Two days after the US told federal civilian agencies to pull DeepSeek from their systems, the Chinese lab reportedly raised about $15 billion. That is roughly 100 billion yuan, and it makes this one of the largest funding rounds in China’s history. If the goal in Washington was to spook capital, it didn’t work.
According to TechCrunch, citing a Bloomberg report, potential investors include battery maker CATL and Tencent, the company behind WeChat. DeepSeek did not immediately respond to a request for comment, so none of this is confirmed by the company.
The numbers climb fast. In June, DeepSeek raised more than $7 billion at a $50 billion valuation. In July it was reportedly seeking another $1.5 billion at around $71 billion, with a possible IPO in China as early as next year. Now there’s a $15 billion round on top. The Hangzhou startup is putting the money into AI infrastructure and open models for both domestic and international markets.
That open-model strategy is the real story. DeepSeek got global attention in early 2025 with its free, open-weight R1 reasoning model, which challenged the assumption that US labs like OpenAI, Anthropic and Google had a lock on frontier reasoning. Open weights let developers run and modify the model themselves, and that is exactly why it spread. It is also why governments worry about it. A ban on federal systems is a security move, but it doesn’t touch the developers who already download the weights.
The round also fits a wider pattern. Chinese AI companies are pulling in serious money. Moonshot AI, the Kimi developer, secured $2 billion at a $20 billion valuation in May. Manus AI raised more than $500 million this month. Zhipu AI (Z.ai) and MiniMax both listed in Hong Kong in January, and their market caps are now roughly $29 billion and $7.2 billion. DeepSeek’s valuation, even at the June figure, sits well above all of them.
For domestic rivals, this is bad news. A war chest this size lets DeepSeek spend more on compute and talent, and that squeezes everyone else in China’s crowded model market. For US labs, it means cheap, capable open models will keep arriving, which pressures API pricing for tools like GPT and Claude.
One more detail matters for control. Founder Liang Wenfeng, who previously co-founded the quant hedge fund High-Flyer, which helped finance DeepSeek early on, holds an estimated 81.6% of the Chinese operating entity after the earlier round, per WireScreen data. New money may dilute that, but he is unlikely to lose the steering wheel.
So is this noise? No. Restrictions in the US haven’t cut DeepSeek off from funding, and the company is clearly playing a global game. Watch whether the round is confirmed, and whether the IPO plan holds.
