OpenAI and Anthropic face AI price war ahead of IPOs

OpenAI is weighing steep cuts to its developer token fees while Anthropic, now valued at $965 bn, is already ahead of the San Francisco‑based lab’s $852 bn estimate [1]. The pricing debate emerges as both companies prepare confidential S‑1 filings that could see them listed as early as September, turning a private AI duel into a public‑market showdown [1].

Anthropic’s surge rests on a rapid rise in enterprise revenue, which it said topped $30 bn annually after a jump from $9 bn a year earlier, driven largely by its Claude Code coding assistant [1]. OpenAI, by contrast, generates roughly $2 bn in monthly revenue—about $24 bn on an annualised basis—and does not expect profitability until 2030 [1]. Analysts note that the two firms now compete not only on model capability but on the economics of token consumption, a battleground that could dictate margins for the whole AI stack [1].

The rivalry is amplified by a paradoxical public stance: both labs have been publishing warnings about the speed of frontier AI development while simultaneously launching their most powerful models and offering free usage incentives to accelerate adoption [2]. Anthropic’s recent paper calls for a coordinated “slowdown or pause” on self‑improving AI, and its CEO Dario Amodei has urged binding regulation [2]. OpenAI’s CEO Sam Altman echoed the sentiment, proposing an international body to align safety and development [2]. Yet each company released a new flagship model—Claude Fable 5 from Anthropic and GPT‑5.5 from OpenAI—alongside expanded free‑tier limits for developers [2].

Investors fear that a token‑price war could compress margins across the AI stack. If Anthropic’s lower prices win enterprise customers, compute demand may shift toward its partners, notably Nvidia and Microsoft, while OpenAI’s higher‑priced tokens could lose traction [1]. Conversely, OpenAI could match Anthropic’s cuts without ceding market share, preserving usage growth and offsetting margin pressure [1]. The outcome will shape not only the profitability of the two IPO candidates but also the broader economics of AI inference spend.

The real question now is whether the pricing contest will force one of the giants to sacrifice short‑term revenue for longer‑term market dominance, or if both will coexist with differentiated pricing strategies as they step onto the public stage.