Key Highlights
- Anthropic's IPO filings show total purchase and Lease commitments of USD 900.9 billion, nearly all tied to computing capacity and infrastructure, a figure that matters for suppliers such as Nvidia (NASDAQ:NVDA).
- Third quarter Revenue reached USD 17 billion, about twelve times the level of a year earlier, while the company recorded an operating loss of USD 567 million.
- Compute and infrastructure spending was USD 9.9 billion in the quarter, about USD 0.58 for every dollar of revenue.
- Anthropic plans to price its IPO on Wednesday, November 18, 2026, with trading on Nasdaq under the ticker ANT expected the following day.
The Scale of the Commitments
Ahead of its listing, Anthropic has disclosed how much computing power it has contracted to secure. Total purchase and lease commitments come to USD 900.9 billion, nearly all of it for compute and related infrastructure, with about half expected to be incurred by the end of 2030. One agreement alone, with Data Center operator Nscale, covers up to USD 44.6 billion of Nvidia-powered capacity through 2033, with fees beginning as capacity is delivered. Those numbers dwarf the company's current revenue, but they also show how rapidly that revenue is growing. Third quarter revenue reached USD 17 billion, roughly twelve times the year-earlier figure, and the operating loss was a relatively modest USD 567 million. Compute spending of USD 9.9 billion in the quarter equaled about 58 cents for each dollar of revenue, a ratio investors will watch closely as a measure of whether AI model providers can reach sustainable margins. The commitments stretch across multiple chip suppliers and cloud partners, which spreads Anthropic's dependence on any single vendor but concentrates a very large future obligation on one company's ability to keep growing sales. For context, a commitment of that size is larger than the annual Capital spending of most of the world's largest technology companies, which is why the disclosure has drawn so much attention from investors in chips and data centers.
Why the Disclosure Matters for Nvidia
For Nvidia, Anthropic's commitments are a visible pipeline of future demand, much of it flowing through cloud providers and specialist operators that buy Nvidia systems. The timing is notable. On the same weekend, an influential independent research firm cut its forecast for global AI accelerator spending by 28% for 2028 and 40% for 2030, while still expecting USD 923 billion of spending in 2030, more than double this year's estimate. Anthropic's filings offer a counterpoint, showing one major buyer locking in capacity years ahead. Nvidia is also an investor in the deal itself. Seven computing suppliers, led by Alphabet and Nvidia, have agreed to buy USD 26 billion of non-voting shares in a private Placement at the IPO price, with Nvidia committing up to USD 10 billion. That ties a portion of Nvidia's Balance Sheet to Anthropic's valuation and links the two companies' fortunes beyond a supplier relationship. Nvidia reports its own third quarter results on Tuesday, November 17, a day before Anthropic prices.
What Investors Weigh Now
The central question is whether revenue can keep pace with commitments. Anthropic's growth rate is exceptional, but a slowdown would leave large fixed obligations, and some of the risk would flow back to data center operators and their lenders. Investors in Nvidia and other suppliers will watch the IPO pricing, early trading, any changes to Anthropic's spending plans and comments on customer Demand in Nvidia's results. A successful listing at a valuation near USD 1.4 trillion would also give Anthropic a liquid currency and fresh capital to support its commitments, reducing near-term funding risk for its partners.
Where Things Stand
Anthropic's USD 900.9 billion of compute commitments and USD 17 billion of quarterly revenue show both the size of AI demand and the Financial Risk behind it. For Nvidia, the disclosures support long-term demand even as forecasts are trimmed, with the IPO and Nvidia's results as near-term tests.






Please wait processing your request...