The wave of intelligent computing driven by generative AI has not only brought business changes to thousands of industries, but also triggered changes in the future of data centers. Recently, at the 11th Data Center Standards Conference hosted by CDCC (China Data Center Working Group), Schneider Electric launched a special forum with the theme of "From the present to the future, insisting on innovation" to discuss and share the reshaping of the future in the AI era. innovation paths and practices.
In the era of AI intelligent computing, changes and changes in data centers
The vigorous development of the digital economy has played an important role in promoting the growth of my country's data center industry. Data from the China Academy of Information and Communications Technology shows that in 2022, the scale of my country's digital economy will exceed 50 trillion yuan for the first time, and the digital economy's proportion of GDP will also increase to 41.5%. The acceleration of the digitalization process is driving the rapid growth of data centers and digital infrastructure construction. In particular, national-level actions to improve computing power supply and technology companies' demand for intelligent computing have further boosted the growth of new data centers such as supercomputing centers and intelligent computing centers. It is expected that by 2025, my country's computing power will exceed 300 EFLOPS, of which intelligent computing will The computing power accounts for 35%, and the market growth potential is huge.
At the level of industrial policy, the trend of integrating energy and computing power continues, making the impact of energy on data centers increasingly ubiquitous. According to the plan of East Data and West Calculation, by 2025, the average power utilization efficiency of newly built large and ultra-large data centers across the country will be reduced to less than 1.3, and the national hub nodes will be further reduced to less than 1.25. At the same time, the level of green energy utilization in data centers will continue to increase, and the proportion of renewable energy sources such as solar energy and wind energy in the energy supply of data centers will rapidly increase.
In terms of market trends, the emergence of generative AI this year has led to a rapid increase in demand for intelligent computing. Over the past 20 years, data center demand has grown primarily from storage and general-purpose computing, as well as the migration from on-premises deployments to the cloud. In the future, AI will become the biggest driving force for the development of data centers, and the training and inference of large models will put forward higher computing power requirements for intelligent computing. According to IDC forecast data, by 2030, the average annual growth rate of my country's intelligent computing power will reach 62%, which together with supercomputing will bring about huge changes in the future growth structure of computing power. In the next five years, AI will sequentially drive the growth of data centers and edge computing, and the growth rate will be 2-3 times that of traditional data centers.
In this regard, Yan Yong, vice president of low voltage system business of Schneider Electric Energy Management Low Voltage Division, said that in this gradual process of coexistence, we must not only continue to adhere to the path of efficient and sustainable development, but also cope with the changes and challenges it brings. , further improve the adaptability of data center design and re-examine the resilience of data centers. "With the changes in the computing power growth structure, future-oriented data centers should be more 'resilient, efficient, sustainable, and adaptable.' As a global leader in data center infrastructure and digital services, Schneider Electric has the expertise in medium and low voltage power distribution to the IT computer room, from interconnected hardware to edge- and cloud-based software integrated solutions, from consulting design to joint development of customized innovative solutions, as well as smart and efficient full life cycle digital solutions, thereby providing data center innovation. Develop comprehensive empowerment to help data center users move better into the future."
Four dimensions to help build a future-oriented data center
In the face of changes in data centers, Schneider Electric used multiple keynote speeches to discuss how to keep pace with the future and enable data centers to keep pace with the era of intelligent computing.
First, improve data center resilience. In intelligent computing centers, the continuous computing power demand brought by large model training makes the peak-to-average ratio of data center load approach 1. In this context, in order to reduce the impact of unexpected downtime, the protection selectivity of data center power distribution becomes particularly important. Schneider Electric believes that the protection selectivity of data center power distribution needs to be examined from the perspective of the entire life cycle of the design stage, construction stage and operation and maintenance stage.
In addition, power supply interruptions caused by equipment failure are still the main cause of data center outages. Ensuring the toughness of individual equipment and the sensitivity of monitoring equipment are still the keys to ensuring the resilience of the data center. For the 2N architecture currently commonly used in data centers, there is still room for further improvement in the stability and reliability of products and technologies. Creating strong intelligent power distribution and assisting intelligent operation and maintenance is an effective way to solve this problem. Data center planners can use ETAP digital software to assist in the design of power distribution systems, reducing the computational workload and workload caused by the numerous levels and complex operation methods of the power distribution system.
Secondly, we need to improve the efficiency of the data center. The high-density computing power brought by generative AI requires a computer room with higher power density to ensure it. Data center efficiency includes effectively optimizing costs, deployment speed and space usage, helping enterprises optimize capital structure and cash flow.
Taking the power module as an example, Schneider Electric has built a power module based on digital twins to provide closed-loop capabilities spanning design-construction-operation and maintenance, bringing customers three major values: safety, reliability, convenient operation and maintenance, and open compatibility. First, the ETAP digital software can be used to achieve a verifiable design; then, by using the high reliability of the BlokSeT standard solution and incorporating it into UPS products, through temperature rise simulation verification and false load testing, verifiable integration can be achieved; based on the medium and low voltage integrated solution , realize linkage management and improve power supply continuity guarantee; and use highly reliable equipment to lay a solid foundation for the safety and reliability of power modules. Schneider Electric's power module solutions based on closed-loop capabilities can minimize the impact of faults, improve efficiency and optimize costs from end to end.
Third, the adaptability of the data center must be improved. Under the new power system architecture, enterprises have put forward higher demands for data center power quality management, peak load increase, carbon reduction and cost saving. In this context, it is just the right time to build a capacity-increasing data center that can handle peak loads.
Utilizing innovative technologies such as UPS energy storage and joint power supply, Schneider Electric builds a capacity-increasing data center for enterprises by creating a UPS peak-bearing solution. Compared with the traditional solution, the capacity expansion data center with the help of UPS peak-carrying solution and chilled water storage tank system can increase the IT load of the data center by 35%. Although the initial total investment increases by 14%, the investment cost per watt can be reduced by 16%. , the ROI time is only 1.5 years.
Finally, we must adhere to sustainable development. As large energy consumers, data centers have always been one of the most important players in achieving sustainable development. With the continuous decline of PUE, if data centers want to achieve truly sustainable development, they need to expand from the consideration of energy consumption to the use of renewable energy, water use, greenhouse gas emissions, waste generation and processing, and local environmental protection. The impact on the ecological environment and other broader dimensions.
As a practitioner and enabler of sustainable development, Schneider Electric not only took the lead in proposing the elements and measurement indicators of sustainable development of data centers by publishing a white paper, but also helped operators at different stages of development assess the sustainability of the data center environment from multiple dimensions. Continuously conducts overall analysis and optimization, and can also provide sustainable development services covering carbon neutral consulting, renewable energy procurement, resource and energy efficiency services, waste management, carbon-efficient supply chain consulting, battery recycling, etc.
At this Data Center Standards Summit, Schneider Electric grandly released more than ten innovative products and solutions for data center applications, covering power supply and distribution systems, power distribution equipment, digital management software, overall energy-saving transformation solutions and other fields. . Facing the advent of intelligent computing, Schneider Electric has launched continuous innovation in products, technologies and services to help customers build future-oriented data centers to quickly respond to market changes. In the new era of change, join hands with industry chain partners to promote the transformation and development of China's data center industry and win-win the future of intelligent computing!
