The ledger remembers every trembling hand, even when that hand is the founder’s.
Anthropic, the AI safety darling backed by nearly $10 billion in venture capital, just signaled its plan to issue super-voting shares to CEO Dario Amodei ahead of a potential IPO this year. On the surface, this is a standard governance play—founders retain control, investors get a seat at the table. But beneath the legal boilerplate lies a deeper tension: the market is being asked to fund a company that explicitly admits it cannot trust its own shareholders with equal voting power.
Let’s sit with that irony. Anthropic was founded on the premise of building safe, aligned AI. Its core narrative is about preventing runaway systems that act against human interests. Yet when it comes to its own corporate governance, the company is building a structure where a single individual can override the collective will of the market. Logic chains break where greed connects, and in this case, the greed is for control—not just of the company, but of the narrative itself.
Context: Why Now?
Anthropic’s IPO has been rumored since late 2024, with whispers of a $60 billion valuation. The company is positioning itself as the “ethical” alternative to OpenAI, which has a complex governance structure involving a non-profit parent and a capped-profit arm. But Anthropic’s super-voting shares are a different beast entirely. They grant the CEO 10x or 20x voting power per share, effectively neutralizing any shareholder revolt—even if the market disagrees with the company’s direction.
This isn’t unprecedented. Google, Facebook, and Snap all used dual-class structures to keep control with founders. But those companies were consumer platforms, not existential-risk AI labs. The stakes are fundamentally different. When an AI company introduces a governance model that concentrates power, it sends a signal about how it will handle decisions around model release, safety thresholds, and even regulatory compliance.
From my experience auditing token distribution curves during the 2017 ICO mania, I saw how quickly investors overlook governance red flags when the narrative is hot. Bancor and Augur both had poorly designed token-based voting systems that gave insiders outsized influence. Investors bought in because the story was compelling—but the returns eventually reflected the structural flaws. The same pattern is repeating here, wrapped in AI hype instead of ERC-20 tokens.
Core: The Technical Anatomy of Control
Let’s break down what super-voting shares actually mean for the market. In a traditional IPO, each share carries one vote. In a dual-class structure, Class A shares (public) get one vote, while Class B shares (insiders) get ten or more. This means the CEO can control the board with as little as 10% of the economic equity.
For an AI company, this has direct implications for capital allocation, research priorities, and risk management. Consider the following data points:
- According to a 2023 Stanford study, dual-class companies underperform single-class peers by 2-3% annually over the long term, after adjusting for industry. The gap widens during periods of disruption—exactly when AI companies need agility.
- In the crypto world, we’ve seen the same effect with DAO governance tokens. Uniswap’s UNI token, for example, gives holders governance power, but the actual voting is dominated by large wallets and early investors. The result is a system that looks decentralized but behaves like a plutocracy.
- Anthropic’s super-voting shares essentially create a “governance oracle” that cannot be challenged. Any decision—from hiring a new safety researcher to approving a model release—can be made unilaterally.
I’ve run my own analysis on IPO structures using a proprietary LLM-agented scrape of SEC filings. The pattern is clear: companies with super-voting shares tend to have higher volatility in the first 12 months post-IPO, as the market struggles to price in the control premium. For Anthropic, that volatility could be amplified by the AI sector’s inherent unpredictability.
Speed wins the trade, clarity wins the war. Right now, the market has speed—the narrative of an AI IPO is strong—but it lacks clarity on who actually makes the decisions when the model starts behaving unexpectedly.
Contrarian: The Unreported Angle
Most analysts are framing this as a negative: investors get less say, risk increases. But there’s a counter-intuitive argument that Anthropic’s super-voting shares increase market confidence. Here’s why.
First, the AI sector is uniquely dependent on long-term vision. Short-term shareholder pressure has already caused problems at OpenAI—the board’s brief firing of Sam Altman in 2023 was a direct result of governance friction. By centralizing control, Anthropic avoids that chaos. The CEO can make unpopular but necessary safety decisions without worrying about a proxy fight.
Second, institutional investors are increasingly comfortable with dual-class structures for tech companies. The rise of passive investing means many funds buy and hold regardless of governance. The real concern is not about voting power but about valuation. If the market believes Amodei is a better steward of the AI safety narrative than a dispersed shareholder base, the super-voting shares become a feature, not a bug.
Third, consider the crypto parallel. Bitcoin’s governance is deliberately decentralized, but it’s also slow and contentious. Decisions about protocol upgrades take years. In contrast, a centralized entity like Coinbase can deploy new features in weeks. Anthropic is betting that the AI race requires a similar speed of decision-making.
Silence is the only honest metadata. And the silence from institutional investors so far suggests they’re willing to accept this trade-off. They’re betting that a controlled Anthropic is better than a democratized one that might fracture under pressure.
But there’s a blind spot: the correlation between control and accountability. When a company with super-voting shares fails, the founder cannot be easily removed. We saw this with WeWork—Adam Neumann’s control allowed the company to pursue reckless expansion until it collapsed. AI companies have even more potential for catastrophic failure. If Anthropic’s model goes rogue, who fires the CEO? The answer is no one.
Takeaway: The Next Watch
This IPO is a litmus test for the entire AI sector. If Anthropic prices successfully and trades well, expect every AI startup to rush to file with similar super-voting structures. If it stumbles, the market will demand governance reforms before the next big AI IPO.
Infinite leverage, finite patience. The market can tolerate control concentration for only so long before the data catches up. Watch the lockup expiration dates, watch the institutional flow, and watch the SEC’s tone. The real story isn’t the super-voting shares themselves—it’s whether the market has learned anything from the blockchain governance failures of the past decade.
As for me, I’ll be running my signal models on the IPO day. The ledger remembers every trembling hand, and I intend to read it before the crowd does.