Anthropic has priced its initial public offering at US$83 a share, above the projected range, in a record offering that raised US$81.3 billion and reportedly values the company at US$1.5 trillion on a fully diluted basis.
The size of the raise changes the competitive picture. A war chest of that scale funds multi-year compute contracts, which in the frontier model business is the single largest line item and the hardest to secure on short notice.
It also imports a new kind of pressure. Public companies must explain spending quarterly, and investors tend to treat safety research and evaluation work as cost centres unless it can be tied to product differentiation.
Expect the pricing conversation to move to the foreground. Rivals in China are competing aggressively on per-token cost, and a newly public Anthropic has to defend gross margin while matching price cuts.
One detail worth noting for observers of the sector: several early backers who spent years warning publicly about AI risk are among the largest beneficiaries of the listing - a position that will be tested the first time a safety decision costs quarterly earnings.
