Rational Analysis of "East Data, West Computing": Where the "Electricity" Is, the "Computing" Is.

　 “East Data, West Computing” is perhaps the hottest concept lately.

　　According to documents from the National Development and Reform Commission and other ministries and commissions, it has been agreed to launch the construction of national computing power hubs in the Beijing-Tianjin-Hebei region, the Yangtze River Delta region, the Guangdong-Hong Kong-Macao Greater Bay Area, the Chengdu-Chongqing region, Inner Mongolia, Guizhou, Gansu, and Ningxia. Ten national data center clusters have also been planned, and the super project “East Data, West Computing” has been fully launched. The capital market quickly became excited, related concept stocks rose sharply, and some even compiled a detailed list, enumerating the leading companies and high-quality targets in the sub-markets of big data, cloud computing, computing power construction, and data security, creating multiple daily limit increases in a short period of time. However, if one only focuses on the concept of “East Data, West Computing” and ignores the layout logic, time schedule, and current industrial situation of this systematic project, even a slight disturbance or negative news may cause the stock prices of those concept stocks to turn from rising to falling, forming a temporary capital bubble.

　　Therefore, the purpose of this article is not to tell everyone about investment opportunities, but to rationally cool down the hype surrounding “East Data, West Computing”.

　　 Traditional infrastructure drives new infrastructure

　　Why launch the “East Data, West Computing” project?

　　The simple explanation is to transfer data from the east to the west for computation and processing. Similar to super projects like South-to-North Water Diversion, West-to-East Power Transmission, and West-to-East Gas Transmission, the root cause of “East Data, West Computing” lies in the unbalanced domestic resource supply, forcing a geographical solution to resource allocation problems. To put it more simply, “computation” is a highly energy-consuming task. In the operating cost of a data center, electricity costs often account for more than 50%. The uneven distribution of electricity resources in China is arguably the direct inducement for the “East Data, West Computing” project. How to solve the problems of location, energy consumption, and distribution of “computing power” is obviously a subject worthy of in-depth study. Let's first look at a set of comparative data on the net power supply of different provinces and cities. Inner Mongolia, Yunnan, Guizhou, and Gansu are provinces and cities where power generation is greater than power consumption, while Beijing, Zhejiang, and Guangdong are provinces and cities with a serious shortage of power generation. To solve the predicament of power shortages in the east, the country launched the aforementioned “West-to-East Power Transmission” project, and ultra-high voltage has also been included in the key projects of new infrastructure, but these measures are still insufficient to solve the fundamental problem.

　　Taking the southern route of “West-to-East Power Transmission” as an example, it mainly transmits hydropower from Yunnan and Guizhou to the Guangdong-Hong Kong-Macao Greater Bay Area. The amount of electricity that can be transmitted in a year is about 300 billion kilowatt-hours, but the annual electricity consumption in the five southern provinces is over 1300 billion kilowatt-hours, far from enough to meet the electricity demand. Recalling the “power shortage” phenomenon that occurred in many eastern provinces and cities in 2021, placing high-energy-consuming data centers in eastern regions is clearly not an ideal choice.

　　Another calculation is the cost. With the current technology level, the transmission loss of long-distance power transmission over 2000 kilometers is about 6%, which is already quite optimistic, but the construction cost of ultra-high voltage transmission lines cannot be ignored. Referring to the publicly available data from the State Grid and the Southern Power Grid, the construction cost of a single ultra-high voltage DC line is about 25 billion yuan; while a 12,000-kilometer submarine optical cable only requires 500 million US dollars. The cost-effectiveness of the two is self-evident. Moreover, data centers also need to consume a large amount of electricity to cool down servers. Some institutions estimate that even if the average industrial electricity price is calculated at 0.5 yuan per kilowatt-hour, a 1°C reduction in the temperature of the data center location can save a standard data center machine room with 100,000 servers 96,000 yuan per day.

　　Microsoft, Amazon, and Facebook have all gone to extremes for “physical cooling,” such as Microsoft submerging the “Northern Islands” data center in the ocean, and Facebook building a data center in northern Sweden near the Arctic Circle. Alibaba Cloud, Tencent Cloud, Baidu, and Huawei in China are also moving their data centers to areas with lower overall temperatures, such as Guizhou and Gansu. Even without the “East Data, West Computing” project, the trend of data center “westward movement” has already taken shape.

　　Against this backdrop, “East Data, West Computing” is a necessary choice for national strategy and an inevitable path for promoting coordinated development between the east and west. However, whether it is the right time to deliberately hype up “East Data, West Computing” concept stocks still requires rational consideration based on the actual situation.

　　Demand determines the speed of launch

　　“East Data, West Computing” is essentially a long-term plan.

　　According to the CDCC’s “2021 China Data Center Market Report,” the total number of data center racks in China in 2021 is expected to reach approximately 4.1506 million, with a total computing power of 120 EFLOPS, and will continue to grow at a rate of over 30%. From the numbers alone, the growth of computing power is undeniably optimistic, but two sets of sub-data should also be noted.

　　First, the current distribution of computing power in different provinces and cities in China. According to the data provided by the China Academy of Information and Communications Technology in the “China Computing Power Development Index White Paper,” Beijing, Guangdong, and Shanghai rank among the top three in terms of computing power, and the growth rate of computing power in eastern provinces such as Zhejiang, Guangdong, and Jiangsu exceeds 60%, while Inner Mongolia, Guizhou, and Gansu are not superior in terms of computing power scale and growth rate.

　　Second, the uneven rack utilization rate of current data centers. The current average utilization rate of data centers in China is about 50.07%, and the direct factor leading to the low utilization rate is the uneven demand. Internet companies, which are major consumers of computing power, are mainly concentrated in Beijing, Shanghai, Guangzhou, and Shenzhen, so the demand for computing power in eastern provinces and cities is generally higher, while the industrial structure in western regions has a weaker demand for computing power.

　　The strategic consideration of the “East Data, West Computing” project also lies in this, that is, to promote the orderly transfer of computing resources westward and help solve the imbalance between the supply and demand of computing power in the east and west. However, in the traditional cooperation model, most cloud vendors use eastern data centers as source stations and central and western regions as CDN nodes and edge computing nodes. Whether transferring data to the west for computation or storage, the prerequisite is to build high-speed network channels between data centers.

　　Therefore, the “East Data, West Computing” project will not be like past infrastructure construction with a “flood irrigation” approach, but a gradual and systematic project. Especially in the initial stage, in addition to the macro layout of eight computing power hubs and ten data center clusters, two indicators are deliberately emphasized:

　　First, the rack utilization rate. The average rack utilization rate of data centers within the cluster must reach at least 65%. Compared with blindly building data centers, the early focus may be on improving the utilization rate of current data centers, especially the utilization rate of several data centers in the west.

　　Secondly, there is green computing. A key indicator for measuring the energy efficiency of data centers is PUE (Power Usage Effectiveness), calculated by dividing the total power consumption of the data center by the power consumption of IT equipment. A PUE value closer to 1 indicates greater energy efficiency. The "East Data, West Computing" project explicitly limits the PUE of Zhangjiakou, Shaoguan, the Yangtze River Delta, Wuhu, Tianfu, and Chongqing clusters to below 1.25, and the PUE of the Linhe, Gui'an, Zhongwei, and Qingyang clusters to below 1.2. Given the shortage of professional and technical personnel in western China and the bottleneck in heat dissipation technology for reducing PUE values, "East Data, West Computing" will be a process of gradually releasing benefits. The Department of High Technology of the National Development and Reform Commission also issued a document in June 2021 stating that a relatively ideal goal is for the proportion of data centers in eastern China to decrease from 60% to around 50%, and the proportion of data centers in western China to increase from 10% to around 25% by the end of the 14th Five-Year Plan period.

　　Therefore, at this stage, promoting or hyping "East Data, West Computing" concept stocks does not have much practical value, and will not change the existing pattern of computing power in the east and west in the short term. If we reorganize the timeline, the "East Data, West Computing" plan has been in the works for a long time. In September 2020, the "East Data, West Computing" Industry Alliance was established. In May 2021, the layout of the computing power network hub nodes was clarified. In December of the same year, the carbon neutrality and green computing plan was finalized, and then the "East Data, West Computing" project was officially launched... This is a deep-level plan to solve the domestic energy and computing power bottlenecks, and should not become a temporary hot topic.

　　 Microclimate in the Macro Environment

　　The connotation of "East Data, West Computing" is not simply a literal understanding.

　　Referring to the statement of Academician Wu Hequan of the Chinese Academy of Engineering: "Due to the delay caused by long-distance network transmission and the influence of related supporting facilities, some businesses with low network requirements, such as background processing, offline analysis, and storage backup, can be transferred to the west first. Some businesses with higher network requirements, such as industrial internet, disaster warning, telemedicine, and artificial intelligence reasoning, can be deployed in eastern hubs such as Beijing-Tianjin-Hebei, the Yangtze River Delta, and the Guangdong-Hong Kong-Macao Greater Bay Area. Within the hubs, it is important to promote the transfer of data centers from first-tier cities to surrounding areas." This actually distinguishes two application scenarios:

　　One is for businesses with low network latency requirements, typical examples being log analysis, daily and monthly report analysis, personalized user algorithm analysis, and video rendering in software systems. Although it increases the cost of data transmission, the advantages of cheaper electricity and land prices in western China are sufficient to offset the increased hard costs of data transmission, attracting more and more enterprises and institutions to embrace data centers in western China.

　　The other is for businesses with high network latency requirements, such as innovative applications represented by industrial internet, autonomous driving, finance and securities, and smart cities. The value of these applications can no longer be considered only from the perspective of cost-effectiveness. They are the main direction of industrial digital transformation, a foreseeable economic growth driver, and have an underestimated driving effect. For a long time, these high-value applications will remain concentrated in the east.

　　According to publicly available statistics, in 2020, the scale of industrial digitalization in China was 31.7 trillion yuan, accounting for 80.9% of the digital economy, with innovative applications contributing a significant proportion. This also means that low latency remains an important indicator of computing resources, and the role of eastern data centers cannot be replaced in the short term. This is probably the reason why the "East Data, West Computing" project has established eight computing power hub nodes, and has not simply deployed computing resources in western provinces. The Beijing-Tianjin-Hebei, Yangtze River Delta, and Guangdong-Hong Kong-Macao hubs cover four national data center clusters, and have clearly stated that they will accelerate the improvement of the data center industry ecosystem.

　　The challenge is actually left to cloud computing vendors, who need to change traditional node deployment and build a hybrid architecture that adapts to the "East Data, West Computing" pattern. For example, users can upload raw data to data centers in the west for AI model training and data storage with high computing density, then synchronize the trained models to data centers in the east for inference services, and then integrate edge computing to achieve ultra-low latency calculations.

　　What also needs long-term consideration are those western provinces and cities that have tasted the "benefits." If electricity is the engine of economic growth in the industrial age, computing power is the new productivity of the digital economy. The opportunity for the "source" of computing power is not simply to export production resources like electricity, but to change the resource-exporting economic structure, promote the transformation of industrial digitalization and digital industrialization, and give rise to new technologies, new industries, new formats, and new models.

　　In other words, the significance of the "East Data, West Computing" project is not simply to balance resources between the east and west. Its hidden mission is to accelerate the orderly transfer of related industries to the west under the macro environment of the eastward layout of computing infrastructure, attract more and more excellent talents to develop in the west, continuously activate the economic vitality of the west, form coordinated development of supply and demand, and use the tailwind of digital transformation to bridge the economic gap between the east and west.

　　Postscript

　　To reiterate this point: The time scale of the "East Data, West Computing" project may be a decade, and many practical problems need to be solved during this period. Simply promoting it in the capital market or using it to hype up stock prices and market capitalization may be suspected of "harvesting anxiety".

　　Article Source: Alter Tech Talk

　　Disclaimer: Some of the publicly available information collected on this website comes from the internet. The purpose of reprinting is to convey more information and for online sharing, and does not represent this website's endorsement of its views or responsibility for its authenticity, nor does it constitute any other suggestions. The content of the article is for reference only. If you find any works on the website that infringe your intellectual property rights, please contact us, and we will modify or delete them promptly.