DeepSeek reportedly raises about $15 billion in one of China's biggest rounds

DeepSeek has reportedly raised about 100 billion yuan, roughly $15 billion, in its latest fundraising round. Reports describe it as one of the largest ever in China. Tencent and CATL are named among the biggest contributors, according to people familiar with the matter.

Coverage frames the raise as a sign of the startup's ambition to compete globally with Anthropic and OpenAI. Some headlines say the round was double what DeepSeek first sought, and others tie it to a possible IPO, though those details come only from headlines in the material and are not confirmed here.

The money lands as DeepSeek keeps shipping. In October 2026 it published a point update to its V4.1 Flash model with MIT-licensed weights, a 1 million-token context window, better results on every official benchmark, and unchanged token prices. Its open-source models are free to access and much cheaper to use than many competing tools. Snowflake, Vercel and OpenRouter offer DeepSeek models to customers, and its latest release has drawn praise online, especially for coding.

The company also faces headwinds. A DHS ban has been announced without a stated explanation of the specific security risks. The Trump administration sees China as a major AI competitor, and OpenAI and Anthropic have accused Chinese companies, including DeepSeek, of copying their models. DeepSeek previously said it spent around $6 million training its first model, versus an estimated $100 million for GPT-4, though those figures cover different costs.

Our take

A $15 billion war chest changes DeepSeek from a cheap, efficient challenger into one that can fund frontier-scale training and compete on compute. The open weights and low prices already pressure US labs, and this money lets it keep that up.