Anthropic Crosses $47B ARR, Overtakes OpenAI in B2B

Anthropic, the company behind the Claude family of large language models, closed May with an annualised recurring revenue of around forty-seven billion dollars, a figure that, for the first time, put it ahead of OpenAI in the enterprise market. The jump was unusually fast: a research report published this week shows ARR rising from roughly one billion in late 2024 to forty-four billion at the start of May, and then to forty-seven billion by month-end. Wall Street analysts now expect Anthropic to post its first quarterly profit in the second quarter of 2026.

Why the surge? Anthropic has stayed narrowly focused on coding, security, and long-context agent tasks, three areas where enterprise buyers are willing to pay premium prices. More than eighty percent of its revenue now comes from corporate contracts, which tend to be sticky once integrated into a company's workflow. Gross margin has lifted to about seventy percent, helped by aggressive prompt caching and a heavier mix of API business. By contrast, OpenAI's adjusted gross margin has fallen to roughly thirty-three percent, dragged down by the rapid growth of inference costs.

The two rivals are now pursuing very different paths. Anthropic keeps doubling down on the enterprise vertical, betting that deep integration with developer tools and security teams will lock in long-term contracts. OpenAI, with its consumer ChatGPT brand now past one billion mobile users, is leaning into advertising, AI hardware, and a broader model lineup. Investors are watching closely to see which monetisation strategy holds up as the cost of serving frontier models continues to climb.

The bigger lesson is about timing. When frontier models start to look similar on benchmarks, the company that owns the most valuable workflow wins. Coding assistants and security agents are now the clearest examples, and they happen to be the verticals where Anthropic is strongest. Expect the next twelve months to be defined less by raw model scores and more by who controls the entry point that enterprises touch first.