Anthropic could surpass OpenAI in annualized revenue by mid-2026

Our revenue data comes from Epoch’s AI Companies database, which combines company disclosures and media reports into standardized annualized revenue records. For this chart, we use OpenAI and Anthropic revenue data starting at the date where each first reported $1B in annualized revenues. You can explore the full data on the AI Companies hub.

Key source outlets include The Information, Bloomberg, CNBC, Financial Times, The Wall Street Journal, Reuters, and The New York Times, alongside direct company disclosures.

We fit simple exponential trends to each company’s annualized revenue data.

• OpenAI annual growth — 3.4×year (90% CI: 3.3× to 3.6×)
• Anthropic annual growth — 10×/year (90% CI: 8× to 12×)
• Estimated crossover point — August 2026 at roughly $43 billion in annualized revenue (90% CI: February 2026 to April 2027)

In the main plot, we show 90% prediction intervals to show the range over which new datapoints are expected to be observed. All reported intervals are based on the statistical noise in our data, and are not all-things-considered forecasts of OpenAI’s and Anthropic’s future revenue.

Our crossover estimate is a simple extrapolation of recent revenue growth. Growth rates this high often slow as companies scale, due to market limits, competition, pricing changes, and operational constraints. Both OpenAI and Anthropic have indicated internally that they expect slower growth in 2026 than our trend-based projection; for example, investor reporting has suggested OpenAI is budgeting roughly 2.2× growth in 2026 and Anthropic projecting 4× growth or less. Their internal forecasts imply a crossover date sometime during 2026 or 2027, consistent with our analysis.

We use simple exponential fits to forecast trends, which fit remarkably well (R² is 0.99 for OpenAI, and 0.98 for Anthropic). However, we find weak evidence that Anthropic’s annualized revenue trend has slowed recently. Using a piecewise exponential fit, we find a breakpoint in July 2025, with a post-breakpoint slope of 7×/year. The breakpoint model is moderately preferred on BIC-based model selection, with a likelihood ratio of 7:1. If this more recent trend continues, the expected crossover date is in December 2026.

Our revenue figures come primarily from media reports, which may have timing or accuracy uncertainties.