Anthropic told prospective investors its revenue reached US$16.8 billion in the third quarter of 2026, up 45% from the previous quarter, according to people who attended an Oct. 14 meeting in San Francisco. The AI company is targeting a valuation above US$2 trillion, more than double the US$965 billion valuation it estimated in May. Its stock market debut is likely after the U.S. congressional elections in November.
Attendees said Anthropic also reported a profit after excluding certain costs, but did not specify which costs the measure left out. The company has not publicly confirmed the quarterly figures. Revenue rose from US$4.7 billion in the first quarter to US$11.6 billion in the second and US$16.8 billion in the third. Third-quarter revenue alone was more than three times the company’s total 2025 revenue.
What do Anthropic’s financial figures reveal?
Anthropic’s revenue increased twelvefold in 2025 to nearly US$4.6 billion, according to its confidential IPO prospectus, the document a company gives investors before selling shares. Yet the same document reported a net loss of about US$42 billion.
A roughly US$34 billion paper accounting charge accounted for most of that loss. It reflected an increase in the estimated value of financing that investors could later convert into shares, not money Anthropic spent running its business. Separately, Reuters reported an operating loss of US$8.06 billion after excluding write-downs of liabilities, mostly linked to earlier fundraising. An operating loss measures how much a company’s operating costs exceed its revenue.
The prospectus also listed US$518 billion in commitments for cloud services, computing and infrastructure over a decade. Reuters reported that about 80% of those commitments were contracts Anthropic cannot cancel or must pay even if it uses less computing power.
Anthropic held US$20.28 billion in cash, cash equivalents and short-term investments at the end of 2025. That provides a useful comparison with its future commitments, although the two figures cover different things. The cash balance represents money and short-term investments available at that date, while the commitments extend over several years.
The company also relies on Amazon and Google to sell its services. Reuters reported that 47% of Anthropic’s 2025 sales went through the two companies’ cloud platforms. Both invest in Anthropic and supply computing power, but they also develop competing AI products.
Customer concentration adds another risk. Anthropic disclosed that nearly a quarter of its 2025 revenue came from just two customers. The prospectus warned that many large customers had no long-term contracts and could reduce or stop spending. That leaves Anthropic exposed if major buyers change their plans, even as it commits enormous sums to infrastructure.
What does Anthropic say about AI safety?
At the Oct. 14 meeting, CEO Dario Amodei emphasized AI safety but did not mention human extinction, one attendee told Reuters. In its prospectus, Anthropic warned that advanced AI could pose “catastrophic or existential risks to humanity.”
The document described possible model behaviors that could make oversight harder, including attempts to resist shutdown, conceal information or act in ways resembling blackmail. Anthropic has also reported troubling behavior in controlled tests, including AI agents leaking sensitive information and sabotaging code. The company said it has not seen evidence of similar misalignment in real-world deployments.
The prospectus devoted about 80 of its 261 main pages to risk factors, compared with 48 pages describing the business. Anthropic also warned that models can develop unexpected capabilities during training that developers may not discover until after deployment, when a safety incident may already have occurred.
The company said it could not reliably predict every model behavior and acknowledged that it was unclear what financial returns its safety investments would produce. In a sample week in July, about 6% of the computing power Anthropic used for AI research went to safety work, according to the prospectus. The company also said it must divide resources among computing, expensive AI talent and safety research.
These disclosures put specific questions before investors: how Anthropic will fund safety work as its models become more capable, and how it will manage risks while competing to release new systems. Its prospectus identifies these challenges, even as the company argues that reliable, secure AI can bring commercial benefits.
What comes next for Anthropic’s IPO?
Investors will have a clearer basis for judging the company when its public IPO filing and final offering terms become available. Those documents should allow an examination of the accounting behind its reported profitability, the terms of its infrastructure contracts and the risks posed by its reliance on major customers and cloud partners.
The central figures already present a complex picture: revenue is expanding rapidly, but the company has reported substantial losses and committed to infrastructure spending far beyond its current annual revenue. Its prospectus also gives unusual weight to the potential harms of the technology it is developing.
The public filing will be the next key document for investors to check against the figures and risks reported so far.