With the accelerated implementation of artificial intelligence applications, as of June of this year, the average daily call volume of words in China has exceeded 500 trillion dollars. Behind the rapid growth in the scale of terms, competition for large global models has heated up, and demand for computing power has exploded. A CCTV financial reporter learned that currently the global big model is in the most intense competition and the fastest iteration, while the big Chinese model is already in the first tier of global competition. Industry insiders said that currently, the flagship model of artificial intelligence is updated almost on a monthly basis, and the focus of competition is shifting from simple “model intelligence” to comprehensive competition between smart device implementation and ecological construction. Major manufacturers continue to increase investment in training, and the number of model parameters and context windows continue to expand. At the same time, artificial intelligence is moving from “being able to chat” to “being able to work”, and smart devices are beginning to be implemented on a large scale. One task requires repeated search, reading context, adjustment tools, and multiple rounds of feedback, and demand for inference computing power has exploded. In order to compete for computing power resources, major Internet companies have accelerated the deployment of AI intelligent computing centers and signed long-term cooperation agreements to lock in resources.

With the accelerated implementation of artificial intelligence applications, as of June of this year, the average daily call volume of words in China has exceeded 500 trillion dollars. Behind the rapid growth in the scale of terms, competition for large global models has heated up, and demand for computing power has exploded. A CCTV financial reporter learned that currently the global big model is in the most intense competition and the fastest iteration, while the big Chinese model is already in the first tier of global competition. Industry insiders said that currently, the flagship model of artificial intelligence is updated almost on a monthly basis, and the focus of competition is shifting from simple “model intelligence” to comprehensive competition between smart device implementation and ecological construction. Major manufacturers continue to increase investment in training, and the number of model parameters and context windows continue to expand. At the same time, artificial intelligence is moving from “being able to chat” to “being able to work”, and smart devices are beginning to be implemented on a large scale. One task requires repeated search, reading context, adjustment tools, and multiple rounds of feedback, and demand for inference computing power has exploded. In order to compete for computing power resources, major Internet companies have accelerated the deployment of AI intelligent computing centers and signed long-term cooperation agreements to lock in resources.