Logo of DeepSeek Photo: VCG

Chinese AI startup DeepSeek reportedly plans to expand its latest funding round to as much as 100 billion yuan ($15 billion), aimed at expanding model training, computing infrastructure and talent recruitment, according to a Shanghai Securities News report on Thursday.

The company's fundraising target had been revised upward in recent discussions, from an initial 50 billion yuan to around 100 billion yuan, Shanghai Securities News reported, citing a source familiar with the matter.

According to the report, Tencent and battery giant Contemporary Amperex Technology Co (CATL) are said to be among the largest prospective investors in the round, which market sources said could be finalized in the coming days. 

DeepSeek did not comment on the media reports when reached by the Global Times.

According to earlier media reports, DeepSeek completed its first financing round in June at a valuation above $50 billion. That round reportedly included major backing from founder Liang Wenfeng, Tencent, CATL-related entities, NetEase, JD.com, and several investment funds.

Computing power is emerging as the central priority in this funding round. Earlier reports said that DeepSeek is building a large-scale data center in Ulanqab, North China's Inner Mongolia Autonomous Region. The Shanghai Securities News report cited an industry analyst as saying that DeepSeek's model parameter scale during training is believed to have already surpassed 2 trillion, and that by next year, most leading domestic models are expected to expand to the 5 trillion to 8 trillion parameter range, with some possibly reaching around 10 trillion. 

Against that backdrop, spending on computing power is likely to rise sharply, potentially by several times. 

Some analysts suggest that DeepSeek's foundation models and open-source full-stack R&D capabilities are leading in China, which is also an important reason why the company was able to secure such a major investment.

On October 17, DeepSeek released DeepSeek-V4.1-Flash-1017, an updated AI model promising stronger benchmark performance, faster task completion, and shorter outputs than the previous version while lowering usage costs. According to the company, the updated model beat GPT 5.6 Sol.

The latest financing buzz also reflects the continuing race among Chinese AI firms to release new products and keep pace with fast-moving global competition.

Over the past year, major domestic players have rolled out upgraded foundation models, multimodal systems, agent-based tools and enterprise AI solutions. The industry's focus has shifted from simply building large models to improving inference efficiency, lowering training costs and embedding AI into real-world applications such as office software, search, cloud services, education, finance and industrial automation, Tian Feng, president of the Fast Think Institute and former dean of the Intelligence Industry Research Institute at SenseTime, told the Global Times.

Global Times