company's board, its decision-making process, its conflict resolution mechanisms—all of these are opaque. In crypto, we have learned the hard way that opaque governance leads to centralization, and centralization leads to capture. The Tornado Cash sanctions taught us that writing code can be a crime if the government decides your code is a tool for money laundering. Anthropic's alignment technology could be similarly weaponized. If the AGI emerges and its alignment is controlled by a small group of people, the risk of regulatory capture is immense. Aschenbrenner's bet is not just a technical bet; it is a bet on the governance of that small group.
Curating the soul in a world of derivative clones. This is the third time I am using this phrase, and I mean it each time. Because the soul of an investment is not its financial return. It is the values it represents. Aschenbrenner's portfolio is a mirror. It reflects his belief that the most important thing in the world is the alignment of superintelligence. In the same way, a DAO's treasury allocation reflects its belief that a particular protocol will bring about a more equitable financial system. The question is not whether the bet will pay off in dollars. The question is whether the bet is an authentic expression of the investor's core values.
I have seen this pattern before. In 2022, during the bear market, I wrote a manifesto on "Decentralization as Emotional Security." I argued that the real value of blockchain is not the technology but the emotional resilience it provides—the ability to survive a crash without losing faith in the system. Aschenbrenner's move is the same. He is not selling because he lost faith in AI. He is selling because he has too much faith in one specific version of it. He is consolidating his emotional security into a single point of belief.
The takeaway for our community is this: The next bull run will not be about infrastructure. It will be about conviction. The projects that survive will be those that can articulate a clear, uncompromising vision of the future. The investors who thrive will be those who are willing to concentrate their capital on that vision, even when it seems irrational. Aschenbrenner's bet on Anthropic is a template for how we should think about governance in the age of AGI. We need to curate our souls, not just our portfolios. We need to ask not "what is the return?" but "what is the belief?"
In the end, the blockchain is a ledger of convictions. Every transaction is a statement of what someone values. Aschenbrenner's transaction is a statement that AGI alignment is the most valuable thing in the world. Whether he is right or wrong, he has given us a gift: a visible, public, and irreversible signal of his deepest belief. That is more than most DAOs can say. Curating the soul in a world of derivative clones.