Anthropic’s 7 Founders Want More Votes Before an IPO. Investors Should Worry.
Anthropic wants public money while seven founders seek outsized control. That can protect a mission — or leave investors funding management without a real say.
Anthropic wants public money while seven founders seek outsized control. That can protect a mission — or leave investors funding management without a real say.
Reports this month say CEO Dario Amodei and Anthropic’s other six co-founders are preparing to receive shares with extra voting power ahead of a possible IPO. Put bluntly: the people building one of the world’s most consequential AI companies appear to be asking future shareholders to fund the machine while accepting a smaller say over where it goes.
I understand why they want it. I’m still not convinced they’ve earned the blank cheque.
This is not just another tech-company governance tweak
The reported plan is for Anthropic to issue a new class of shares with enhanced voting rights to Dario Amodei and the other founders. The stated purpose, according to reporting by The Information and Reuters, is to insulate the founders from outside shareholder pressure as the company approaches a potential public listing.
That phrase — “outside shareholder pressure” — does a lot of heavy lifting.
Outside shareholder pressure can mean activists demanding a quick profit at the expense of long-term product work. Fair enough. It can also mean investors asking why the company is burning cash, taking reputational risks, making grand political claims, or allowing a small group of founders to retain power long after their economic stake has been diluted.
Those are not the same thing. One is short-termism. The other is accountability.
Anthropic already has an unusual structure. It is a public benefit corporation, and its Long-Term Benefit Trust has a formal role in governance. Anthropic says the trust exists alongside stockholders in electing the board, with the company’s purpose being responsible AI development for the long-term benefit of humanity.
That was already a serious attempt to build something other than the standard “maximise quarterly earnings and call it leadership” corporate model. The new reported super-voting plan would add another layer: founder control.
A trust with a mission is one thing. A trust plus a board plus super-voting founders is another. At some point, you are not creating resilient governance. You are creating a fortress.
Dario Amodei is trying to solve a real problem
Let’s be fair before we get carried away with the pitchforks.
Frontier AI is not a normal business. If Anthropic’s systems become more capable, the decisions made by its executives and board may affect cybersecurity, employment, education, national security, intellectual property and who controls essential infrastructure. A normal public-company incentive system is poorly designed for that.
Public markets are brilliant at imposing discipline on businesses that sell widgets, mortgages, beer or logistics software. They are less brilliant at rewarding a CEO who says: “We could ship this product sooner, but we won’t because the risk is too high.”
That is precisely why founder control appeals to companies such as Anthropic. Management wants room to make unpopular decisions without a hedge fund manager ringing up in six months demanding a margin expansion plan.
Amodei has publicly argued that powerful AI systems need stronger safeguards and that governments should be able to block dangerous deployments. Anthropic’s pitch has long been that it is trying to build capable AI without becoming reckless in the race against OpenAI, Google and everyone else throwing billions at the problem.
There is a legitimate leadership argument here: if your mission matters, do not hand the steering wheel to the most impatient bidder.
But here is the bit founders hate hearing: a noble mission does not make you permanently right.
The overlooked risk is not control. It is uncorrectable control.
Founder control is fashionable because people remember the winners. Mark Zuckerberg retained voting control at Meta and built an extraordinary business. Larry Page and Sergey Brin retained control at Alphabet. Plenty of investors made fortunes anyway.
What gets ignored is survivorship bias. We remember the controlled companies that became giants, not every founder who kept power too long, surrounded themselves with believers, missed a market change, backed the wrong lieutenant or confused personal conviction with strategy.
Every founder says they want protection from short-term pressure. Very few say, “I’d also like protection from being wrong.” Yet that is often what concentrated control delivers.
The best boards do not exist to clap for the CEO. They exist to make sure the CEO does not drift into a private reality where every objection is ignorance, disloyalty or a failure to understand the mission.
Anthropic’s proposed setup deserves extra scrutiny because its founders reportedly hold relatively small economic stakes compared with founders at some earlier-generation tech companies. If that is true, then the gap between who bears the economic risk and who holds the voting control could become very large after an IPO.
That is the deal future investors need to assess honestly. They may own the financial downside, while seven founders hold disproportionate power over leadership, board composition and strategic direction.
There is nothing illegal or automatically immoral about that. But don’t dress it up as democratic capitalism. It is a controlled company. Call it what it is.
The contradiction Anthropic now has to explain
Anthropic’s public identity is built around safety, restraint and long-term responsibility. Those are serious ideas. They are also easy words to use when you are asking for exceptional authority.
The uncomfortable question is simple: who decides whether management is actually acting responsibly?
If the answer is “the founders,” that is not governance. That is faith.
If the answer is “the board,” then investors should want to know exactly how independent that board remains when founders have extra voting rights and a long-term trust has substantial influence over director selection.
If the answer is “the market,” then super-voting shares deliberately weaken the market’s ability to intervene.
And if the answer is “government,” well, that introduces a different set of problems entirely — especially for a company selling powerful AI to major enterprises and operating in politically charged sectors.
I’m not saying Anthropic should be run by quarterly earnings tourists. I am saying safety claims should require more accountability, not less.
The more consequential the company’s technology becomes, the less acceptable it is for leaders to say, “Trust us, we are the adults in the room.” History is packed with adults in rooms making expensive mistakes.
Why this matters far beyond Anthropic
This is the IPO template to watch.
If Anthropic succeeds in going public with a fortified governance structure, every high-growth AI company will take notes. Founders will argue that advanced technology is too important to be left vulnerable to ordinary shareholder democracy. They will say speed, safety, national competition and long-term research justify extra control.
Some of that will be true. Some of it will be a convenient excuse for founders who never want to be challenged.
The second-order effect is bigger than one cap table. It is about whether public investors are buying ownership or simply providing liquidity and capital to an elite management class.
That changes how you value a company. A share with limited influence over governance is not identical to a share with meaningful voting power, even if both receive the same economic dividends on paper.
Investors need to stop treating governance as a boring appendix at the back of a prospectus. It is the operating system for every future decision: acquisitions, executive pay, risk appetite, succession, layoffs, capital allocation and what happens when the founder loses the plot.
I have watched enough businesses grow to know this: the biggest management risk is rarely incompetence on day one. It is insulation over time.
What this means for you
If you are an investor, founder or operator, take three practical lessons from Anthropic’s move.
First, read the control structure before you get excited about the product. Ask who elects directors, what voting rights each share class carries, whether those rights expire, and what genuinely independent checks exist when management is wrong.
Second, if you are building a company, do not confuse mission protection with founder worship. You can protect long-term thinking through clear board mandates, staged voting rights, independent directors and sunset clauses. You do not need permanent immunity from challenge.
Third, if you are raising money, be brutally clear about the bargain. If you want control, say it. Explain why it is necessary, what limits apply and how investors are protected. Sophisticated capital can accept an unusual structure. What it will not forgive is discovering later that “mission-led governance” was code for “the founders answer to nobody.”
Anthropic may be right to protect its ability to make hard, long-term decisions. But the company should remember the basic rule of leadership: power is easiest to justify before you have it.
The real test is what you do once nobody can take it away.