OpenAI is weighing significant token price cuts after Anthropic's valuation eclipsed its own for the first time, signaling a potential price war between the two most valuable AI companies at the worst possible moment for either's finances.
The Wall Street Journal reported Wednesday that OpenAI's deliberations are not yet final, but the directional intent is clear: cut per-token prices before Anthropic does, or risk losing more of the developer market that has visibly shifted toward Claude. The catalyst is hard to ignore. Anthropic closed a $65 billion funding round in late May, pushing its valuation to roughly $965 billion and, for the first time, eclipsing OpenAI's $852 billion mark from its own March 2026 round. The company that Sam Altman built now finds itself playing defense on price.
Claude Code is what tilted the balance. Anthropic's coding assistant crossed $1 billion in revenue within six months of its launch, drawing engineers away from OpenAI's ecosystem in measurable numbers. By April 2026, ChatGPT's share of global generative AI web traffic had fallen from 77.6% in May 2025 to 53.7%, according to web analytics data. On the Ramp AI Index, which tracks enterprise software spending, more companies are now paying for Anthropic than for OpenAI. That is a first, and OpenAI knows it.
For developers and startups building on top of these APIs, cheaper tokens would arrive at a moment of genuine need. The cost of running AI-powered applications has become a real burden at scale. Uber's CTO confirmed publicly that the company burned through its entire 2026 AI budget in four months, with per-engineer monthly API costs ranging between $500 and $2,000 as agentic workflows compounded consumption far beyond initial projections. Any meaningful price reduction from OpenAI would ease that pressure and accelerate adoption at layers of the stack where cost remains the primary friction. The startups that have been throttling usage to stay solvent would have room to breathe.
The problem is that both companies can barely afford the generosity. OpenAI and Anthropic are projected to spend a combined $65 billion on compute and operations in 2026 alone, and neither is profitable. Anthropic expects to reach breakeven by 2028; OpenAI is staring at a timeline roughly two years beyond that. Both companies have filed confidentially for IPOs, with Anthropic targeting an October 2026 debut. Voluntarily compressing revenue per token in that window is a decision public-market investors will scrutinize, particularly when the argument amounts to: we are cutting prices to defend market share from a rival that is already worth more than us.
Google is watching all of this from a comfortable distance. Gemini has been cutting API prices aggressively throughout 2026, and sustained price competition between OpenAI and Anthropic validates what Google has argued all year: that frontier model capabilities are converging fast enough that price will increasingly decide the outcome. The DeepSeek disruption earlier in the year made the same point from a different direction, demonstrating that capable models can be built at a fraction of the capital the American incumbents have deployed. If the two dominant players race to the bottom on token pricing, value migrates upward to application-layer companies and downward to whoever controls the cheapest compute. The enterprise AI infrastructure startups positioned in the middle get squeezed from both ends.
Both Altman and Dario Amodei will need to tell coherent profitability stories on IPO roadshows. The bet behind any price cut is that volume growth from newly cost-competitive APIs offsets per-token margin compression, and that winning developer loyalty now converts into durable enterprise contracts later. It is a defensible wager, but it depends on usage scaling faster than pricing falls. Given that even Uber exhausted its AI budget in four months at current rates, there is a real question about how much further consumption can actually expand when the bills come due.
Also read: MediaTek doubles its AI chip target to $2 billion and enters the data center • Palantir's NHS battle reveals the political risk enterprise AI investors ignore • Software Stocks Are Posting Record Earnings While Investors Price In Their Obsolescence
