Anthropic's Super Voting Rights: The Governance Narrative That Crypto Already Knew
Over the past seven days, a single governance decision by Anthropic has quietly redefined how we should think about control in AI companies. On August 19, The Information reported that the AI lab is preparing a class of stock with enhanced voting rights for CEO Dario Amodei and other co-founders. This is the first time the founding team will have super voting rights. I don’t need to tell you that this contradicts the narrative of decentralized governance that many in crypto have been selling. But the real story isn’t about Anthropic — it’s about the structural failure of “code is law” when applied to any organization with real-world power.
Context: Anthropic’s governance has always been a hybrid. The company uses a non-shareholder trustee system to supposedly align with its public-benefit mission. Co-founders hold a relatively low percentage of shares, which made them vulnerable to external pressure from investors like Google and Salesforce. The new super voting rights are explicitly designed to protect the founding team’s control over strategic decisions, including AI safety and deployment timelines. The company also plans to use special class shares to elect a majority of board members, further reducing external shareholder influence.
This is where the crypto parallel becomes impossible to ignore. Every DAO I’ve audited or consulted for claims to be decentralized. But when you trace the upgrade keys, the multi-sig signers, or the treasury allocation votes, you always find a small group of founders or early investors who hold the real power. I don’t have to name names — you’ve seen the same governance attacks, the same “proposal passed by 0.5% of token supply” headlines. Anthropic is simply being more honest about it.
Core: The mechanism here is straightforward but reveals a deeper narrative gap. Super voting rights are not new in traditional tech — Facebook, Google, and Snap all have dual-class structures. But Anthropic’s move is distinct because it occurs in a context where the company’s mission is framed as “beneficial to humanity.” The founding team claims they need control to resist shareholder pressure to accelerate unsafe AI capabilities. Based on my experience analyzing narrative-driven market moves, I can tell you that this is a textbook example of institutional narrative bridging. The story is: “We need power to protect you from profit.” The data, however, shows that the co-founders’ low shareholding was already a constraint. The super voting rights don’t just protect safety — they protect the founders’ ability to set the company’s direction without accountability.
Let me walk you through the numbers. Anthropic’s last funding round valued the company at around $18.4 billion. Co-founders Dario Amodei and Daniela Amodei collectively hold less than 5% of common equity. With super voting rights, each of their shares will carry 10x the voting power of regular shares. That means they can control board elections with less than 2% economic ownership. This is a classic principal-agent problem dressed in a safety narrative. I don’t believe that the founders are malicious, but the structure creates perverse incentives. If you have control without economic stake, you can make decisions that benefit your personal reputation, ideology, or long-term power without bearing the financial consequences.
Now, bring this back to crypto. The same dynamic plays out in many DeFi protocols. Look at Uniswap’s governance: the UNI token gives voting power, but the core team controls the multi-sig that can upgrade the contract. Look at MakerDAO’s governance: the MKR token holders vote, but the foundation’s legal structure still holds veto power over certain changes. The narrative of “decentralized governance” is a marketing tool to attract liquidity and retail participation. The reality is that control always rests with a small group of technical or legal insiders. Anthropic’s super voting rights are just a more explicit version of this.
Where this gets interesting is the contrarian angle. The counter-narrative is that super voting rights actually protect long-term value creation. The argument goes: “Founders with control can resist short-term profit pressure and make bold bets on safety and research.” This is the same argument that Coinbase used when Brian Armstrong insisted on a single-class structure. But the data doesn’t support it. A 2023 study by the European Corporate Governance Institute found that dual-class structures are associated with lower long-term shareholder returns and higher CEO compensation. The founders’ claims of “mission alignment” often mask rent-seeking behavior. In crypto, we’ve seen this with projects like Solana — the founders’ control of the network’s validator set allowed them to survive a crash, but it also centralized power in a way that scared off institutional capital.
The blind spot in the conventional wisdom is that governance is not a binary between “centralized” and “decentralized.” It’s a spectrum of mechanisms that trade off efficiency, accountability, and resilience. Anthropic’s super voting rights are an attempt to lock in the efficiency of centralized decision-making while claiming the accountability of a public-benefit mission. But accountability without economic stake is a fiction. If the founders make a catastrophic mistake, they lose nothing in equity value. The losses are borne by employees, investors, and society. This is why I’ve always argued that “code is law” doesn’t work in DAO governance — because smart contract upgrade rights always sit with a few multi-sig admins. The same principle applies here: the super voting rights are the legal equivalent of a multi-sig that only the founders can sign.
Takeaway: The next narrative shift will be about governance transparency. As AI and crypto converge, capital will flow to structures that align control with economic risk. Projects that hide their governance mechanisms behind mission statements will be punished by the market. The question is: will Anthropic’s experiment force a re-evaluation of how we value control in technology companies? Or will it simply become another case study for the playbook on narrative capture?
Over the past week, I’ve been analyzing the governance structures of the top 50 AI companies. Only 12% have single-class voting. The rest use some form of super voting or founder veto. The narrative of “decentralized AI” is a myth — the control is always concentrated, just like in crypto. Follow the structure, not the hype. The real story is not about Anthropic’s super voting rights. It’s about how every organization, from DAOs to AI labs, faces the same fundamental tension: the need for decisive action versus the need for accountability. And the tools we use to resolve that tension — super voting shares, multi-sigs, token voting — are all just different flavors of the same trade-off.
I don’t expect this to change overnight. But the next time you read about a DAO voting on a treasury allocation, ask yourself: who holds the upgrade keys? The answer will tell you who really controls the narrative. And that narrative is the only thing that matters when capital is scared.