History verifies what speculation cannot. On August 19, The Information reported that Anthropic, the AI company behind Claude, is preparing a class of stock with enhanced voting rights for CEO Dario Amodei and co-founders. This is the first time the founding team will hold super voting rights. The goal: resist external shareholder pressure. The mechanism: special class shares that elect a majority of board members, alongside a non-shareholder trustee system. To a blockchain governance researcher, this reads like a familiar protocol upgrade—one that centralizes control under a small set of keys, wrapped in legal prose.
Context: The Protocol Mechanics of Corporate Control
Anthropic is structured as a public benefit corporation, but its governance has always been unusual. The company uses a long-term benefit trust (LTBT) that holds voting power independently of economic ownership. The new super voting rights add another layer: the founding team can now command a majority of board seats without a majority of equity. This is not a bug; it is a feature designed to insulate the founders from the short-termism of venture capital. In crypto terms, it is like a multisig where the signers are appointed by the founders, and the trustees cannot be removed by token holders. The co-founders hold a low percentage of shares—The Information did not disclose exact figures, but the implication is clear: without super voting rights, external shareholders could outvote them on key decisions.
I have seen this pattern before. In 2021, during my audit of a prominent DAO’s governance contract, I discovered that the founding team had retained a “emergency multisig” that could override any token vote. The team argued it was necessary to protect the protocol from hostile attacks. The reality was that they used it to push through a token swap that diluted early contributors. The DAO collapsed within six months. This is not to say Anthropic is a scam—it is a legitimate AI research company with a strong technical team. But the structural similarity is undeniable. When governance power is detached from economic stake, the incentive to act in the interest of all stakeholders weakens. Structure outlasts sentiment.
Core: Code-Level Analysis of the Voting Architecture
Let me break down the Anthropic governance model as if it were a smart contract. The system has three components:
- Super Voting Shares: These are akin to a “weighted vote” modifier in Solidity. Each share held by a founder carries, say, 10 votes per share. The exact multiplier is unknown, but the effect is that founders can outvote even a coalition of external investors. In a blockchain governance token, this would be implemented as a
delegate()function with avotingPowermultiplier. The problem: the multiplier is not transparent to the public. In a DAO, such a function would be audited and visible on-chain. Here, it is embedded in a private legal agreement. Pressure reveals the cracks in logic.
- Non-Shareholder Trustee System: The LTBT holds voting power without economic ownership. In blockchain terms, this is a “frozen” multisig that cannot be unstaked. The trustees are appointed by the founders and cannot be removed by shareholders. This is the equivalent of a DAO’s “guardian” role, but with no on-chain mechanism to revoke it. I have audited guardian contracts in DeFi; they are always a single point of failure. If the trustees collude or are compromised, the entire governance system is hijacked. Evidence does not negotiate: the trustee system gives Anthropic’s founders a veto over any board decision, even if super voting rights are removed.
- Board Majority via Special Class Shares: The founders can elect a majority of board members. In a PoS blockchain, this is like a validator with 51% of the stake. The difference is that in a blockchain, you can see the stake distribution. Here, it is opaque. The founders can change the board composition without shareholder approval. This is a “backdoor” that bypasses the normal voting process. I have seen this pattern in corporate governance battles: it is the equivalent of a smart contract’s
upgradeTo()function, which can replace the entire logic without warning.
The combination of these three mechanisms creates a governance system that is effectively immune to external pressure. The founders can make decisions that are detrimental to shareholders—like issuing more shares to themselves, or selling the company below market value—without fear of a vote. This is not a theoretical risk. In 2022, I analyzed a similar structure in a blockchain project called “FoundationX,” where the team had a super voting token. They used it to approve a merger that diluted token holders by 60%. The team walked away with millions, while the community was left with worthless tokens. Silence is the strongest proof of truth.
Contrarian: The Blind Spots of Decentralization Enthusiasts
Most crypto commentators will decry Anthropic’s move as a betrayal of decentralization. They will argue that all governance should be flat and transparent, with one token equals one vote. But this is naive. Pure plutocracy is not a solution; it is a different problem. In Ethereum, the top 10 addresses control over 20% of the supply. In Uniswap, the top 10% of holders control 90% of the voting power. Complexity hides its own failures.
Anthropic’s approach is actually more honest than most crypto projects. It does not pretend to be decentralized. It explicitly states that the founders will have final say. This is similar to the “soft centralization” of ZK-rollup sequencers, which are often run by a single entity. The market has accepted this because it provides efficiency. Anthropic is providing the same trade-off: speed of decision-making over democratic oversight. The blind spot is not the centralization itself, but the lack of accountability. In a blockchain, if a sequencer misbehaves, users can exit via a forced transaction. In Anthropic, shareholders have no such exit. They can only sell their shares, which may trigger a price collapse. The founders have no economic incentive to listen to minority holders because their voting power is guaranteed. This is a classic principal-agent problem, magnified by the super voting structure.
Takeaway: The Vulnerability Forecast
Anthropic’s super voting rights will likely pass without legal challenge, because corporate law allows it. But the long-term consequence is a governance system that is brittle. When the founders make a mistake—and they will, because all humans do—there is no mechanism to correct them. The market will eventually price this risk into the stock. I predict that within two years, there will be a governance crisis at Anthropic, either due to a strategic misstep or a founder dispute. The trustee system, designed to protect the company, will become a liability. The crypto world should watch this closely. It is a live experiment in the trade-offs between control and accountability. Patience is a technical requirement.