When Bill Ackman, the activist investor who once called Bitcoin a “bubble,” repositions his portfolio to make Amazon his fourth-largest holding while dumping Alphabet entirely, the crypto community should pause. Not because this signals a bullish spin on Amazon stock—that’s obvious—but because the underlying logic reveals a preference that threatens the very premise of decentralized AI: the belief that centralized platforms can offer model neutrality without the need for permissionless infrastructure.
I’ve spent the last decade dissecting the moral architecture of trust in code. In 2018, I audited a fledgling DeFi protocol called EtherTrust and found a reentrancy vulnerability that would have drained $200,000. That experience taught me that trust is not a credential—it’s a fragile property of systems that must be verified at every layer. Now, as I watch a traditional value investor bet on Amazon’s AI monetization pathway, I see the same pattern: the market is choosing centralized model neutrality over decentralized sovereignty, and the implications for blockchain’s AI narrative are profound.
Context: The Hidden Signal in Ackman’s 13F
Pershing Square’s quarterly 13F filing showed Amazon moved to its fourth-largest holding while Alphabet (Google’s parent) was sold. The immediate narrative in financial media was “AI monetization clarity”—AWS’s pay-per-token model for AI inference is a proven revenue engine, while Alphabet’s search-ad business faces structural disruption from conversational AI. But there’s a deeper layer: the choice between model neutrality and model ownership.
Amazon Web Services (AWS) offers a “bring your own model” approach through Bedrock, allowing customers to run Anthropic, Meta, or even open-source models on its infrastructure. Alphabet’s Vertex AI, by contrast, heavily promotes its own Gemini models. In a world where enterprises fear vendor lock-in, AWS’s strategy is a centralized version of the open-multichain vision that blockchain advocates have long promoted. Ackman is betting that this model-friendly stance will win the cloud AI war—and the market is rewarding that bet.
Core Insight: The Irony of Centralized Openness
Here’s the uncomfortable truth for decentralized AI proponents: AWS is effectively acting as a “permissionless” platform for AI models, but it’s permissionless within a centralized, censored, and surveilled environment. It’s the same paradox I uncovered in 2021 when I traced the NFT metadata of CryptoSculptures to centralized servers—the promise of permanence was an illusion. Today, that illusion is being sold as “model neutrality.”
During the 2020 DeFi Summer, I facilitated community discussions for a lending protocol called LendPool, witnessing how truly permissionless finance empowered the unbanked. But the subsequent greed and wash trading revealed a darker truth: permissionless systems attract speculation, not just liberation. Similarly, AWS’s model neutrality may attract enterprise adoption, but it does not solve the core problems of censorship, data sovereignty, or algorithmic bias. It merely offers a better UX for the same centralized structure.
Ackman’s bet is a vote for efficiency over sovereignty. And that vote is rational—AWS’s AI revenue is already exceeding $100 billion annualized, while decentralized AI protocols like Bittensor or Render Network are still battling for adoption. The institutional capital is flowing to the path of least friction, not the path of maximum freedom.
Contrarian Angle: The Uncomfortable Case for Centralized Model Neutrality
Let me play the devil’s advocate. If the market is rewarding a centralized platform that offers model neutrality, then what is the unique value proposition of decentralized AI? The blockchain community often argues that “trustless” systems are inherently superior, but trustlessness is a feature that only matters when the alternative is broken. Amazon’s AWS has a 32% market share in cloud, a 99.99% uptime SLA, and a multi-billion-dollar compliance apparatus. The “trust” in Amazon is not blind—it’s backed by contracts, insurance, and regulatory oversight.
During my bear market solitude in 2022, I taught blockchain fundamentals to underprivileged teenagers in Milan. I realized that decentralization’s true value is not in efficiency but in equity—giving agency to those who cannot access centralized systems. But for a hedge fund manager like Ackman, efficiency is the only metric. And in that metric, AWS’s centralized model neutrality is winning.
This is the blind spot in the crypto narrative: we assume that “open” is always better than “closed,” but the market is showing that “open within a closed system” can be good enough for enterprise. The question is whether that “good enough” will suffocate the demand for genuine decentralization.
Takeaway: A Call for Sovereign AI
I’m a born idealist, but I’ve learned that ideals without evidence are fantasies. The Pershing Square signal is a warning: if decentralized AI cannot offer a clear differentiation beyond “not centralized,” it will remain a niche. The killer app for decentralized AI is not model neutrality—that’s already commoditized by AWS. The killer app is sovereign identity—the ability to prove that a human, not a bot, is making the inference request, and that the data remains private. This is the “Proof of Soul” I wrote about in my 2026 manifesto.
Ackman’s bet is a stress test for the crypto community. Will we respond with hand-wringing about the evils of centralization, or will we build infrastructure that makes sovereignty as convenient as AWS’s model neutrality? The answer will not come from a 13F filing—it will come from the code we write and the values we encode. The ghost in the machine is still waiting for a soul.
