news

on september 12, at the 2024 china international fair for trade in services, the general artificial intelligence computing power forum was held in shougang park. with the theme of "building a foundation for unbounded intelligent computing to unleash new productivity", this forum attracted a group of industry experts and scholars to jointly discuss new possibilities for the development of artificial intelligence computing power.

nowadays, all large, medium and small enterprises in the domestic market are undergoing digital transformation. behind the demand for "everything on the cloud" lies huge market potential. how to expand the boundaries of the computing power ecosystem and how to foster more efficient models are both industry problems and opportunities.

multi-dimensional and multi-level

at the forum, zheng weimin, academician of the chinese academy of engineering and professor of the department of computer science and technology at tsinghua university, said that it is crucial to develop artificial intelligence and build domestic computing power. the key factor lies in developing a software ecosystem and fully unleashing the potential of the underlying software.

zhang hongke, academician of the chinese academy of engineering and professor at china university of transportation, said that computing power network is a new type of information infrastructure, showing characteristics such as diversity and ubiquity, intelligence and agility, security and reliability, and green and low-carbon. creating a new internet system architecture is an urgent need to seize the main battlefield of the future economy.

beijing business daily reporters learned that my country's computing industry is currently developing rapidly, but it also faces a series of challenges, including mismatch between supply and demand, and problems with energy consumption, security, and coordination between software and hardware. these difficulties are necessary for initial exploration and will be gradually resolved in future practice.

according to experts attending the meeting, my country's ai computing power is still in the development and exploration stage. a number of companies have flocked into the market to provide their own solutions for ai computing power, which are generally small in scale but rich in levels.

according to the data, as of the end of 2023, there are 128 projects with "intelligent computing centers" nationwide, of which 83 projects have disclosed scale, with a scale of more than 77,000p; the scale of different intelligent computing centers varies greatly, and the computing power scale is generally 50p, 100p, 500p, and 1000p; there are also scales as high as 12,000p or more.

in terms of project scale, 63% of the projects in the intelligent computing center have a computing power scale of less than or equal to 500p, and 17.7% of the projects have a computing power scale of 500-1000p, showing a small-scale and multi-level trend.

this "let a hundred flowers bloom" trend also indirectly demonstrates the vitality of the market. it is not difficult to see from the market that today's small and micro enterprises are also undergoing digital transformation and achieving "everything on the cloud". it is through digital transformation and the development of cloud computing that the potential of the cloud service market has been driven, and market opportunities have been transferred, which will be shared by data service companies of different levels and sizes.

rich applications and business opportunities

at the forum, a number of companies also demonstrated cutting-edge computing capabilities and put forward their own views and solutions to current industry trends.

for example, qishang online and beilong yunhai have brought together the technical strength of the chinese academy of sciences and the computing industry to jointly launch the intelligent computing center operation and maintenance service products. the service system covers a full range of operation and maintenance services from infrastructure, network to it equipment. including the maintenance and performance optimization of key computing resources such as gpus, roce and ib professional network formation, and also provides value-added services such as network optimization and model tuning.

wang yi, chairman and ceo of qishang online, said that currently, the computing industry is facing challenges such as limited gpu supply, insufficient closed-loop development ecosystem, and difficulty in model implementation. as a service provider of data-computing integration, the company actively responds to challenges, focuses on computing network and scheduling, explores innovative business models, and builds an open and win-win computing integration ecosystem.

at the forum, li fei, vice president of h3c group and dean of the artificial intelligence research institute, interpreted the mutual response between ai computing power and new quality productivity. he said that new quality productivity is a fundamental subversion of the original technology route. h3c takes "computing power × connection" as its core strategy, based on large-scale model customization and solution capabilities of industry data and application scenarios, to fully assist users in the end-to-end implementation of aigc.

in addition to serving enterprises and individual customers, how to make ai computing power play a greater role in municipal management is also the focus of this forum. yang xuqing, chief technology officer of softcom wisdom, discussed in depth the dual-engine driving model of "data + ai" and elaborated on the ideas of building an ai computing power flexible service system and an ai-oriented data governance framework. he emphasized that the application of large models will effectively promote the intelligent transformation of urban services and inject new vitality into urban development.

from the industry's point of view, through cooperation among enterprises, institutions, colleges and universities, and governments, my country will next build a computing power internet that is more connected, flexible and efficient, and promote the deep integration of computing power and the real economy.

beijing business daily reporter tao feng and wang zhuli