China calls for AI self-reliance to counter U.S. tech curbs

China's top leadership just made its AI playbook crystal clear: double down on self-reliance and build a homegrown tech stack from the silicon up.

State media reported on Saturday that President Xi Jinping called for self-reliance and self-strengthening in artificial intelligence during a Politburo study session, urging the country to leverage its new whole national system to push development forward.

The tone was notably pragmatic. Xi acknowledged that China must recognize the gaps and redouble its efforts, promising state policy support across government procurement, intellectual property rights, research, and talent cultivation.

Silicon, software, and the DeepSeek effect

For nearly two years after OpenAI released ChatGPT, the conventional wisdom was that aggressive U.S. export controls on advanced chips would stall China's AI ecosystem.

That narrative cracked wide open in January when Chinese startup DeepSeek rolled out an advanced reasoning model. DeepSeek claimed the system was trained on less advanced chips at a fraction of the cost of Western competitors, proving that infrastructure software engineering could compensate for raw hardware shortages.

Xi pointed directly at that foundation on Friday. He stressed that China must concentrate its efforts on mastering core bottlenecks like high-end chips and basic software, aiming to build an independent, controllable, and collaborative hardware and software ecosystem.

He also pushed for faster AI legislation, calling for a risk warning and emergency response system to keep the technology safe and controllable, echoing his previous stance that AI should not become a game of rich countries.

Why it matters

We think this is a crucial acknowledgment of the post-DeepSeek reality. China is no longer just complaining about U.S. sanctions; it is turning efficiency-first engineering into official national policy.

For enterprise tech and developers in China, government procurement will now aggressively favor domestic hardware and homegrown software infrastructure. Even if local chips lag Nvidia's bleeding edge, the state is creating guaranteed domestic demand to force those platforms to mature.

Globally, this solidifies the bifurcation of the AI stack. The West is pursuing brute-force scale with trillion-dollar compute clusters, while China is attempting to leapfrog compute deficits through surgical architecture tweaks and state-coordinated software layers.

Building an entire independent silicon-to-software stack is notoriously grueling, and catching up on leading-edge fabrication remains China's biggest hurdle. But by aligning state resources behind foundational engineering rather than just surface-level wrapper apps, China is ensuring that the transatlantic AI moat will be much harder to sustain.