Chinese AI laboratory DeepSeek has reached $400-500 million in annualized recurring revenue, the company revealed as it seeks to raise approximately $7.4 billion in a new funding round targeting a $74 billion valuation, according to The Information.
The Hangzhou-based startup, which burst onto the global AI scene with its open-weight reasoning models, disclosed its first concrete revenue figures—marking a significant milestone from its earlier status as a primarily self-funded research lab. The revenue stems from sales of DeepSeek’s AI models, which have gained traction among developers and enterprises seeking alternatives to OpenAI and Anthropic offerings.
The new funding round would value DeepSeek at roughly $74 billion, representing a substantial jump from the approximately $71 billion valuation figures previously reported for the same fundraising effort. The company is also reportedly preparing for a mainland China IPO, with a late-2026 filing targeted for a 2027 market debut.
DeepSeek has achieved over 50% gross margin from its model sales, attributed to infrastructure improvements the lab has implemented over the past year. This profitability metric stands out in an industry where many AI companies continue to burn cash pursuing compute-intensive model development.
The revenue disclosure comes amid intensifying competition in the global AI market, with DeepSeek positioning itself as a viable open-source alternative to Western AI providers. The company’s models have attracted attention for their strong performance at competitive price points, challenging the dominance of OpenAI’s GPT-series and Anthropic’s Claude lineup.
If the funding round closes at the reported valuation, DeepSeek would join the ranks of China’s most valuable private AI companies, rivaling valuations achieved by ByteDance and other technology giants in the region.