OpenAI’s Price War with Anthropic Could Undermine Its IPO

The Wall Street Journal reported on June 10 that OpenAI is weighing significant reductions to the prices it charges for tokens.

The discussions are described as still in flux, but this shows that OpenAI sees Anthropic closing the gap and is considering using price as a weapon.

The Numbers Behind the Pressure

Anthropic’s annualized run rate stood at roughly $1 billion at the start of 2025. By April 2026, that figure had reached $30 billion, a trajectory that CEO Dario Amodei described as outstripping the company’s own forecasts by a factor of eight.

OpenAI is cooked…Anthropic won the enterprise market thanks to their narrow focus on coding.It now projects higher revenue than OpenAI in 2029, and targets turning FCF positive by 2028 while OpenAI doesn’t expect to turn FCF positive before 2030.OpenAI is a sinking ship. pic.twitter.com/AC2ISdaMUC

— Oguz Erkan (@oguzerkan) January 28, 2026

Claude Code, Anthropic’s AI coding agent, hit $1 billion in annualized revenue within six months of its public launch in May 2025 and surpassed $2.5 billion by February 2026.

Business subscriptions to Claude Code quadrupled in the first quarter of 2026 alone. Some analysts now estimate Anthropic’s annualized revenue may have crossed $47 billion by May 2026.

OpenAI, by comparison, reported a revenue run rate of approximately $13 billion in 2025 and does not expect to turn a profit or generate positive free cash flow until 2030.

That gap in growth trajectory, not just in current revenue, is what makes Anthropic’s rise threatening enough to prompt a pricing response.

The IPO Paradox

Both OpenAI and Anthropic are understood to be eyeing public listings. OpenAI has been in discussions for a funding round that would value the company at $750 billion. Anthropic closed a $30 billion Series G in February 2026 at a $380 billion post-money valuation.

Deliberately cutting the price of your core product immediately before that process is a decision many would view as damaging to its IPO goals.

The counterargument is that lower prices drive volume, which in turn drives revenue. That logic holds in markets where demand is highly elastic and where the company’s cost base is manageable.

AI infrastructure is neither. Training and serving frontier models is expensive at a level most industries never encounter. OpenAI’s own leadership has acknowledged that the cost of training a single competitive model is approaching, and in some projections, far exceeding $1 billion.

Cutting token prices in that environment does not automatically translate into higher profits. It can just as easily mean more loss, spread across a larger customer base.