The Chinese AI startup is raising $7.4 billion not for servers or compute, but to hand out equity and keep its 150-person team from walking out the door.
DeepSeek, the Chinese AI startup that shook the industry with its efficient open-source models, is doing something unusual with its first-ever external funding round. It’s telling the very investors writing checks not to recruit its staff.
The instruction underscores a reality that founder Liang Wenfeng has been grappling with for months. In a market where Alibaba, ByteDance, and other cash-rich Chinese tech giants are aggressively hunting AI talent, DeepSeek’s lean team of roughly 150 employees has become a prime target. The fundraise isn’t really about buying more GPUs or scaling infrastructure. It’s about survival of the roster.
A valuation that tripled in two months #
Back in April 2026, DeepSeek initially targeted a $20 billion valuation for its maiden funding round. By May, valuation discussions had climbed to around $45 billion, with China’s so-called “Big Fund” reportedly involved in talks. Then came the latest figures: DeepSeek is now set to close a roughly $7.4 billion round that would value the company at somewhere between $52 billion and $59 billion post-money.
The investor roster reflects that momentum. Tencent and CATL, the world’s largest electric vehicle battery maker, are among those participating in the round.
The real product is the people #
Liang Wenfeng had previously rejected outside capital entirely. His reasoning was straightforward: external investors bring external pressure, and that pressure could compromise the company’s research-first, “science-project ethos,” as it’s been described.
DeepSeek’s hiring model is distinctive. The company recruits heavily from recent graduates, prioritizing passion and raw ability over years of experience. It pairs those hires with competitive compensation packages and access to GPU resources. But Alibaba and ByteDance, both sitting on massive war chests, have been running aggressive recruitment drives targeting AI talent across China.
The $7.4 billion round is designed to solve this problem directly. The primary goal is issuing equity and stock options to existing employees.
A broader talent war with global implications #
The no-poaching instruction to investors is a practical acknowledgment of this dynamic. Venture capital and strategic investors often maintain deep networks across their portfolio companies. Liang is essentially drawing a line: you can own a piece of the company, but you can’t dismantle the team that makes it valuable.
This approach also reveals something about DeepSeek’s competitive moat. The company’s breakthrough moments, including its R1 reasoning model that rattled US tech stocks earlier in the year, weren’t built on proprietary data advantages or massive compute budgets. They were built on clever engineering by a small group of talented people.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our