On September 10, during the 2026 China Unicom Partner Conference, Chen Weiliang, the founder of MetaX Inc., gave a speech. He highlighted that AI is experiencing rapid development, propelled by the synergistic effects of data, computing power, and algorithms. Nevertheless, the computing power industry is confronted with a dual challenge: there exists a substantial deficit in computing power on the demand front. The parameters of large models have soared to the trillion or even hundred-trillion mark, with the computing power consumption for both training and inference escalating exponentially. Simultaneously, as AI applications transition from the laboratory to large-scale commercial deployment, the utilization of Tokens has surged, resulting in a persistent shortage of computing power supply.
