Industrial News

Analysis of the size of the intelligent computing power market in 2024: The size of the intelligent computing power market will reach 70 EFLOPS 

Intelligent computing power has the advantages of high performance, low energy consumption and flexibility, and the application scenarios of intelligent computing power have also been broadened under the continuous progress of science and technology, and the intelligent computing power market in first-tier cities is developing rapidly. The following is an analysis of the size of the intelligent computing power market in 2024.

The market size of intelligent computing power  

After years of development, China's computing infrastructure construction has reached the world's advanced level, with the total scale of computing power second only to the United States and ranking second in the world. As of 2023, the number of racks in use that provide computing power services in China will reach 8.1 million standard racks, more than 5,000 computing power providers of various types, and the computing power scale will reach 230EFlops (exascale floating point operations per second). In 2023, the scale of intelligent computing power will reach 70EFLOPS, an increase of 29EFLOPS year-on-year, accounting for more than 30% of the total scale of computing power.

At present, the integration of AI technology into thousands of industries is accelerating, and the demand for computing power for ultra-large-scale AI models and massive data continues to rise. According to the analysis and prediction of the scale of the intelligent computing market, the scale of China's intelligent computing power will continue to grow rapidly, and it is expected that by 2026, the scale of China's intelligent computing power will reach 1271.4EFLOPS, with a compound growth rate of 52.3% in the next five years, and the compound growth rate of the scale of general computing power in the same period will be 18.5%. By 2030, mankind will enter the YB data era, with a 10-fold increase in general computing power and a 500-fold increase in artificial intelligence computing power. Demand is accelerating, and supply is insufficient, and intelligent computing power has become a scarce resource. 

China's intelligent computing power is mainly distributed in national computing hub nodes such as Helinger, Guiyang, and Wuhu, as well as AI super first-tier cities such as Beijing, Shanghai, and Shenzhen. Due to the advantages of strong economic foundation, high level of science and technology, and talent gathering, these regions have become important areas for the development of intelligent computing power. The Beijing-Tianjin-Hebei region, the Yangtze River Delta, the Guangdong-Hong Kong-Macao Greater Bay Area, the Chengdu-Chongqing Economic Circle and other regions are in the leading position in terms of computing power development. Guangdong, Beijing, Shanghai and surrounding provinces generally have a high computing power development index, among which Guangdong, Beijing, Jiangsu, Zhejiang, Shandong, and Shanghai rank high in the computing power development index.

Market demand of intelligent computing power 

The analysis of the market size of intelligent computing power points out that in the evolution of the demand for intelligent computing power, the key driving force comes from the development and application of large models. Pu Wei, CFO of Hongbo Co., Ltd., mentioned that the computing power cluster based on the general large model is leaping from the scale of 10000 calories to the scale of 10,000 cards, especially the 10,000 card clusters being built by domestic giants such as Ali and ByteDance. 

For the pre-training of large models, the requirements for scenarios are increasing day by day, especially for enterprises lacking Vanka computing power, it is difficult to achieve further iteration of the model. 

The rapid development of artificial intelligence, especially deep learning, natural language processing, computer vision, and other technologies, has greatly increased the demand for computing resources. Training and deploying AI models requires powerful computing power, especially for large-scale data processing and high-dimensional model training, and intelligent computing power has become an indispensable infrastructure. Deep learning and large-scale pre-trained models such as GPT-4 and BERT require a large amount of computing resources for training and inference, which drives the demand for specialized hardware such as GPUs and TPUs. 

The popularity of cloud computing has led more and more enterprises to migrate computing resources from traditional data centers to the cloud, and intelligent computing power, as the core component of cloud computing, meets the needs of on-demand allocation and flexible expansion. Cloud computing platforms (such as AWS, Azure, and Google Cloud) provide computing instances based on hardware such as GPUS, TPU, AND FPGAs, which can support users' AI training, data analysis, and other tasks. As the cloud computing market expands, so does the demand for intelligent computing power, especially in AI training and high-performance computing (HPC).

Overall, in the next few years, the demand for intelligent computing power will continue to maintain rapid growth, and will penetrate more into traditional industries, promoting the digital and intelligent transformation of the industry.