DeepSeek Is Raising $12 Billion as China's AI Bet Gets Bigger

Chinese AI startup DeepSeek is closing in on one of the largest private funding rounds the industry has ever seen, with more than 80 billion yuan, roughly $12 billion, committed by investors including Tencent and battery giant CATL ahead of a possible domestic IPO.

The round has blown past DeepSeek's original target of about 50 billion yuan, according to people familiar with the fundraising. Reuters reported that the company expects to complete the round in October, with investor commitments already exceeding its initial goal. Bloomberg reported that the final tally could approach 100 billion yuan, or roughly $15 billion.

The news comes less than two weeks after DeepSeek reached a $1 billion annualized revenue run rate, more than double its pace from just a few months earlier, as rising demand for its models and higher prices turn last year's breakout into a sizable business.

DeepSeek began seeking new capital in July at a targeted valuation of 500 billion yuan, or about $74 billion, just a month after raising roughly $7.4 billion in its first external funding round. Tencent and Contemporary Amperex Technology Co. Ltd., better known as CATL, are among the largest investors in the latest financing.

The models are closing the gap

The fundraising momentum follows the release of DeepSeek-V4.1-Flash in September, which the company said was built for higher throughput, faster inference and scaling to larger workloads. Bloomberg Intelligence said this week that the performance gap between leading Chinese and U.S. AI models has narrowed to about 3 percent on benchmark scores, down from roughly 9 percent in May, with DeepSeek among the companies driving the gains.

The new capital comes as DeepSeek prepares for a potential listing on Shanghai's STAR Market. The company has hired CITIC Securities to work on IPO preparations, though the timing, valuation and size of the offering have yet to be finalized, Reuters reported.

Building the stack at home

DeepSeek has been moving deeper into China's effort to build an AI stack that depends less on U.S. technology. The startup recently announced a partnership with Huawei to develop programming tools optimized for Huawei's Ascend AI chips, including open-source software aimed at giving developers an alternative to Nvidia's CUDA ecosystem, which remains the dominant software platform for AI accelerators.

The company is also building a data center with at least 160,000 AI chips from Huawei, and is working on its own inference chip to reduce dependence on both Nvidia and Huawei. Founder Liang Wenfeng has said he wants to keep developing open models with freely available weights.

The round has not been without drama. DeepSeek briefly paused the fundraising after comments by Liang about the company's reliance on Nvidia chips went viral, a reminder of how politically charged the U.S.-China chip relationship has become. The pause did not last. Investor demand, stoked by the V4 releases and the revenue inflection, pulled the process back on track within weeks.

For U.S. observers, the raise is another data point in a year that has steadily eroded assumptions about America's AI lead. When DeepSeek released its R1 model in early 2025, it was treated as a one-off efficiency breakthrough. Twenty months later, the company is raising at a scale that rivals the largest U.S. AI financings, with a revenue base to help justify it. The question for 2027 is no longer whether Chinese labs can keep pace, but whether they can do it on an increasingly domestic technology stack.

That makes the latest funding round about more than financing another generation of models. A $12 billion raise would give DeepSeek a substantial war chest as China tries to reduce its reliance on U.S. suppliers, and if the round reaches the reported 100 billion yuan upper range, the company could enter its expected 2027 IPO with close to $15 billion in fresh capital from this financing alone.