China AI model developer startup DeepSeek is in talks for its first venture funding round, and its valuation has jumped in a matter of weeks from $20 billion to $45 billion, the Financial Times and Bloomberg News reported on Tuesday.

The round is led by China’s state semiconductor investment fund, China Integrated Circuit Industry Investment Fund, and Tencent and Alibaba are also in discussions to take part, the reports said.

DeepSeek drew attention in early 2025 after releasing a large language model trained with far fewer computing resources and costs than major U.S. models. DeepSeek models are available for free on Hugging Face under an open-weights approach.

DeepSeek was founded by Chinese hedge fund billionaire Liang Wenfeng (량원펑) and has so far kept its distance from raising external funding.

The Financial Times, citing sources, said Liang decided to raise funds to offer shares to employees after facing a situation in which rivals were poaching DeepSeek researchers.

China is supporting domestic AI development with funding as it struggles to secure U.S. technology, especially semiconductors. DeepSeek’s recently unveiled V4 model runs efficiently on chips from China’s Huawei, and the combination is seen as a strong pillar in China’s AI development strategy to stand up to the United States, the Financial Times said.