DeepSeek raises again weeks after $7B round to fund data centers, chips

DeepSeek, the Hangzhou-based AI startup, closed its first funding round in late May at roughly $7 billion（約1.1兆円）, valuing the company at $52 billion（約8.3兆円） according to reporting by the Financial Times. Now, just weeks later, the company is in early talks with new investors for a subsequent round at a pre-money valuation of about $71 billion（約11兆円）. The fresh capital is earmarked for building the company's own data centers and purchasing AI chips. Additionally, DeepSeek is developing its own inference chip as part of a strategy to reduce reliance on Nvidia and Huawei.

The capital-raising push is directly tied to DeepSeek's aggressive expansion strategy and, in particular, its pricing model. The company recently released V4-Pro and V4-Flash, described as the largest open-weights models with up to 1.6 trillion parameters. The V4-Pro pricing has been set at permanent rock-bottom levels—roughly eleven times cheaper than GPT-5.5 on input. This ultra-competitive pricing is beginning to pay off; according to data from US financial services firm Ramp, which tracks real spending across more than 50,000 companies, DeepSeek was among the fastest-growing software vendors among US businesses in June. Ramp did flag security risks, since companies are sending data directly through DeepSeek's platform.

Despite its rapid adoption, DeepSeek is not closing the performance gap with Western leaders quickly enough. OpenAI's GPT-5.6 Sol and Anthropic's Claude Mythos have reached a new tier that DeepSeek cannot match yet, though the gap in performance is far smaller than the gap in price. Within China, the competitive landscape is intensifying. Zhipu AI released GLM-5.2, an open-source model that trails Anthropic's Opus models by only a few percentage points on hours-long coding tasks and is gaining traction with enterprises. MiniMax is reportedly working on a 2.7 trillion-parameter model expected to ship as early as Q3. Moonshot AI, the company behind the Kimi product, is seeking fresh capital at a valuation of up to $30 billion（約4.8兆円）. Founder Liang Wenfeng has already committed about $3 billion（約4800億円） of his own capital to DeepSeek, making him the largest backer. Other investors include CATL, Tencent, JD.com, NetEase, and China's state-backed AI fund.

DeepSeek's immediate need to raise fresh capital after closing its first round just weeks prior underscores the tension between aggressive market expansion and capital requirements in the AI infrastructure business. The company's pricing strategy—offering models at a fraction of Western competitors' rates—is clearly resonating with US businesses (Ramp's data shows DeepSeek among the fastest-growing software vendors in June), but sustaining that strategy requires enormous capital expenditure on data centers and chips. By developing its own inference chip, DeepSeek is also attempting to reduce dependency on US-controlled suppliers like Nvidia, a move likely motivated both by cost and by geopolitical supply-chain concerns.

The intensity of competition within China appears to be a major driver of this funding cycle. Zhipu AI's GLM-5.2, MiniMax's imminent 2.7 trillion-parameter release, and Moonshot AI's $30 billion（約4.8兆円） valuation target all signal that Chinese AI labs are competing ferociously on both capability and price. DeepSeek's performance gap with OpenAI's GPT-5.6 Sol and Anthropic's Claude Mythos has widened, yet the price gap remains enormous—a dynamic that may be pushing the company to invest aggressively in R&D and infrastructure to close the capability gap while maintaining its low-cost advantage.