A recent report from a Washington-based research group highlights the current state of China’s artificial intelligence infrastructure. According to the findings, the total computing power available to Chinese entities is expected to rise from 4.6 million to 8.8 million 'H100-equivalents' by 2027. This metric is used to benchmark performance against Nvidia’s high-end data-center graphics processing units.

Despite the significant expansion in capacity, the report underscores a persistent dependency on foreign technology. US-designed chips are projected to make up roughly 65% of China's total AI computing capacity in the coming years. This reliance persists even as China seeks to bolster its domestic semiconductor industry and cloud computing capabilities. The data suggests that while China is successfully increasing its access to AI resources, it has yet to achieve self-sufficiency in the hardware necessary to power advanced artificial intelligence models. The report also notes that offshore cloud access remains a critical component of China's current strategy to meet its growing computational demands.