Anthropic, the company behind the Claude model family, has disclosed preliminary financials ahead of a public listing expected as early as November, positioning itself as the first profitable frontier AI lab to reach the market.
The numbers behind the filing
Revenue surged more than tenfold in 2025 to $4.6 billion, though a $42 billion net loss was dominated by a $34 billion accounting charge tied to the revaluation of convertible shares. The operating loss came in at $8.06 billion, with $7.33 billion directed to compute and infrastructure.
A profitable quarter changes the comparison
In the second quarter of 2026 the company recorded $11.5 billion in revenue, a fourteen-fold increase from the same period a year earlier, and an adjusted operating profit of $559 million. OpenAI, by contrast, posted $6.7 billion in revenue and a $12.3 billion operating loss for the quarter. SpaceX, which listed earlier this year, was unprofitable in its second quarter, though analysts project a turn in the current period.
Valuation math and what it signals
Anthropic is targeting a $2 trillion valuation, roughly in line with SpaceX's $1.75 trillion debut, but at 45 times run-rate sales versus the roughly 100 times sales SpaceX commanded. The gap reflects both a faster growth rate, second-quarter revenue rose more than 1,000 percent year over year, and the rarity of a profitable model developer at this scale.
The safety narrative and market implications
Chief executive Dario Amodei has publicly advocated slowing AI development to address safety risks, a stance that may temper the retail enthusiasm that surrounded Elon Musk's SpaceX offering. A successful debut at the targeted valuation would nonetheless reinforce confidence in the AI infrastructure chain, particularly semiconductor suppliers such as Nvidia, by demonstrating that the largest model builders can sustain the capital intensity required to stay at the frontier.