Advanced AI models require massive computing power, and that means high-end chips. This is where China faces its greatest vulnerability. U.S. export controls have made it harder to access cutting-edge semiconductors, slowing progress in training powerful foundation models. Still, China’s response has been twofold: the government is doubling down on industrial policy to support local champions, while private players like Moore Threads and Enflame push toward IPOs to expand domestic capabilities.
China has laid out an ambitious blueprint to build a national computing backbone connecting data centers across regions and optimizing energy usage. But the rollout is bumpy. Power supply mismatches, technical fragmentation, and institutional inefficiencies hamper the system’s effectiveness. Without faster implementation, the gap between ambition and capacity will persist.
China holds a clear advantage in raw data volumes. Yet in training large-scale AI models, quality trumps quantity. Much of the available data is fragmented, low-quality, or siloed. This limits the reasoning capabilities of Chinese AI systems. Recognizing the problem, Beijing is ramping up efforts to improve access to curated, high-quality training datasets an essential step if China wants to keep pace in AGI.
China’s model landscape is moving fast. Companies like DeepSeek are leading breakthroughs in efficient training and open-source model development. DeepSeek-R1, for instance, achieved top-tier performance with relatively limited compute. This trend (fueled by competition and a new openness) positions China to play a more visible role in the global AI community, not just as a user, but as a contributor.
Where China already shines is in rolling out AI at scale. From consumer apps and enterprise tools to urban systems and robotics, the country excels at moving from concept to implementation. The goal now: replicate the “killer app” playbook in the AGI era, expanding from digital agents to embodied intelligence across industries.
This whitepaper breaks down China’s AGI strategy across five key pillars: Chips, Computing, Data, Models, and Applications.
Focusing on Large Models as the dominant technological path, we assess China’s strengths, weaknesses, and strategic potential. The paper intentionally avoids speculative debates about AGI risks and instead offers grounded insight for multinational companies seeking to understand China’s evolving AI ecosystem.
