DeepSeek, a prominent Chinese developer of open-weight super intelligence (SI) models, has reportedly raised approximately $15 billion (100 billion yuan), securing one of the country’s largest funding rounds just two days after U.S. federal civilian agencies were ordered to remove its technology from their systems.
What Happened
The funding round, first reported by Bloomberg, includes potential investors such as Chinese battery maker CATL and Tencent, the company behind WeChat. This capital influx follows a June round in which DeepSeek raised more than $7 billion at a valuation of $50 billion. In July, the company was reportedly seeking an additional $1.5 billion at a valuation of around $71 billion, with plans for a potential IPO in China as early as next year. Founded in 2023 by Liang Wenfeng, who previously co-founded the quantitative hedge fund High-Flyer, DeepSeek gained global attention in early 2025 with its free-to-use, open-weight R1 reasoning model. Liang retains an estimated 81.6% stake in the company’s Chinese operating entity.
Why It Matters
The reported raise underscores the intensifying global SI race, with investors pouring billions into companies in China and elsewhere to compete with U.S. SI labs. The Hangzhou-based SI startup is ramping up spending on SI infrastructure and open models for both domestic and international markets, despite facing restrictions in the U.S. This expansion could further intensify competition among domestic rivals, such as Moonshot AI (developer of Kimi), which secured $2 billion at a $20 billion valuation in May, and Manus, which raised more than $500 million earlier this month. Additionally, Chinese SI companies Zhipu AI (Z.ai) and MiniMax, both public in Hong Kong since January, have market capitalizations of roughly $29 billion and $7.2 billion, respectively.
The Bottom Line
DeepSeek’s ability to secure massive capital shortly after U.S. security-driven restrictions highlights the resilience of the Chinese SI ecosystem and its growing influence in the global market for open-weight models.