In the fast-moving world of artificial intelligence, a technical report is usually a victory lap for engineering prowess. However, DeepSeek’s recent release of its V4 model technical report has sparked a different conversation. Among the nearly 300 names listed as contributors, ten are marked with an asterisk—a footnote indicating they have already departed the firm. This rare moment of corporate transparency highlights a growing crisis within China’s premier AI startup: its struggle to retain top-tier talent in an environment of predatory recruitment.
DeepSeek has earned a reputation as the ‘Whampoa Military Academy’ of Chinese AI, a nickname referencing the historic officer training school. High-profile departures include core researchers like Wang Bingxuan, who joined Tencent, and Guo Daya, who was poached by ByteDance’s Seed team. These researchers were instrumental in developing the foundation, reasoning, and multimodal capabilities that allowed DeepSeek to punch above its weight class, often rivaling Silicon Valley giants on a fraction of the budget.
While DeepSeek’s technical philosophy is celebrated for its efficiency, its business model creates a distinct disadvantage in the talent market. Unlike rivals like Moonshot AI, which has raised over $2.5 billion, DeepSeek remains entirely self-funded by its parent company, High-Flyer Quant. Founder Liang Wenfeng has famously rejected external venture capital to avoid the pressures of short-term commercialization. However, without the high-valuation stock options offered by VC-backed competitors, DeepSeek struggles to compete with the ‘golden handcuffs’ used by China’s internet titans.
The talent war is not just anecdotal; it is structural. Market data reveals that new AI job postings in China have surged twelvefold in early 2026, with average monthly salaries for core roles exceeding 60,000 RMB. For senior experts in deep learning, annual packages now reach as high as 3 million RMB. As Tencent, Alibaba, and Xiaomi aggressively scale their AI departments, the pressure on independent labs to offer liquid equity and competitive bonuses has become an existential threat to technical autonomy.