Chinese AI lab DeepSeek is close to completing a fundraising round worth roughly 100 billion yuan — about $15 billion — double the size it originally sought, according to Reuters reporting citing two people familiar with the matter. The round, expected to conclude in the coming days, would rank among the largest private fundraisings in the Chinese AI sector and underscores how much capital the race to build frontier models now demands.
The figure has swelled dramatically since the round first came into view. Earlier this month, reports indicated DeepSeek was targeting at least $12 billion from investors including Tencent and battery giant CATL. Reuters now reports the round stands at approximately 100 billion yuan, twice the company's original 50 billion yuan target. Bloomberg first reported the $15 billion scale of the fundraising, citing sources familiar with the matter. For more context on this story, see our ongoing artificial intelligence updates.
Both outlets caution that the final details remain subject to change, and DeepSeek has not publicly confirmed the reported figures.
Tencent and CATL Step Up for China's AI Champion
What makes the round unusual is the composition of its backer list. Rather than relying primarily on specialist venture capital, DeepSeek has drawn in some of China's largest corporations: Tencent Holdings, the gaming and social media giant behind WeChat, and Contemporary Amperex Technology (CATL), the world's dominant electric vehicle battery maker, are among the largest investors in the round, according to one of Reuters' sources.
The participation of industrial and consumer giants signals how far AI investment appetite has spread beyond the technology sector in China. State-backed entities and large corporate groups seeking exposure to the AI buildout are joining the cap table, Reuters reported — a pattern that mirrors, in its own way, the flood of non-tech capital flowing into AI infrastructure in the United States.
DeepSeek, Tencent, and CATL had not responded to requests for comment at the time of the report.
An IPO Moves Closer
The fundraising also advances DeepSeek's path toward the public markets. Reuters reported in September that the company had hired CITIC Securities, one of China's leading investment banks, to prepare for a potential initial public offering on the STAR Market, Shanghai's tech-focused listing board. AI Buzz Wire covered those IPO preparations earlier this month, when reports indicated a 2027 listing timeline was under discussion.
A listing would hand DeepSeek a renewable source of capital for research, computing infrastructure, and model development — resources that matter given the scale of spending among its competitors. The company's home city of Hangzhou has been positioning itself as a hub for China's AI industry, and a successful STAR Market debut would make DeepSeek one of the most valuable Chinese AI companies to go public.
The timing of any listing remains unclear, and the structure of the company's shareholding would need to be sorted out before regulators sign off.
Where the Money Goes: Models, Chips, and the US Rivalry
DeepSeek's technical output has continued through the fundraising process. The company released its latest model, DeepSeek-V4.1-Flash, in September, which it says was designed for greater capability, faster inference, higher throughput, and improved scaling to larger models.
The capital comes as Beijing pushes to strengthen China's position in the global AI race against US firms, and as the cost of staying competitive climbs relentlessly. The contrast with American rivals is stark: Meta alone reported $34.9 billion in capital expenditures last quarter as it builds out AI infrastructure, pushing its free cash flow negative for the first time since 2012. Against that backdrop, a $15 billion war chest reads less like extravagance and more like table stakes for a lab with frontier ambitions.
DeepSeek has also been diversifying its compute base, with earlier reporting detailing the company's work with Huawei on Ascend-based training infrastructure and open-source software aimed at reducing reliance on NVIDIA's CUDA ecosystem. Earlier coverage of the funding round also flagged plans for a data center housing some 160,000 Huawei chips, one of the largest single-site deployments of domestic accelerator hardware reported anywhere. NVIDIA, meanwhile, is courting European buyers at its GTC Berlin event this week — evidence that the compute race DeepSeek is buying into now spans continents.
The company's rise has been one of the sharper plot twists in the industry. Emerging from the stable of quant hedge fund High-Flyer, DeepSeek built a reputation for extracting frontier-adjacent performance from constrained compute, then open-sourcing much of its work. That formula shook US markets in early 2025 and continues to shape how both Chinese and American labs think about efficiency — a pedigree that helps explain why investors were willing to double the check.
Relief and Rough Edges
The funding news arrived alongside a reminder of the operational strain rapid growth brings: Yicai Global reported Thursday that DeepSeek suffered one of its worst service outages of the year, disrupting access for users of its chatbot and developer APIs. The company has previously attributed capacity problems to surging demand for its models.
Whether DeepSeek can convert the fresh capital into a durable advantage will depend on execution: delivering model advances, scaling infrastructure without repeated outages, and navigating an IPO process under intensified scrutiny of Chinese tech listings. But with the round closing at roughly double its original size, investors are clearly betting the company can.
Stay Ahead of AIFollow AI Buzz Wire for breaking AI news, model releases, and industry analysis.
Read more AI news →
