DeepSeek’s first funding round has been set at about RMB 50 billion, or $7.4 billion, with the company valued at RMB 350 billion to RMB 400 billion, equivalent to roughly $52 billion to $59 billion. That top-end figure is higher than the market talk circulating at the end of May.
The central detail in the round is founder Liang Wenfeng’s own commitment. He is putting in RMB 20 billion, accounting for 40% of the full raise. According to the report, Liang currently holds 84% of DeepSeek, and the personal capital injection is meant in part to limit dilution of the founding team and keep control firmly in place.
Founder capital shifts the structure of the deal
Earlier speculation had pointed to China’s national AI industry investment fund as the lead investor. With Liang stepping in as a major source of capital, that fund is now described as a regular participating investor instead. The change matters because it shifts the logic of the round: the deal is being anchored by the founder’s own money rather than by a state-backed lead.
Among outside investors, Tencent is said to be considering an investment of RMB 10 billion, while CATL is interested in contributing RMB 5 billion. If completed, those two would account for most of the external capital in the round. DeepSeek is also in final talks with NetEase and JD.com, and the total number of investors is expected to stay below 10. Hong Kong-based IDG Capital and Monolith Capital were also named as potential participants.
Valuation keeps climbing as API pricing turns aggressive
DeepSeek’s valuation has moved up quickly over a matter of weeks. TechCrunch first reported a $45 billion valuation in early May, followed by market chatter around $51.5 billion in mid-May, then $55 billion by the end of the month, and now a ceiling of $59 billion. The pace of those revisions shows how rapidly pricing expectations have changed around the Chinese large language model company.
On the product side, DeepSeek recently announced a permanent price cut for its V4-Pro API, bringing output pricing to less than TWD 30 per million tokens. That places it among the most aggressive pricing tiers in the mainstream large model market. At the same time, the company has faced talent losses. The report says four core researchers were hired away by outside institutions including Google DeepMind, a development some observers see as one trigger behind the financing process.
DeepSeek is backed by quant hedge fund High-Flyer. Public details in the report say Liang Wenfeng was born in 1985, graduated from Zhejiang University, and later moved from quantitative arbitrage into large-model development.