DeepSeek in talks for first venture round as implied valuation jumps to $45B

DeepSeek is in discussions to raise its first venture capital round, with sources saying investor demand has driven the company’s implied valuation from roughly $20 billion to about $45 billion in a matter of weeks. The talks would mark the lab’s first acceptance of outside capital after operating without external investors since its launch.

The lab emerged in early 2025 after releasing a large language model that was trained using a fraction of the compute and at a fraction of the cost of major U.S. models from OpenAI and Anthropic. Some versions of DeepSeek’s models have been published on Hugging Face; evaluations cited by sources say they keep reasonable pace with top models on tasks such as reasoning and coding.

Sources say the fundraising round is said to be led by the state investment vehicle China Integrated Circuit Industry Investment Fund. Chinese backers are described as motivated to develop homegrown AI to reduce reliance on U.S. technology and chips. DeepSeek’s models have been optimized to run on chips made by Huawei, and cloud operators Tencent and Alibaba are reportedly in talks to participate in the deal.

Founder Liang Wenfeng, who controls nearly 90% of DeepSeek, had previously avoided outside investors; sources told reporters the decision to pursue external funding reflects pressure from competitors poaching researchers and a need to offer employees equity. Backing from the state fund and major cloud players could strengthen China’s domestic AI stack and lessen dependence on foreign suppliers. DeepSeek could not be immediately reached for comment.