China's AI Hardware Lags US Despite Model Competitiveness

Chinese artificial intelligence models have become increasingly competitive with their US peers, but the country’s AI hardware still lags significantly behind. While domestic chips are now widely adopted for model inference, none of China’s top models are known to have been pre-trained on homegrown silicon.

To understand this gap, it helps to look at the three stages of AI model development. First is pre-training, the most computationally demanding phase, where a model feeds on massive data sets to learn basic patterns. Next is post-training, a less intense process that fine-tunes the model to follow specific human instructions.

Finally comes inference, the everyday act of running the finished AI to answer user queries and instructions.

Driven by Washington’s escalating export controls and Beijing’s push for technological self-sufficiency, a growing number of Chinese AI labs are now experimenting with shifting these earlier training phases onto domestic hardware.

While relying on indigenous suppliers meant Chinese AI labs “may not develop as quickly and efficiently as their US counterparts”, in the long run, the country was building an entire domestic AI supply chain, which was “quite rare worldwide”, said Natixis economist Gary Ng.

Here is how China’s latest AI models are using domestic computing hardware across these different stages.