As Xi Jinping visits DC, Chinese economists warn that Beijing is pouring too many resources into AI, which creates few jobs, and is doing little for the economy
Context & Ripple Effects
Beijing had already framed its technology strategy around practical AI applications rather than Silicon Valley-style AGI pursuits, while also showcasing AI self-sufficiency efforts alongside investment in power generation and skills training. The economists' criticism tests whether that industrial strategy is delivering broad economic value rather than only technical capacity.
The warning follows a February debate over AI productivity gains versus automation-driven job displacement and renewed concern that workers may bear the costs of the national AI mission. It makes employment outcomes a central measure of success for Beijing's AI allocation.
First-order effects
- Beijing's AI investment program faces unusually public criticism from economists over its job creation and economic payoff, raising the political importance of demonstrating practical returns.
- Chinese workers worried about automation gain a domestic economic critique of an AI push that has been presented as a national priority.
Second-order effects
- The criticism strengthens the case for Xi's application-oriented technology strategy, putting more pressure on AI projects to show deployment beyond compute and infrastructure buildout.
- Skills-training and power investments associated with AI self-sufficiency become more consequential: they must support employment and productive use, not merely expand technical capacity.
Third-order effects
- If employment concerns continue to constrain the AI program, China's industrial policy may shift from maximizing AI capacity toward judging projects by diffusion into job-supporting applications.
- The episode points to a broader tension in national AI strategies: concentrated infrastructure investment can advance capability while leaving governments accountable for uneven labor-market gains.
The trend: National AI strategies are increasingly being judged not only by compute and model capability, but by whether their economic benefits reach workers and operating industries.