share_log

东方证券:AI迅速发展持续拉动算力需求 算力调度运营进程加速

Oriental Securities: The rapid development of AI continues to drive computing power demand, computing power scheduling, and the operation process accelerates

Zhitong Finance ·  Nov 17, 2023 16:56

The rapid development of AI continues to drive demand for computing power. Computing power scheduling is expected to solve problems such as the imbalance between computing power supply and demand, and empowering computing power as a basic resource for the benefit of the public.

The Zhitong Finance app learned that Orient Securities released a research report saying that the rapid development of AI continues to drive demand for computing power. Computing power scheduling is expected to solve problems such as the imbalance between computing power supply and demand, and empowering computing power as a basic resource for the benefit of the public. Currently, computing power sensing and scheduling are not yet mature, and the central government and local authorities continue to publish documents to jointly promote the development process of computing power scheduling. In addition, governments in many regions have issued documents to promote integrated scheduling of computing power, and have built computing power scheduling platforms with the support of technology companies, starting with regional scheduling. Recommended attention: Zhongke Shuguang (603019.SH), Qingyun Technology-U (688316.SH), Capital Online (300846.SZ), Uker-W (), Shenzhen Sunda A (000032.SZ), Stacky (300608.SZ). 688158.SH

Orient Securities's views are as follows:

The rapid development of AI continues to drive demand for computing power. Computing power scheduling is expected to solve problems such as the imbalance between computing power supply and demand, and empowering computing power as a basic resource for the benefit of the public.

This year, we have entered the era of big models, and the demand for intelligent computing power has greatly increased. Continued breakthroughs and applications of AI algorithms such as ChatGPT will greatly drive computing power demand. However, our country's computing power has problems such as an imbalance between supply and demand, high costs, and insufficient robustness. How to improve the utilization rate of computing power and reduce the difficulty of using computing power, so that small and medium-sized enterprises and individuals can use low-threshold, inclusive computing power becomes the key to the country's technological and economic development. Achieving flexible scheduling across regions and on demand is expected to continue to empower computing power inclusion, making computing power a basic resource that can be used as needed, such as water and electricity.

Currently, computing power sensing and scheduling are not yet mature, and the central government and local authorities continue to publish documents to jointly promote the development process of computing power scheduling.

Currently, the perceptual scheduling of computing power is still difficult. How to allocate appropriate computing resources to the demand side is the key. The diversity of computing power also challenges the interconnection of nodes and the unified scheduling of computing power. Countries and localities have introduced many policies to continue to advance the development process of computing power scheduling. In 2023, policies such as the National Supercomputing Internet and the “Action Plan for the High-Quality Development of Computing Infrastructure” were introduced one after another, all of which proposed exploring mechanisms for collaborative scheduling of computing power. At the same time, local authorities have also issued a number of documents promoting the integrated scheduling process of computing power.

The scheduling and operation of computing power is similar to the role of a platform. It can match appropriate computing power to the corresponding demand side, improve the utilization rate of computing power resources, and make it a basic resource for the benefit of the public.

Computing power scheduling can deploy computing power tasks and deliver results across service providers, architectures, and regions. It can sense the computing power needs of customers in real time and automatically allocate optimal computing power that meets their needs according to the situation in the computing power resource pool. Obtaining computing power will be as convenient and inexpensive as obtaining water and electricity. Computing power scheduling has far-reaching significance in improving the utilization rate of computer resources, reducing resource waste, and ensuring the efficient execution of tasks.

Many local governments have taken the lead in building computing power scheduling platforms, and many enterprises and scientific research institutions have entered the market one after another.

Many local governments have issued documents to promote integrated scheduling of computing power, and have built computing power scheduling platforms with the support of technology companies, starting with regional scheduling. At the same time, many enterprises, institutions, and research institutes are also actively exploring computing power scheduling. The Institute of Computing of the Chinese Academy of Sciences, the Chinese Science Department represented by Shuguang of the Chinese Academy of Sciences, the Pengcheng Laboratory of Huawei, the Huawei Pengcheng Department represented by Huawei, communication operators, parallel technology, Qingyun Technology, and Capital Online tried first to explore computing power scheduling technology and operation models. It is expected that they will gain a first-mover advantage in the wave of computing power and continue to benefit.

Risk warning: Policy implementation falls short of anticipated risks; technology development related to computing power scheduling falls short of anticipated risks.

The translation is provided by third-party software.


The above content is for informational or educational purposes only and does not constitute any investment advice related to Futu. Although we strive to ensure the truthfulness, accuracy, and originality of all such content, we cannot guarantee it.
    Write a comment