- Anthropic’s IPO prospectus shows revenue rose 1,304% to $16.29 billion in the six months ended June 30, from $1.16 billion a year earlier, after full-year 2025 revenue jumped 1,089% to $4.59 billion. The company also lost $80.7 billion in the first half, versus $12.9 billion a year earlier, a total it said largely reflects accounting charges on convertible notes as its enterprise valuation rose.
- The balance sheet is getting heavier fast. Anthropic had $6.2 billion of debt at the end of June, mainly convertible notes issued to Amazon since 2023, while non-cancelable purchase commitments rose from $54.6 billion at year-end 2025 to $265.6 billion by June 30, roughly a 4.9-fold increase (Lumida calculation), and future undiscounted lease obligations climbed from $50.2 billion to $321.4 billion.
- Compute deals tie Anthropic to the same companies it competes with. It agreed to pay SpaceX about $3.1 billion a month for Nvidia-powered compute once capacity is delivered, up to $120.5 billion into 2029, intends to spend at least $44.9 billion with Microsoft between November 2026 and May 2033, and gave Google an option to buy up to $10 billion of stock at the IPO price.
- The more than 90 pages of risk factors flag existing and proposed U.S. moratoriums on new data center construction, which 16 states had formally weighed as of last month, along with dependence on TSMC in Taiwan and possible legal liability if its models help develop weapons. A Founder LLC of seven co-founders will hold 50.1% of the voting power, a structure Anthropic says has “no direct public company precedent.”
What Happened?
Anthropic filed its S-1 on Monday, with Goldman Sachs as lead-left underwriter and Morgan Stanley alongside it, at the top of a syndicate that also includes JPMorgan, Citigroup, Barclays and Bank of America. The filing shows consumption-based revenue of $13.1 billion, or 80% of the total, in the first half, with subscriptions making up the other $3.2 billion. As of Sept. 30 it had more than 11,900 customers generating $100,000 or more of trailing twelve-month revenue, more than 1,900 above $1 million and roughly 250 above $10 million, and 5,300 employees versus 940 at the end of 2024. No single customer represented more than 10% of revenue in the nine months ended Sept. 30, though five customers accounted for about a quarter of it. Anthropic closed its last private round in May, raising $65 billion at a $965 billion valuation.
Why It Matters?
The filing gives public investors their first detailed look at the economics of a frontier AI lab, and the numbers cut both ways. Revenue is compounding at a pace few companies have ever reached, but the off-balance-sheet commitments to compute and leases now run into the hundreds of billions of dollars, which means the bet only works if demand keeps scaling. MIT Sloan’s Sinan Aral told CNBC that a $1.5 trillion to $2 trillion valuation implies roughly 22 to 30 times revenue, so future growth and profit are already being priced in. The risk section also matters for the wider AI trade, because data center moratoriums, chip supply from Taiwan and power availability are constraints that apply to every hyperscaler and neocloud building capacity.
What’s Next?
Investors will watch for the price range, share count and listing date, which will set the valuation against the $965 billion private mark and OpenAI’s $852 billion valuation from March. Alphabet’s decision on its $10 billion option, and how quickly Anthropic turns its compute commitments into revenue, will shape views on its funding needs after the IPO. The governance structure, which Anthropic says may make it more difficult to replace directors and management, is likely to draw scrutiny from large institutional investors. State-level data center legislation is another variable to track, since Anthropic says moratoriums could delay or prevent the facilities it depends on for compute.
