DeepSeek Relaunches Second Funding Round at RMB 500B Pre-Money Valuation
DeepSeek, the Chinese AI large language model startup that has drawn intense investor interest, has quietly resumed its second round of external fundraising after a brief suspension, according to an exclusive report by a leading Chinese financial publication. The development underscores the accelerating pace of capital formation around China's top-tier AI model companies.
According to Caijing magazine's exclusive report published on August 5, 2026, multiple deal participants confirmed that DeepSeek has relaunched its Series B fundraising effort, targeting RMB 50 billion yuan (approximately US$6.9 billion) at a pre-money valuation of RMB 500 billion yuan (approximately US$69 billion). The company is aiming to finalize agreements by late August.
The fundraising timeline has been turbulent. Sources told Caijing that the round initially opened in mid-July but was abruptly halted around July's end, with some prospective investors on the waitlist notified that signing plans had been put on hold. The suspension was reportedly linked to DeepSeek founder Liang Wenfeng's dissatisfaction over widely circulated content purportedly based on leaked investor meeting transcripts. DeepSeek did not respond to Caijing's request for comment.
Now that the round has restarted, both the company and its negotiating partners are said to prefer a low-profile process. Notably, some institutions that had been in active discussions say they have yet to receive formal notice of the resumption, suggesting the outreach remains selective.
The current pre-money valuation of RMB 500 billion represents a roughly 43% premium over the Series A valuation of more than RMB 350 billion. DeepSeek closed its inaugural funding round — raising RMB 50 billion yuan — in June 2026, making it the largest first-round raise in Chinese AI history. Investors in that round included the National Artificial Intelligence Industry Investment Fund, Tencent, CATL, NetEase, JD.com, and IDG Capital, among others. If the current round closes successfully, DeepSeek will have raised a cumulative RMB 100 billion yuan across two rounds.
Investor demand has consistently outpaced supply. Sources indicated that capital expressing interest in the first round exceeded RMB 100 billion yuan, nearly double what was ultimately raised, leaving a substantial pool of sidelined investors eager to participate in subsequent rounds.
The fundraising activity comes amid intense model competition. DeepSeek's V4-Flash, which entered public beta on July 31, scored 50 on Artificial Analysis's Intelligence Index — second among domestic models — while maintaining an output price of just US$0.28 per million tokens, a fraction of rivals' pricing. The model topped OpenRouter's weekly token consumption rankings with 7.1 trillion tokens processed.
The broader sector is also heating up. Moonshot AI closed its Series F at a valuation of approximately US$31.5 billion in late July and has since launched a Pre-IPO round at a US$50 billion valuation, with sources indicating a potential listing filing before year-end.
As one deal participant observed to Caijing, pricing large AI model companies is "essentially an options trade, not a cash-flow-based financial model" — a dynamic that is likely to keep valuations volatile as model capabilities continue to evolve rapidly.
Related Coverage:
DeepSeek Closes $6.9B Round as Liang Wenfeng Lays Out AGI Roadmap and Chip Strategy
DeepSeek-V4-Flash Punches Above Its Weight, Undercutting OpenAI on Cost by 60%