Key Points:
- Chinese AI startup DeepSeek is advancing a massive 70 billion yuan ($10 billion) funding round.
- The financing deal will value the Hangzhou-based open-source pioneer at about $45 billion to $50 billion.
- Founder Liang Wenfeng is contributing up to 20 billion yuan of his own money to control 84.29% of the company.
- Senior management has prioritized achieving artificial general intelligence (AGI) over short-term commercialization.
DeepSeek, the Chinese artificial intelligence startup that upended Silicon Valley last year, is currently advancing a massive 70 billion yuan fundraising round. This translates to roughly $10 billion. The startup’s founder, Liang Wenfeng, has privately told potential investors that the company will prioritize groundbreaking research and the pursuit of artificial general intelligence (AGI) over short-term commercialization and quick profits.
The massive deal is in its final stages. If completed, the funding round will value the Hangzhou-based artificial intelligence company at about $45 billion to $50 billion before the investment. This would set a record for the largest first-time financing round by a Chinese technology startup, proving that the era of low-cost, self-funded artificial intelligence miracles is giving way to massive infrastructure demands.
The state-backed National Artificial Intelligence Industry Investment Fund is in advanced talks to lead the round with an investment of about 10 billion yuan, or roughly $1.38 billion. Beijing established this state-backed vehicle specifically to propel strategic technology projects and build domestic computing stacks. Other notable participants close to securing their slots include the Chinese internet giant Tencent Holdings and the prominent venture firms IDG Capital and Monolith Capital.
Surprisingly, the largest single check in this historic round is coming from the founder himself. Liang Wenfeng reportedly plans to inject up to 20 billion yuan of his own money into the startup, representing about 40% of the total fundraising. By injecting this capital directly, Liang will raise his direct personal stake from 1% to 34%. When factoring in his indirect holdings through parent companies, Liang will control approximately 84.29% of DeepSeek.
For years, DeepSeek adhered to a strict policy of rejecting outside capital. Liang spun out the startup from his quantitative hedge fund, High-Flyer Capital Management, in 2023. High-Flyer, which managed over 100 billion yuan at its peak, entirely funded DeepSeek’s early operations. This allowed it to focus strictly on open-source research without commercial pressures. However, the soaring cost of modern computing power and a recent wave of talent attrition have forced the company to change its course.
In recent months, several core researchers left DeepSeek to join rival Chinese technology giants like Xiaomi and ByteDance. The startup’s management team hopes this new funding round will establish a clear, public market valuation for employee stock options. By giving employees highly valuable, liquid equity, the company hopes to retain its core talent pool and prevent further staff departures.
DeepSeek shocked global technology markets in January 2025 when it launched its R1 reasoning model. R1 matched the performance of OpenAI’s top systems but reportedly cost only $6 million to train. This ultra-efficient architecture used a “mixture-of-experts” design that activated only a fraction of its total parameters, causing Western tech giants to scramble and re-evaluate their massive infrastructure spending assumptions.
The startup has continued to prove its technical relevance on the global stage. In late April 2026, DeepSeek released a preview of its V4 model just hours after OpenAI announced GPT-5.5. The new V4 model demonstrated impressive efficiency, requiring only 27% of the computing power for long-context tasks compared to its predecessor. This constant back-and-forth illustrates the relentless pace of the global artificial intelligence race.
The funding round also highlights a major strategic shift in how China builds its artificial intelligence industry compared to the United States. While Western AI development relies on mega-corporations like Microsoft, Google, and Amazon recycling cloud profits, China is increasingly using state-backed capital to sponsor domestic champions. This hybrid model allows private labs to secure their local supply chains and close the technological gap with the West.
Liang Wenfeng’s message to investors makes it clear that DeepSeek remains focused on the ultimate prize of building artificial general intelligence. While Western competitors face immense pressure to show commercial profits on their investments, DeepSeek will continue to use its massive new war chest to push the physical boundaries of open-source technology.
