Anthropic put a number in front of prospective shareholders this week: $16.8B in third-quarter revenue, up roughly 45 percent from the $11.6B it booked in the second quarter. The figure surfaced at a private meeting with institutional investors ahead of a listing widely expected in November.
The company also told the room it turned profitable on an adjusted basis and generated $1.70 for every dollar spent on computing power. Executives are said to be targeting a valuation near $2 trillion.
The growth is real, the accounting is fuzzy
The Financial Times, which first reported the figure, could not confirm whether $16.8B represents gross or net revenue, or how Anthropic defines profitability. The company is still private and under no obligation to publish audited statements.
That ambiguity matters more than usual right now. The same outlet recently corrected OpenAIās annualized run rate downward from about $68B to $50B after a calculation error, a revision that knocked the Nasdaq down 1.25 percent and dragged Oracle and other AI-linked names with it.
Analyst Holger Mueller of Constellation Research called the disclosure a presentation exercise. āIt is a new number, it is bigger, but it is still circled by tons of questions about how it was arrived at,ā he said.
Momentum heading into a listing
Anthropic closed a $65B round in May that valued it near $965B. Claude Code, its coding agent, has become a mainstay in enterprise engineering teams and is widely credited with pulling the company ahead of OpenAI on commercial traction.
The run-up has not been smooth. Safety researcher Jacob Coxon resigned in September, warning publicly that the people building frontier systems believe the technology could pose catastrophic risks before the decade ends.
