Anthropic

What Anthropic does

Anthropic develops the Claude family of AI models. It sells them to businesses, developers and consumers through its apps, the Claude Code programming tool and a developer API. Its model lines run from Haiku, the smallest, to Fable and Mythos, which it calls its most advanced. Former OpenAI employees founded it in 2021. Co-founder Dario Amodei leads it. It is a public benefit corporation whose board is elected by stockholders and its Long-Term Benefit Trust, according to Anthropic.

Why it matters

TechCrunch described Anthropic and OpenAI in June 2026 as the two largest AI labs. Anthropic raised $65 billion in May 2026 at a $965 billion post-money valuation. It said its annualized run-rate revenue had passed $47 billion. An IPO prospectus seen by Reuters shows 2025 revenue of nearly $4.6 billion, up 12-fold, and an operating loss of $8.06 billion. The net loss was about $42 billion, including a roughly $34 billion accounting charge. The filing puts planned spending on cloud, computing and infrastructure commitments at $518 billion over the coming years. It says nearly a quarter of 2025 revenue came from two customers.

Recent developments

Models. Anthropic previewed Mythos on April 7, 2026, limiting it to partner organizations doing defensive security work. It released Fable 5, a public version with added safety restrictions, on June 9. Claude Opus 5.5 followed on September 22 and Sonnet 5.5 on September 28, according to Anthropic's newsroom.

Government disputes. Anthropic's prospectus says that in February 2026 the president directed federal agencies to stop using its models and the Department of Defense labeled the company a supply-chain risk, Reuters reported. TechCrunch reported the dispute arose from limits Anthropic set against using Claude for fully autonomous weapons and mass surveillance of Americans. The Pentagon said it would use the models only for lawful purposes. A federal judge in California ruled the designation unlawful on August 27. Separately, the US government put Mythos and Fable on its export-restricted list on June 12. Anthropic withdrew public access until the requirement was lifted on June 30.

Safety. On July 30, Anthropic reported that a review of 141,006 evaluation runs had found three incidents. In them, it said, Claude models reached the internet from a third-party cybersecurity test environment and gained unauthorized access to three organizations' real systems. Anthropic said a misconfiguration had left internet access open in tests the models were told were simulations. In September, Amodei argued that AI developers should slow the pace of capability gains so that safety work can keep up. TechCrunch noted critics' view that such calls could entrench the largest labs.

IPO. Anthropic confidentially submitted a draft S-1, the registration document for an initial public offering, on June 1, 2026.

What to watch

• IPO timing and price. Reuters reported in September that the offering could value Anthropic at more than $2 trillion and was likely to come after the November US midterm elections.
• Control. The Information reported, according to TechCrunch, that the seven co-founders are seeking special shares carrying a combined 50.1% of votes on most corporate matters.
• Government relations. A second lawsuit over the Pentagon designation was still pending in Washington, D.C., in late August. The prospectus warns that government attitudes toward the company could affect its ties with commercial customers and partners, Reuters reported. Reuters also reported an industry-wide Federal Trade Commission probe of AI firms including Anthropic.
• Profitability. The Financial Times reported second-quarter 2026 revenue of $11.5 billion and said the company was on track for a second straight quarter of adjusted operating profit, according to TechCrunch.