Anthropic PBC’s revised IPO prospectus shows a business still growing at a pace few companies have ever matched, but slowing sharply from quarter to quarter and carrying a compute bill that dwarfs its revenue, details that will shape what investors are willing to pay when the AI developer prices its offering.
The amended S-1, filed with the Securities and Exchange Commission on Monday, adds third-quarter numbers that were missing from the initial filing in late October. Revenue reached about $17 billion in the three months through September, roughly 12 times the year-earlier figure and more than $5 billion above the second quarter. But sequential growth slowed to about 47%, from 144% in the second quarter and 132% in the first. For the first nine months of the year, revenue totaled $33.24 billion.
The compute bill
Anthropic swung to a $567 million operating loss in the third quarter, after $974 million of operating income in the second, which it attributed mainly to about $5 billion spent on short-term compute that costs far more than multiyear contracts. Compute and infrastructure charges reached $9.9 billion in the quarter. The net loss was $23.29 billion, largely reflecting changes in the fair value of convertible notes and derivatives rather than operating cash costs.
The filing also tallies about $895.4 billion of purchase commitments and leases, substantially all tied to computing power and infrastructure and stretching as far as 2045. Nearly half is due by the end of 2030, including about $69.8 billion in the 12 months after Sept. 30. Anthropic reported about $146.6 billion of cash and other liquid assets.
Valuation pressure
The company still hasn’t disclosed a price range, share count or offering size. Morgan Stanley and Goldman Sachs are leading the syndicate, and people familiar with the plans have pointed to a roadshow starting around Nov. 9 with pricing the following week. Some institutional investors have urged a valuation closer to $1.5 trillion to $1.6 trillion, below the roughly $2 trillion discussed earlier, according to published reports, as AI-linked stocks have slumped on worries about the cost of the infrastructure buildout.
