While the US currently leads in AI with superior chips, China's state-controlled power grid is growing 10x faster and can be directed towards AI data centers. This creates a scenario where if AGI is a short-term race, the US wins. If it's a long-term build-out, China's superior energy infrastructure could be the deciding factor.

The primary constraint on US AI leadership relative to China isn't the ability to build models, but the slow pace of developing necessary compute and energy infrastructure. China faces fewer regulatory barriers, allowing it to scale these critical inputs more rapidly.

China cannot overcome its semiconductor disadvantage by simply applying more energy to its lagging-edge chips. No frontier AI model has been trained on hardware older than 5nm, suggesting leading-edge nodes provide an essential, non-linear advantage in training efficiency that cannot be compensated for with sheer power, a major hurdle for China's AGI ambitions.

While the US faces power constraints, China can build new energy sources like nuclear power plants in just a few years. This ability to rapidly scale power gives it a fundamental, underappreciated advantage in the energy-intensive AI war, alongside its talent pool and government support.

Beyond algorithms and talent, China's key advantage in the AI race is its massive investment in energy infrastructure. While the U.S. grid struggles, China is adding 10x more solar capacity and building 33 nuclear plants, ensuring it will have the immense power required to train and run future AI models at scale.

A critical, under-discussed constraint on Chinese AI progress is the compute bottleneck caused by inference. Their massive user base consumes available GPU capacity serving requests, leaving little compute for the R&D and training needed to innovate and improve their models.

The performance gap between Chinese and American frontier AI models is not due to a lack of talent or different training techniques. Instead, it is primarily constrained by access to massive-scale compute and the capital required to procure it.

Despite developing frontier-level AI models like Kimi K3, Chinese labs are severely compute-constrained. The Kimi K3 launch quickly overwhelmed servers, revealing a lack of GPU infrastructure for large-scale inference. This shifts the US-China competition focus from model benchmarks to industrial capacity and data center dominance.

China's superior ability to rapidly build energy infrastructure and data centers means it could have outpaced US firms in building massive AI training facilities. Export controls are the primary reason Chinese hyperscalers haven't matched the massive capital spending of their US counterparts.