AI Large Models Thrive as Listed Companies Race to Accelerate Computing Power Development
-
Since the beginning of this year, AI technologies represented by ChatGPT have sparked a global wave of innovation, with over 80 large AI models publicly released in China so far.
The flourishing development of large AI models has directly driven the demand for intelligent computing power, unleashing enormous expansion opportunities in the computing power industry. Reporters have noted that recently, many listed companies have announced substantial investments in the computing power sector.
Jiang Qin, Director of the Industrial Empowerment Center at SenseTime's AI Infrastructure, told the Shanghai Securities News that large models have sounded the clarion call for the rapid development of AI, leading to an explosive demand for intelligent computing power, which is increasingly becoming the indispensable 'lifeblood' of AI operations.
"The explosive growth of large models, combined with policy initiatives like the 'East Data West Computing' project, will drive China into an era of large-scale expansion in intelligent computing infrastructure under the dual forces of strong demand and policy support. Leading companies with existing technological and resource advantages will benefit significantly," said an AI industry researcher from Hetai Asset Management.
Multiple Companies Expand Computing Power Investments
On the evening of July 27, Zhenshi Tong announced that it had signed a strategic cooperation framework agreement with Shandong Shuifa Ziguang Big Data Co., Ltd., aiming to collaborate on the construction, operation, AI computing services, sales, and promotion of next-generation intelligent computing centers.
Just two days earlier, Zhenshi Tong had signed a strategic cooperation agreement with Tom Cat to collaborate on AI computing services and artificial intelligence, building an industrial ecosystem of computing power, models, and applications, and promoting the adoption of large AI models across various fields.
Xinjiang Daqo New Energy has also been actively investing in AI computing power. On June 9, the company announced a strategic cooperation agreement with Zhejiang Jiuzhou Cloud Information Technology Co., Ltd. to collaborate on the construction and operation of energy computing centers. The latter is a company specializing in edge computing and private cloud services.
In May, Xinjiang Daqo signed a strategic cooperation agreement with the Xiangcheng District Government of Suzhou to establish the global headquarters of its energy computing center in the Suzhou Computing and Security Industrial Park. The company plans to build 15 energy computing centers worldwide by the end of 2024, with investments exceeding 5 billion yuan.
Xinjiang Daqo began its foray into the computing power industry as early as April. On April 2, the company stated that computing power is now and will continue to be a key driver of productivity. Integrating computing power with digital energy will lead to exponential growth in industrial value chains, helping companies seize opportunities in the AI era. To embrace the digital economy, the company's subsidiary invested 300 million yuan in Xinglin Technology, acquiring a 10% stake.
Additionally, Hongxin Electronics plans to invest 1 billion yuan in building an AI computing server manufacturing base; Weixing Intelligence intends to jointly construct a computing center with an investment of 3 billion yuan; Litong Electronics plans to invest 450 million yuan in AI computing leasing services; and Chaoxun Communications aims to raise 1.079 billion yuan for the Ninghuai Green Digital Economy Computing Center project.
Market Demand and Policy Support Drive Growth
With breakthroughs in large models and the expansion of application scenarios, the demand for intelligent computing power is growing rapidly.
Jiang Qin noted that since 2012, the computational requirements of the largest AI models have doubled every 3.5 months. Compared to Moore's Law, which predicts a doubling period of 18 months to 2 years, the growth rate of computing power demand far exceeds the capacity of chips to keep up.
Data from the Ministry of Industry and Information Technology shows that China's computing power industry has grown by nearly 30% annually in recent years, with the total computing power scale ranking second globally. By the end of 2022, China's total computing power reached 180 EFLOPS (18 quintillion floating-point operations per second), with the core industry valued at 1.8 trillion yuan.
According to the China Academy of Information and Communications Technology, global computing power will enter a phase of rapid growth from 2025 to 2030, with a compound annual growth rate exceeding 70%.
From a policy perspective, China places high importance on the development of the AI industry, laying a solid foundation for the growth of intelligent computing power.
In February last year, four ministries jointly issued a notice approving the launch of national computing power hub nodes in eight regions and planning 10 national data center clusters. This marked the completion of the overall layout design for China's national integrated data center system. With the full implementation of the 'East Data West Computing' project, intelligent computing center construction has entered a new phase of accelerated development.
Zhao Zhiguo, spokesperson and chief engineer of the Ministry of Industry and Information Technology (MIIT), recently stated that the development of general artificial intelligence has raised higher requirements for computing power. The MIIT will focus on three key areas: continuously promoting computing infrastructure construction, advancing breakthroughs in core technologies and industrial upgrading, and stimulating the enabling value of computing applications.
Additionally, reporters noted that since the beginning of this year, AI concept stocks have been wildly speculated, with the computing power sector being the hottest area, attracting many listed companies to cross over into AI computing. For example, companies like Hongxin Electronic, GCL Energy, and Litong Electronics originally operated in industries outside the computing power industrial chain.
An interviewed expert warned that with computing power construction booming nationwide, industry practitioners, investors, and local governments need to think calmly and guard against the overheating risks of computing infrastructure investment. They cautioned against rushing headlong into low-end redundant construction that could lead to capital and resource waste.