China’s AI Talent Crunch: DeepSeek Sparks Fierce Competition for Top Minds
February 21st 2025
China is experiencing a significant talent shortage as the demand for AI professionals surges, driven by the rapid rise of DeepSeek and the broader expansion of the AI sector. Headhunters, such as Jason Yang from Shenzhen’s Touch HR, are scrambling to recruit top-tier talent both locally and overseas. Many candidates are ethnic Chinese with advanced degrees from top universities in China and the U.S., often holding PhDs.
DeepSeek’s recent success with its open-source reasoning model has invigorated China’s AI ecosystem, intensifying competition among companies like Xiaomi, Alibaba, and ByteDance. The result has been soaring salaries - AI PhD graduates now command annual salaries between 800,000 and 1 million yuan, while DeepSeek offers up to 1.54 million yuan for deep learning researchers.
China’s talent gap is stark. By 2030, the country will need six million AI professionals but is expected to fall short by four million workers, according to a McKinsey report. Despite initiatives from over 500 universities offering AI majors since 2018, China still struggles to produce enough PhD-level AI experts.
While China’s tech giants actively poach talent from rivals, some experts warn of rising costs and shrinking profit margins. Australian economist Kailing Shen notes that AI’s long-term adoption will be non-linear, with market saturation eventually driving down salaries.
Moreover, U.S. chip curbs threaten China’s ability to scale its AI infrastructure, with computing power and data accumulation posing significant challenges. Still, industry experts like Renmin University’s Prof. Zeng Xiangquan believe China’s AI talent shortage could be resolved within five to ten years as education systems catch up with market needs.
This talent race highlights a broader trend: AI is not just reshaping technology but transforming the global job market, creating new opportunities while intensifying competition for specialized skills.
Source: Straits Times
