DeepSeek, the Chinese AI startup behind the R1 large language model, has reportedly seen its annualized revenue surge to between $400 million and $500 million. That's roughly double the company's run rate from earlier in 2025, according to figures cited by multiple outlets.
Doubling the run rate
Annualized revenue is a projection based on recent monthly or quarterly performance, so the new range suggests DeepSeek is pulling in around $33 million to $42 million per month. The jump from its previous run rate — which was around $200 million to $250 million on an annualized basis — marks a sharp acceleration in just a few months.
The company has not confirmed the numbers publicly. But the report, which first surfaced in industry newsletters, has been picked up by tech and financial media. DeepSeek's growth comes as the broader AI sector sees intense competition, with players like OpenAI, Anthropic, and China's Baidu all vying for enterprise and consumer customers.
What the figure means
A $400 million-plus run rate puts DeepSeek in a different league from most AI startups, though still far behind the billions that OpenAI is believed to generate. For a company that only launched its flagship model in late 2024, the pace is notable. DeepSeek's models have gained attention for their performance and relatively low cost, which may be driving adoption among developers and businesses.
The revenue is likely coming from API access, enterprise licensing, and possibly consumer subscriptions, though DeepSeek hasn't broken down the sources. The company is based in Hangzhou and has raised significant venture capital, including from Chinese backers.
No official word yet
DeepSeek has not issued a press release or statement about the revenue figure. The report's accuracy remains unverified by the company itself. Investors and competitors will be watching for any official update, especially as the AI market continues to evolve rapidly.
