Tracing the code back to the conscience behind it.
Leopold Aschenbrenner, the man who wrote the most influential AGI manifesto of the decade, just made a move that screams conviction—and blind faith. He liquidated his entire portfolio of AI infrastructure stocks, including Nvidia and Amazon, and poured everything into a single private company: Anthropic. The reported figure? A staggering $45 billion. But that number smells like a mirage, and the decision itself reveals a deeper tension between the promise of decentralization and the reality of centralized power in the AI race.
Aschenbrenner is not a casual investor. He is the author of Situational Awareness, a tract that argues, with relentless logic, that AGI is coming by 2027–2030 and that it will require trillion-dollar compute clusters. He is a former researcher at OpenAI, a safety-first thinker, and now a hedge fund manager. When he clears his portfolio of the very companies that supply the compute he claims is necessary, you have to ask: What does he see that the market doesn’t? Or is he simply choosing a different kind of centralized bet?
Context: The Man Behind the Bet
Aschenbrenner’s background is a cocktail of technical brilliance and missionary zeal. He left OpenAI in 2021 to write Situational Awareness, a 165-page essay that became the bible of the “AGI near-term” movement. He argues that scaling laws will continue to hold, that we will hit AGI within a few years, and that the geopolitical implications are catastrophic. He then founded a fund to back his beliefs. His first play was AI infrastructure—Nvidia, Amazon, Microsoft—because if AGI is coming, the compute providers win. But now he has reversed course entirely.
Why? Because he believes that the real value in the AGI era will not be in the picks and shovels, but in the single company that builds the first AGI. And he has chosen Anthropic.
Anthropic is the AI company founded by former OpenAI employees who left over safety disagreements. Their flagship model, Claude, competes with GPT-4. But more importantly, Anthropic has built its entire identity around “constitutional AI” and responsible scaling. They prioritize alignment research, interpretability, and safety protocols. For Aschenbrenner, a safety researcher at heart, this is the perfect technical alignment.
Core: The Technical Bet and the Commercial Reality
Every line of code is a hand extended in trust.
From a technical perspective, Aschenbrenner is betting on Anthropic’s ability to execute the “scaling law + safety alignment” dual path. He believes that the company that cracks AGI first will be the one that can maintain safety while scaling. Anthropic’s constitutional AI approach, which uses a set of principles to guide the model’s behavior, is seen as a more robust path than OpenAI’s RLHF (reinforcement learning from human feedback). But is that enough to justify a $45 billion single-stock concentration?
Let’s look at the numbers. The $45 billion figure is suspicious. Aschenbrenner’s fund, as of my last knowledge cutoff in mid-2024, was managing around $1 billion. A $45 billion portfolio would represent a massive jump that is inconsistent with public records. Even if he raised additional capital, that number feels like a leaked exaggeration. But the direction is clear: he is all-in on Anthropic.
Commercialization wise, Anthropic is still unprofitable. They generate revenue through API access and enterprise contracts, but they burn cash on compute and talent. Their valuation has been estimated at $15–20 billion in private rounds. A $45 billion bet would value the company far above any plausible current worth. This suggests that Aschenbrenner is not just investing in the company as it is, but in a future monopoly on AGI. He is betting that Anthropic will become the next trillion-dollar company, and that the public markets (Nvidia, Amazon) are already priced for that future.
But here is the blockchain angle: Aschenbrenner is consolidating power into a single centralized entity. He is an open source evangelist’s nightmare. He is putting all his chips on a company that controls its own models, its own safety protocols, and its own destiny. There is no decentralization here. No open source community. No democratic governance. It is a bet on a benevolent dictator.
And based on my audit experience—having spent four months in 2017 auditing ERC-20 contracts for Cape Town projects—I know that centralized systems have hidden vulnerabilities. The reentrancy bugs I found in those projects were not just code flaws; they were trust flaws. Aschenbrenner may trust Anthropic’s safety team, but trust in a single point of failure is not a robust security architecture.
Contrarian: The Pragmatism Test
Open source is not a license; it is a promise.
Let me offer a contrarian view: Aschenbrenner might be right. If AGI is indeed near, the first mover will capture an enormous share of value. Concentration of capital in the most likely winner is a rational strategy. And Anthropic, with its safety-first approach, might be the most responsible steward of AGI. But the problem is that this model entrenches power. It creates a single point of failure for humanity. If Anthropic’s alignment fails, there is no alternative. There is no decentralized backup.
Aschenbrenner’s move also reveals a blind spot: he sold his infrastructure stocks because he believes the compute providers will not capture the full value of AGI. But that is a short-term view. In the long term, if AGI is built by a single company, that company will be heavily dependent on its compute suppliers. Amazon and Nvidia will still have leverage. By selling them, he is betting that Anthropic will become vertically integrated or that it will switch to alternative compute (TPU, Trainium). But that is a risky bet.
Furthermore, the absence of any open source component in his strategy is telling. The AI community is increasingly divided between closed-source companies (OpenAI, Anthropic) and open source efforts (Llama, Mistral). Aschenbrenner, despite his background in safety research, has chosen the closed path. This is a betrayal of the decentralization ethos that many of us in the blockchain space hold dear.
Takeaway: The Vision Forward
We build bridges, not just blocks, between people.
Leopold Aschenbrenner’s concentration of his portfolio into Anthropic is a powerful signal. It tells us that the smartest money in AGI is betting on a centralized, safety-first, closed-source future. But for those of us who believe in sovereignty, in open protocols, in education as the only true decentralized currency, this is a wake-up call. We need to build alternative paths. We need to fund decentralized AI research, open source models, and community-governed compute.
The crypto world has spent years building decentralized finance. Now we need to build decentralized intelligence. Because code without conscience is just chaos, and conscience without code is just a dream. The future of AGI should not be owned by one company. It should be built by all of us.
Let me end with a question: If AGI is coming, who will hold the keys? Not the keys to the model, but the keys to the governance? Aschenbrenner has chosen his answer. It is time for the blockchain community to choose ours.
