DeepSeek's annualized revenue run rate has more than doubled to roughly $1 billion, the Chinese AI lab told investors, as the company finalizes a second funding round and prepares to go public on the Shanghai Stock Exchange. The figure, shared by DeepSeek chief executive Liang Wenfeng, was first reported by The Information and is up from about $500 million just months earlier.
Nearly all of that revenue comes from a single line of business: access to DeepSeek's models through its API. The jump follows a round of steep price increases the company introduced last month, when it raised API prices by 2.3 to 4.5 times across its model lineup. Wenfeng told investors that customer demand held up despite the increases, and DeepSeek continues to position its pricing as among the lowest of any major AI provider, even after the hikes. Wenfeng also said the company is still devoting more than 70% of its computing capacity to training rather than to serving paying customers.
The revenue disclosure comes as DeepSeek finalizes a funding round targeting roughly 50 billion yuan, or about $7.5 billion, at a valuation of 500 billion yuan, or roughly $75 billion. That round is expected to close by the end of October. Reports on the fundraising say DeepSeek plans to file its Shanghai listing application before year's end, with an IPO on the exchange's STAR Market targeted for 2027.
DeepSeek's growth trajectory has been unusually steep even by AI-industry standards. The Information reported in August that the company's revenue for the first seven months of 2026 — about 475 million yuan, or roughly $70.7 million — was already ten times its total revenue for all of 2025. The company first drew global attention in early 2025 with a model that matched the performance of leading U.S. systems while reportedly using far fewer Nvidia chips than its rivals, unsettling both Silicon Valley and Wall Street.
DeepSeek did not respond to a request for comment relayed through The Information's reporting. Its pricing strategy has also drawn attention from rivals: when it introduced peak-hour pricing for its flagship V4 models in August, Bloomberg reported the new rates would roughly quadruple prior levels, even as DeepSeek maintained a cost advantage over Western competitors such as OpenAI and Anthropic.