The companies building the agent economy are going public with a governance structure that lets founders keep the keys even after selling the house.

The Summary

The Signal

Anthropic is preparing to go public with a governance model that's become standard in AI: founders keep control, public shareholders get economic exposure but no power. Dual-class stock structures give company founders super-voting shares, often with 10 votes per share compared to 1 vote for ordinary shareholders. It's how Larry Page and Sergey Brin controlled Google with less than 50% ownership. It's how Elon Musk maintained control of SpaceX through multiple funding rounds.

Now it's the expected structure for AI companies going public. The logic: AI development requires long-term thinking that quarterly earnings calls and activist investors would destroy. The counterargument: unchecked founder control has historically led to terrible decisions at public companies, from media conglomerates to tech platforms.

"The AI boom is eroding the principle of 'one share, one vote' that checks the power of company founders."

Here's the financial wrinkle. Anthropic's IPO will generate billions in demand from index funds, which are required to buy shares of companies added to major benchmarks like the S&P 500 or Nasdaq-100. But Anthropic faces a longer wait than SpaceX did to join these indexes because index providers have tightened their rules around dual-class shares since SpaceX went public.

The delay matters for three reasons:

  • Index inclusion creates automatic buying pressure, lifting the stock price
  • Passive funds hold roughly 40% of U.S. public equities, making index inclusion a liquidity event
  • Companies excluded from indexes face higher volatility and lower institutional ownership

The timing gap gives Anthropic founders an extended period where they control a public company with less external pressure from the index-fund complex. That's either a feature or a bug depending on whether you think Dario Amodei should have unilateral control over one of the three companies racing toward AGI.

The Implication

If you're buying shares in the next wave of AI IPOs, understand what you're actually buying: economic exposure without governance rights. You own a piece of the upside, but founder decisions, good or catastrophic, happen without your input. For agents-focused companies, this matters more than traditional tech. The gap between a well-aligned AI system and a misaligned one isn't a product miss. It's existential.

Watch how Anthropic structures its dual-class terms. If they include sunset provisions that convert super-voting shares to regular shares after a set period, that's a signal they view founder control as temporary scaffolding, not permanent architecture. If there's no sunset, they're telling you this is how it stays.

Sources

Bloomberg Tech