Artificial intelligence firm Anthropic generated $1.70 in revenue for every $1 spent on computing power during the third quarter, presenting the metric to investors weeks before its planned November initial public offering (IPO). Lead underwriters Goldman Sachs and Morgan Stanley are preparing an offering that targets a valuation exceeding $2 trillion.
Compute Economics Flip as Revenue Reaches $16.8 Billion
Quarterly revenue surged 50% to $16.8 billion in Q3, up from $11.5 billion in Q2 and $4.8 billion in Q1, marking the company's second consecutive quarter of adjusted profit. The financial figures, reported by the Financial Times, demonstrate a sharp turnaround from 2025, when Anthropic generated only $0.63 in revenue per $1 of compute spending—earning $4.6 billion against $7.33 billion in compute costs and incurring an operating loss exceeding $8 billion.
Executive leadership, including Chief Executive Dario Amodei and Chief Financial Officer Krishna Rao, confirmed an operating margin of 11% and positive cash flow, though Q2 adjusted profits excluded stock-based compensation and gross margins of 80% omitted training expenses. Anthropic has committed at least $518 billion over a decade to cloud infrastructure, compute capacity, and biomedical research.
Growth Headwinds and Pricing Pressure
Despite positive quarterly results, Anthropic's revenue expansion showed signs of cooling compared to earlier this year, when Q2 revenue more than doubled Q1 figures. Research firm SemiAnalysis revised its year-end annualized revenue projection to $74 billion, up from $65 billion in July, but down from summer investor estimates that anticipated passing $100 billion.
Several key factors are applying pressure to growth:
- Client Independence: Major corporate clients Meta and Microsoft are increasingly prioritizing internally developed AI models over third-party providers.
- Distillation Restrictions: Anthropic limited model distillation—where rival developers pay to train models on Claude outputs—which previously accounted for 5% to 7% of overall revenue.
- Price Competition: Intensifying competition from low-cost Chinese laboratories and OpenAI drove price reductions, leading Anthropic to price its Claude Haiku 5.5 model 75% lower than its predecessor this month.
Why It Matters
Anthropic's shift to positive unit economics demonstrates that massive frontier model infrastructure costs can yield cash flow as hardware efficiency improves. However, slowing sequential growth rate highlights how rapidly enterprise customers shift toward in-house solutions when API costs remain high. With capital markets closely tracking market concentration amid broader discussions on how public markets evaluate tech valuations and rising enterprise borrowing costs, Anthropic's $2 trillion pitch will test investor appetite for long-term infrastructure commitments in an increasingly commoditized AI software market.
