Hook
Last week, Anthropic filed its S-1 draft with a governance clause that sent shockwaves through both the AI and crypto communities: CEO Dario Amodei will hold supervoting shares, controlling 60% of voting power post-IPO. The move is classic Big Tech paternalism—think Zuckerberg's dual-class structure at Meta. But for those of us who have spent years building decentralized governance systems, it raises a deeper question: Is concentrated power always the enemy of mission-driven organizations, or can it sometimes be the only way to protect a vision from the short-term whims of markets?
Context
Anthropic, the AI safety company behind Claude, is preparing for what could be the most consequential tech IPO of 2025. The company's charter explicitly prioritizes responsible AI development over shareholder returns. Super-voting shares, in this context, are designed to insulate the CEO from activist investors who might demand faster commercialization or relaxation of safety protocols. This is not a new mechanism—Google, Facebook, and Snap all used similar structures to preserve founder control. What makes Anthropic's case unique is the intersection of high-stakes AI governance and the growing influence of blockchain-based governance models that reject such centralization outright.
In the crypto world, we have built entire ecosystems around the principle that no single entity should hold disproportionate power. DAOs, quadratic voting, and token-weighted decision-making are all attempts to distribute authority. But after years of designing and observing these systems—from my work on UnityDAO's quadratic voting framework to the Human-First Protocols initiative—I've come to a uncomfortable conclusion: the dogma of decentralization can sometimes be just as dangerous as the dogma of centralization.
Core
Let me start with a hard truth from my own experience. In 2020, I co-designed the governance structure for UnityDAO, a $5 million treasury managed by 3,000 members. We implemented quadratic voting to prevent whale dominance, and for a while, it worked. Participation increased by 300% compared to industry averages. But we also faced a persistent problem: when the market turned volatile, voter turnout dropped below 5%, and the few who remained were often the most financially motivated—not the most aligned with the mission. We ended up with governance by apathy, not by community.
Code without compassion is cold. That's a phrase I've used in my workshops since 2017, when I started Ethical Ledger to teach retail investors about smart contract risks. The compassion is not just about user experience—it's about recognizing that governance structures are human systems first, technical systems second. Super-voting shares, despite their anti-democratic appearance, may actually be a form of compassion: they protect the long-term mission from the cold calculus of quarterly earnings.
Anthropic's situation is a stark contrast to the crypto world. In most DAOs, the 'community' is a misnomer—voter turnout is below 5%, and whales or VCs pull the strings behind the curtain. On-chain governance is often a theater of participation. The illusion of decentralization masks the same concentration of power, but without the accountability that comes with a named leader. At least with Amodei, we know who is responsible. In a DAO, when a proposal passes that drains the treasury, everyone blames the smart contract.
From a technical perspective, super-voting shares are a form of 'emergency brake' governance. They allow the CEO to override shareholder votes on critical issues—like AI safety protocols or acquisition decisions. This is analogous to the 'multisig with timelock' mechanism used in many DeFi protocols, where a core team has the power to pause contracts in case of an exploit. The difference is transparency: Anthropic's structure is disclosed in the S-1, while many DAOs hide their centralization behind a veneer of smart contract code.
Code without compassion is cold. In my 2022 Rebuild Chicago initiative, I saw how the bear market destroyed communities that had no human fallback. When FTX collapsed, the 'code is law' crowd was silent. The people who suffered were not the whales—they were the retail investors who trusted the system. Anthropic's super-voting shares are a bet that human judgment, even when concentrated, is better than the cold, indifferent logic of markets or code.
Contrarian Angle
But here is where the crypto perspective adds value: concentrated power is only safe if the leader's incentives are aligned with the mission. Amodei's reputation as an AI safety advocate is the only guarantee. If he ever sells his shares or changes his mind, the super-voting structure becomes a trap. In the crypto world, we have seen this pattern repeatedly—founders who start with noble intentions and later abandon the project, leaving token holders with nothing.
My work on the 'Values First' coalition in 2025 taught me that even the most well-intentioned centralized power needs checks and balances. We negotiated a $10 million grant from BlackRock's venture arm, conditioned on their adoption of transparency protocols. The lesson was clear: trust, but verify. Anthropic's IPO prospectus needs to include explicit mechanisms for accountability—like a sunset clause on the super-voting shares, or a board veto on existential decisions. Without that, the structure is just a monarchy in corporate clothing.
From a market perspective, the IPO's success will depend on whether institutional investors accept this trade-off. The crypto community has historically rejected dual-class shares, but we are also watching the rise of 'liquid democracy' experiments that try to balance delegation with direct participation. Perhaps the real innovation is not in the voting mechanism itself, but in the cultural norms that surround it. Anthropic's bet is that the market will reward a clear, accountable leadership structure over a messy, opaque one—even if that means less decentralization.
Takeaway
As I write this, I am reminded of a conversation I had during the 2026 Human-First Protocols initiative. A young developer asked me why we needed a manual verification layer for AI-generated proposals. 'Can't we just trust the algorithm?' he asked. I told him: Code without compassion is cold. The algorithm cannot feel the weight of a decision that affects thousands of people's livelihoods. It cannot understand the nuance of a moral trade-off.
Anthropic's super-voting shares are a bet on human judgment. Whether that bet pays off depends on the integrity of the person holding the power—and on the willingness of the market to accept that some decisions are too important to be left to votes. For the crypto world, the lesson is not to abandon decentralization, but to recognize that it is a means, not an end. The goal is not the distribution of power, but the alignment of power with purpose. If we forget that, we are just building more efficient systems for making the wrong decisions.