Business

OpenAI and Anthropic Prepare for a Token-Based Price War

OpenAI is reportedly considering significant cuts to its per-token pricing to compete with Anthropic, even though both companies currently operate at heavy losses because compute costs exceed revenue per query.

OpenAI and Anthropic Prepare for a Token-Based Price War

OpenAI is reportedly considering steep reductions to its per-token charges in an effort to win customers away from Anthropic, according to the Wall Street Journal. Tokens are the billing units used by AI providers; a single request can consume multiple tokens and most firms price usage per token.

Why this price war is unusual

Unlike typical price wars fought from profitable positions, this confrontation begins from a loss-making baseline. Both companies currently operate at a loss because the compute and infrastructure required to serve queries cost more than the revenue per request. Lowering token prices further would intentionally increase those losses.

Sam Altman, CEO of OpenAI, has recently described AI costs as “a huge issue.” Meanwhile, Anthropic — developer of the Claude family of models — has for the first time overtaken OpenAI’s valuation, helped by the viral success of Claude Code. OpenAI is pushing its own Codex offering to respond.

Practical effects on the market and customers

The strategy appears to be a bet that locking in enterprise customers is worth accelerating losses: larger short-term deficits in exchange for a bigger installed base that could produce long-term benefits. Repricing tokens is being treated like adjusting a kilowatt-hour charge, but unlike a utility, the service cannot break even at current or proposed price points, raising questions about how long such a strategy is sustainable.

This dynamic forces questions about the industry’s business models: how long can investors and owners tolerate deeper losses to secure market share, and will customers respond to lower prices by increasing usage or by switching providers?

Stakes and implications

If the two most advanced AI companies accept losing money faster as a path to victory, technical superiority ceases to be a protective moat and instead becomes the fuel they burn to stay competitive. The outcome could reshape pricing norms, influence vendor strategies, and alter investor expectations across the AI sector.

Observers should watch how enterprise clients react, what countermeasures competitors take, and the longer-term impact on the sector’s financial sustainability.