OpenAI is preparing to slash token prices in anticipation of Anthropic making the same move, according to a Wall Street Journal report Thursday. The company expects its rival to blink first on pricing and wants to be ready to follow immediately. The discussions are still in flux, the WSJ reported, citing people familiar with the matter.
But CEO Sam Altman has already telegraphed the direction publicly. At a recent event, he acknowledged that AI costs have become "a huge issue" for business customers. "I think we'll have a lot of ways we can help people get more value for less spend," he said.
The problem: both companies are already losing billions of dollars. Computing costs to run large-scale AI models are enormous, and cutting token prices would compress margins further at the exact moment both OpenAI and Anthropic are pursuing IPOs that will expose their economics to public investors for the first time.
OpenAI filed confidentially for an IPO this week. Altman told employees the company plans to go public within the next year.
Anthropic has also filed and both are expected to list this fall. The competitive pressure driving the price discussion is real. Anthropic's revenue surged after its coding tool Claude Code gained traction with software engineers, briefly pushing the startup's valuation past OpenAI's.
Anthropic closed its Series H funding round on May 28 at a $965 billion valuation, edging out OpenAI's $852 billion valuation from March. OpenAI responded by making its own coding tool, Codex, a company priority.
Investors have long flagged one structural vulnerability: customers can switch between the two products with relative ease. A price war tests that weakness directly. Whoever blinks first sets the floor for an industry that has not yet figured out how to grow profitably.
Enterprise enthusiasm for AI is already hitting budget ceilings. An Uber executive said earlier this year the company had exhausted its 2026 spending on agentic AI.
Another executive noted last month it was hard to connect AI-driven coding gains to actual product improvements customers could see. Those admissions have sparked a debate in Silicon Valley about "tokenmaxxing", burning through tokens at high volume without a clear return.
Altman's promise of "more value for less spend" runs into hard math. According to the WSJ, OpenAI plans to invest $121 billion in computing in 2028 but is expected to post a deficit of $85 billion that year alone. Anthropic is targeting its first profit around 2028. OpenAI is expected to burn cash through 2030.
Price cuts could signal to public investors that growth depends on discounting rather than sustainable pricing power. For an industry that has captured most of the revenue flowing into new AI products but hasn't proven it can grow profitably, that's the last signal either company wants to send ahead of a public debut.
