Chinese AI startup DeepSeek has resumed its second funding round, aiming to raise 50 billion yuan ($8 billion) at a $74 billion valuation, just weeks after founder Liang Wenfeng paused the deal over a leaked investor transcript. Concurrently, DeepSeek has warned developers of a significant, across-the-board API price hike. The price increase highlights the infrastructure and financial pressures facing the startup as it attempts to sustain its ultra-low-cost AI model strategy amid surging global demand.
Second funding round resumption
- ▪Chinese artificial intelligence startup DeepSeek has resumed its second fundraising round, aiming to raise 50 billion yuan (approximately $8 billion) at a pre-money valuation of 500 billion yuan (approximately $74 billion)
- ▪DeepSeek and its prospective investors expect to complete the signing of the second-round investment agreements in late August 2026, with both parties seeking to proceed in a low-key manner
- ▪The list of parties negotiating for DeepSeek's second funding round includes venture capital and industrial capital firms that invested in the first round, alongside institutions that were not selected in the first round
Founder concerns over leaks
- ▪DeepSeek suspended its second funding round in late July 2026, barely two weeks after it was initiated in mid-July, due to founder Liang Wenfeng's frustration over online reports regarding comments he made to investors
- ▪The suspended funding round was triggered by a leaked transcript of a May 20, 2026 closed-door meeting on GitHub, where DeepSeek founder Liang Wenfeng admitted China trails the U.S. in AI and conceded DeepSeek's heavy reliance on Nvidia hardware
- ▪DeepSeek founder Liang Wenfeng maintains near-total voting control and holds roughly 84% of the company, having routed nearly all first-round investor funds into a limited partnership he controls with zero voting rights and a five-year lock-up
API price increases announced
- ▪The API price hike announcement came one week after DeepSeek released its DeepSeek-V4-Flash-0731 model, a 284-billion-parameter lightweight version that gained rapid traction for its cost-performance ratio
- ▪DeepSeek announced on August 6, 2026, that it will implement a significant, across-the-board price increase for its application programming interface (API) services in the near future
First round valuation growth
- ▪DeepSeek's first-round investors included China's National Artificial Intelligence Industry Investment Fund, Tencent, NetEase, JD.com, Monolith Management, IDG Capital, and CATL
- ▪The proposed 500 billion yuan valuation for DeepSeek's second funding round represents an approximate 43% increase over its first-round valuation completed just weeks prior
- ▪DeepSeek completed its first funding round in June 2026, raising 50 billion yuan (approximately $7 billion to $7.4 billion) at a valuation exceeding 350 billion yuan (approximately $52 billion)
Cost sustainability challenges
- ▪DeepSeek faces competitive pressure from American rivals, including OpenAI's GPT-5.6 Luna and Meta's Muse Spark, which have reduced prices to compete directly with DeepSeek's cost-effective models
- ▪DeepSeek's aggressive low-cost strategy, which initially featured pricing of $0.14 per million tokens, faces sustainability challenges as surging global demand overwhelms its infrastructure
Story comments
Loading comments…