Anthropic's S-1 asks the public to buy shares in a company that a trust of financially disinterested trustees helps govern. The filing calls Anthropic a research lab first, with a mission of the responsible development of advanced AI for the long-term benefit of humanity. Its business is Claude, which it says has seen the fastest enterprise adoption of any technology in history. By September 30, 2026, about 250 customers each paid it $10 million or more a year.

Seven Researchers and a Charter

Seven co-founders started Anthropic in 2021, and it was incorporated as a Delaware public benefit corporation that January. Claude reaches customers through Anthropic's own models, platform and apps, and through all three big clouds: AWS, Google Cloud and Microsoft Azure. Claude Code, its coding agent, plans, writes, tests and debugs software on its own. The company trains and serves its models on four kinds of chips: AWS Trainium, Google TPUs, NVIDIA GPUs and AMD GPUs.

Four Classes of Stock

The governance is the most unusual part of the filing. Public buyers get Class A shares. One Class F share goes to an LLC owned by the seven co-founders and carries 50.1 percent of the votes on most matters. The Long-Term Benefit Trust holds Class T stock, which lets it elect a set of directors, and employees get Class E shares with limited votes. Anthropic filed confidentially on June 1, 2026, then publicly on October 26. The price range is still blank.

Cover: The Beacon, Odilon Redon. Art Institute of Chicago, public domain. Chosen for a beacon (a guiding light, built with care).