DeepSeek Weighs Doubling Funding Round to $15 Billion

China's DeepSeek is considering expanding its latest funding round to as much as 100 billion yuan (about $14.9 billion), roughly double the target it set when the raise opened, according to people familiar with the talks who spoke to CNBC. The move would rank among the largest private AI financings ever assembled outside the United States.

What happened

DeepSeek began taking outside capital for the first time this year, initially setting a 50 billion yuan target for the current round. State-backed funds, the investment arms of listed Chinese companies and venture firms have since oversubscribed the deal, and the company now expects the final tally to approach 100 billion yuan. Two names have emerged as the largest backers: WeChat operator Tencent and battery maker CATL, both of which Bloomberg reported had committed capital earlier on Tuesday. CNBC's sources added that CATL, Geely Auto, Monolith Management and Loyal Valley Capital have all signed on.

DeepSeek is screening investors carefully. It is turning away private funds raised from individual investors and limiting participation largely to government and corporate money, the people said. The round was initially expected to close by the end of August, but dragged on as some prospective backers balked at the price. DeepSeek is seeking a valuation of roughly 500 billion yuan, or about $75 billion, ahead of a possible listing on Shanghai's STAR Market, and has hired CITIC Securities to prepare. Bloomberg reported the fresh money could lift the company's post-money valuation well past that mark.

Why it matters

The timing is not accidental. DeepSeek's September release of V4.1 Flash, a model the company says is faster and cheaper to serve, reset expectations for how much performance open-weight systems can deliver per dollar. Investor interest followed the model, not the other way around. Completing a round at this size would give DeepSeek a war chest for compute, talent retention and a data centre build in Inner Mongolia that reportedly leans on 160,000 Huawei AI chips, sidestepping US export controls on advanced Nvidia hardware.

The builder angle

For teams building on AI, the signal is the price floor. This funding is a bet that open-weight models keep compressing the cost of capable inference, so capital is being raised to defend that position rather than to monetise quickly. Founder Liang Wenfeng has said he prioritises pushing toward AGI over profit, and the round was briefly paused after he told investors that Chinese labs still lag their US counterparts and depend on Nvidia chips. Combined with rival Moonshot AI reportedly targeting its own early-2027 IPO, the picture is a Chinese cohort scaling through public markets while its models stay cheap and downloadable. Builders evaluating model providers should read this as evidence that the open-weight tier is becoming a durable supply layer, not a temporary discount.