A sharp decline on a trading screen, rather than a press release, was the first indication that something had gone wrong. Within hours of the company’s initial blog post earlier this month, funds marketing exposure to Anthropic shares decreased—a move that traders notice before they comprehend. Brokers in New York and London were already on the phone, attempting to determine whether the shares in their books still had any significance, by the time the majority of people had actually read the notice.
The post itself was remarkably straightforward. Anthropic identified eight secondary platforms and stated that any stock sales made through them would be null and void and would not be recorded in the company’s financial records. It was applicable to both common and preferred stock. Rather than hiding behind the legalese that every Silicon Valley startup stuffs into its shareholder agreements, no significant AI company had ever done this so openly, naming names.
The list then quietly shrank. Half of the names vanished. The only companies left are Upmarket, Pachamama, Unicorns Exchange, and Open Door Partners. The most well-known brand in private market trading, Hiive, had vanished. There was only an edit, not a formal apology or an explanation. It’s the kind of action that raises concerns, particularly for a business in early IPO negotiations with Wall Street’s largest banks.
The CEO of Hiive, Sim Desai, didn’t let the occasion go unnoticed. He stated on LinkedIn that his platform does not allow share transfers “without the company’s approval,” and following the removal, he added a more direct statement: if Anthropic had contacted him prior to going public, the two could have sent a cohesive message. They didn’t. The subtext was clear, but the criticism was measured. A business that is about to go public shouldn’t be picking fights with the very brokers that provide its employees with liquidity.
Anthropic seems to have overestimated how well it could manage its own cap table. Transfer restrictions are a type of fine print that buyers ignored when pursuing pre-IPO exposure, and both Anthropic and OpenAI have long included them in their shareholder agreements. That neglected language became a market event when certain platforms were given names. The legal point landed later than the damage spread.
It’s even stranger because of the timing. Anthropic raised its valuation to $965 billion and finally surpassed OpenAI on Thursday when it announced the largest private funding round in its history, totaling $65 billion. In February, the Series G closed at $380 billion. Secondary markets were quoting it close to a trillion by April. The awkward question of whether the warning was actually about price discovery or about who gets to participate is raised by the fact that the new round is almost exactly where those private quotes had landed.
Most likely both. In order to keep engineers whose compensation is largely dependent on equity, Anthropic needs secondary liquidity. Prior to a potential October listing headed by Goldman Sachs, JPMorgan, and Morgan Stanley, it also requires governance discipline. It’s challenging to balance those two demands simultaneously. It is even more difficult to do it through a blog post that is edited a week later.
The contradiction is difficult to ignore. Despite raising more money than any private company in history, the company is still unable to decide who is permitted to sell its shares. When Bloomberg asked Anthropic for comment, the company did not reply. The four remaining platforms have not said anything. The answers are now available to investors who purchased through the names that were removed. The remaining four shareholders are left in the dark, which seems like the wrong kind of suspense for a company this close to going public.⁖※
