China Integrated Circuit Industry Investment Fund, commonly known as the 'Big Fund', is in negotiations to lead the first financing round for AI lab DeepSeek. According to the Financial Times, the deal could value DeepSeek at about $45 billion.
Potential investors and deal specifics
Tencent is also reportedly considering participating in the funding round, but the final investor group has not been decided. DeepSeek's founder, Liang Venfeng, may personally invest in the round; he currently owns 89.5 percent of the company.
Rapid rise in valuation
The company's valuation has jumped sharply in a short period: only a few weeks earlier, when talks began, market estimates placed DeepSeek at roughly $20 billion. What was initially planned as a token fundraising to assign market value to employee share options may now become a substantially larger financing round that could fund expanded compute capacity and other operational needs.
Technology and products
DeepSeek became widely known in January 2025 after unveiling R1, an open-source large language model; the company claims R1 was trained using a fraction of the compute required by some U.S. competitors such as OpenAI. Since then, the lab has focused on developing cutting-edge AI models and has not yet built a significant commercial business.
The firm's latest V4 model has been optimized for Huawei Ascend 950PR chips, with tuning specifically aimed at inference — the computation stage when the model generates responses.
Why Big Fund support matters
Backing from the Big Fund would materially strengthen DeepSeek's position within China's AI development ecosystem and could accelerate the creation of an ecosystem composed of purely Chinese models, software and chips. That aligns with Chinese leadership priorities on technological self-reliance, championed by President Xi Jinping.
The Big Fund's previous activity
The Big Fund has raised capital in three phases to support China's semiconductor industry. In its third round in 2024 it secured $47 billion from the finance ministry, local governments and state-owned banks, deploying that capital mainly into semiconductor equipment and materials. To date, the fund has not supported any company focused on developing large language models.
International context and competition
Sales of Huawei's Ascend 950PR chips have surged this year, and in China they have overtaken Nvidia's shipments, which continue to face U.S. export restrictions. Nonetheless, China’s overall AI-chip production still trails that of the United States, and Chinese processors remain several generations behind top U.S. designs in technological maturity.
Beijing is therefore betting that close cooperation between chipmakers and model developers can create an independent domestic ecosystem that sustains China’s competitiveness in AI despite export controls. Jensen Huang, Nvidia’s CEO, has warned about this risk — noting that the day DeepSeek runs effectively on Huawei hardware would have serious implications for the United States.
Founder’s original plan and current funding purpose
Liang Venfeng originally intended a symbolic capital raise to put an explicit market value on DeepSeek’s stock options and help retain key researchers. With the company’s valuation having risen quickly, the present funding round may be much larger and provide the extra capital needed to expand computing resources.
Next steps
Negotiations are ongoing and the final investor lineup and transaction terms have not been disclosed. If the deal proceeds, DeepSeek’s resources and standing within China’s AI ecosystem could be substantially strengthened.
