The Ripple Effect of China's National Data Infrastructure Project (Part 1)
A deeper dive into Eastern Data, Western Computing
Interested in Party lore? Check out my translation of a 1989 essay by Xi Jinping on the relationship between politics and the arts, which was recently featured in ChinaTalk.
Education, science and technology, and talent are the foundational and strategic pillars for comprehensively building a modern, socialist nation. We must hold science and technology as our top productive force, talent as our top resource, and innovation as our top motivation.
— Xi Jinping, 20th Party Congress (Oct. 16, 2022)
As you may recall, a few months back I published an article discussing China’s Eastern Data, Western Computing project, a large-scale project that aims to build a comprehensive and standardized data infrastructure throughout China.
While the original article provided an overview of the project and its purpose, I’ve decided to launch a two-part series that explores the Eastern Data, Western Computing initiative in greater depth. This is due to two reasons:
The Chinese government’s repeated emphasis that technology plays a critical role in China’s development, underscored by Xi Jinping’s speech at the recent 20th Party Congress.1
A recent 80-page report2 discussing the Eastern Data, Western Computing project, its impact on China’s national infrastructure efforts, the challenges it faces, and what the future may hold.
(Note that I’ll refer to the project as EDWC from now on.)
In this series, I’ll share many of the findings from the above-mentioned report while exploring the EDWC project in greater depth, in addition to providing additional context and data as I see fit.
This week’s article will dive deeper into the project, its significance, and its predicted short-term impact on China’s technological and national development. Next week’s installment will continue to dig into the project’s impact and also examine some of the biggest challenges faced by new capital projects for computing.
EDWC: A quick overview
The EDWC project, which officially launched in February 2022, aims to build a deliberate and comprehensive infrastructure for handling China’s data storage and computation needs. As the name suggests, a major function of this project is to handle the increasing data bottleneck faced in China’s more crowded eastern regions by outsourcing data storage and computation to data centers in the west.
Among other benefits, this project takes advantage of the comparatively abundant access to renewable energy sources in central and western China, thus easing the environmental impact of China’s technological infrastructure.
In more specific terms, the project aims to establish hub nodes (枢纽节点) in eight regions throughout China (including near major eastern regions such as Beijing and Shenzhen), each of which contains at least one data center cluster (数据中心集群).
Why EDWC is so important
Due to the sheer scope of the initiative, the EDWC project is tied to many of China’s priorities for its growth, such as the following.
Data economy
China has named the development of its data economy a national priority. Current initiatives in this regard include standardizing its data and data infrastructure,3 as well as removing data silos and broadening access to data. With its national scope, EDWC is helping to build the backbone of the data economy. If science and technology are China’s key productive force, as Xi Jinping stated this past weekend, computing power is the data economy’s key productive force.
The report underscores this further by stating that for the first time, computing power is being viewed as a fundamental resource like water, electricity, or gas.
EDWC is also tied to the similarly named digital economy, which is also one of China’s national priorities. This project’s impact on the digital economy will be discussed next week.
National data infrastructure
EDWC’s ultimate goal is to create an integrated computing network linked by its eight hub nodes, helping China establish a national computing infrastructure. The project isn’t as simple as throwing data centers at China’s computing needs, but rather deliberately distributing computing power where and when it’s needed.
It’s important to keep in mind that while the project’s name emphasizes the use of western computing resources for data produced in the east, the resulting infrastructure will be far more flexible. As the report notes:
[The project’s] methods are not set in stone. Therefore, "Eastern Data and Western Computing" may not need to be overly emphasized; in certain scenarios, models of “eastern data and eastern computing” or “southern data and northern computing” may also exist. Methods should be implemented in a way that fits their circumstances. However, regardless of the method used, they all share two common objectives:
To maximize the sharing, distribution, and usage of data center resources.
To promote the attainment of China’s carbon-peak and carbon neutrality strategies through the systematic arrangement of data centers.
Environmental goals
As hinted at in the above quote, EDWC is crucially linked to China’s environmental goals. On its current timeline, China intends to reach peak carbon by 2030 and carbon neutrality by 2060. Data centers are energy hogs,4 so prioritizing efficient and carbon-free energy sources for these new data centers is essential so as not to jeopardize these environmental targets.
The hub node regions of Inner Mongolia, Gansu, Ningxia, and Guizhou have plenty of clean energy. Except for of Guizhou, which has ample hydropower, these regions all have plenty of wind and solar energy.
Greater integration
One of the core concepts of the EDWC initiative is its emphasis on integration in a variety of areas.
There are five specific types of integrations that the project focuses on, which are understandably labeled the “five integrations.” These are covered in more depth in an explainer document published in March by the National Development and Reform Commission (NDRC). In short, each type of integration aims to maximize the scalability and interconnectedness of China’s developing data infrastructure.
Below is a short explanation of each type of integration, as translated from the NDRC document:
Network integration
Data centers will be built in clusters with direct network connections. Sensible network settlement mechanisms will be established; network bandwidth and transmission speeds will be increased; and transmission costs will be lowered. New types of internet exchange points and internet backbone access points will be built and introduced to these clusters in a dependable and orderly fashion.
Energy integration
Focusing on China’s “double carbon” policy [carbon peak by 2030, carbon neutrality by 2060], the project will fully explore the abundant wind, solar, and other renewable resources in the western regions. It will properly handle the issue of fluctuations in renewable energy, expand the scope of transactions in the developing clean energy market, and promote the construction of marketization mechanisms for clean energy consumption. In terms of overall planning, the project will apply unified energy consumption quotas to each data center cluster.
Computing integration
For computing power based both in data center clusters as well as inside urban environments, the initiative will promote integrated connection scheduling across data centers in each industry; boost resource linkage between different clouds, between clouds and data centers, and between clouds and networks; and construct resource pools for computing services.
Data integration
The project will construct facilities and platforms for open data circulation, such as for open data sharing or the merging of government and enterprise data. It will experiment with technological models such as secure multi-party computation, blockchain, private computing, and data sandboxes, and it will create environments for the trusted circulation of data.
Application integration
This initiative will launch the construction of integrated city brains.5 It will experiment with launching “data target ranges” for selected urgent and emergency scenarios, such as public sanitation, natural disasters, and urban supervision. It will explore the regulations for data usage and mechanisms for data coordination in different emergencies.
EDWC’s effects on computing infrastructure projects
The report also predicts how the EDWC initiative will affect upcoming capital construction projects for computing and data facilities that are not directly part of the project. Below are some key predictions, as well as existing trends that have already begun to play out.
Changes in locations for new data centers
As a result of the launch of the EDWC initiative, as well as other historical factors in the growth of China’s data infrastructure, new considerations will be made in selecting the locations for future data centers.
Concentrating on areas with computing needs
By the end of 2021, the majority of the total number of racks in China’s data centers were located in the greater Beijing and Shanghai regions. However, this is going to change.
Tier one cities like Beijing, Shanghai, Guangzhou, and Shenzhen have less room for construction and tighter restrictions against data centers. As a result, the cities surrounding these key urban areas have become top choices for building new data centers. As one example, Ali Cloud’s five hyperscale data centers are all located on the peripheries of major cities rather than inside them.
Optimizing for energy efficiency
Recently, regions with greater access to renewables have already pulled in big tech companies:
In south-central Inner Mongolia, the city of Ulanqab has attracted attention from Apple, Tencent, and Huawei, earning the nickname “Silicon Valley of the Prairie.”
In 2021, Gui’an New District had more hyperscale data centers than most regions in the world, accounting for 34% of the country’s data economy
Gansu has built 6 hyperscale data centers for AWS, Meili Cloud Computing, China Mobile, China Unicom, and others, with plans for 4 more.
The construction of EDWC’s eight hub nodes will also help distribute data centers more efficiently. According to the report:
The planning and construction of these 10 data center clusters will optimize the supply structure of data centers and expand computing power while increasing available space. For infrastructure projects that meet certain conditions and have been integrated into the scope of data center clusters for EDWC hub nodes, there are also clear policies requiring them to support energy consumption quotas as needed and achieve a sustainable balance between large-scale computational deployment and the usage of resources like land, energy, water, and electricity.
Moving hot data closer to users
Another trend spurred by the EDWC project involves handling “hot data,” which requires frequent and immediate access, closer to users.
With the widespread use of high-network-demand technologies such as ultra-high definition video, virtual reality, augmented reality, online payments, stock and bond investing, automated driving, industrial manufacturing, and remote medicine, increasing proximity to users is gradually becoming a more prominent trend in the construction of edge computing data centers and small-to-medium data centers, which act as “edge” locations for computing.
Correspondingly, cold data (which requires less frequent and less urgent access) like offline analysis, back-end processes, or storage, which isn’t visited that often, can be processed in the hub nodes located in more western and less populated regions. The same can apply to “warm data,” which falls between these two extremes.
According to Wu Hequan of the China Academy of Engineering, the data produced in China is typically 80% cold data, 15% warm data, and 5% hot data.
There are several examples of recent policies that also support these points:
Overall Arrangement of 5G Base Stations and Data Centers in Guangdong (2021-2025), published by the government of Guangdong, discusses the delegation of different tasks to different data centers based on delay tolerance:
In principle, only mid-sized and smaller data centers should handle edge computing and low-delay tasks.
Infrastructure for mid-delay tasks would be gradually migrated to the eastern, western, and northern regions of the province.
Infrastructure for high-delay tasks should continue to be moved outside the province.
The Beijing city government’s Beijing Data Center Planning and Development Implementation Plan (2021-2023) lays out the following points:
Renovating unused resources and spaces to build edge computing centers that can support low-delay tasks and service key applications such as city brains and autonomous vehicle networks.
With the exception of edge computing centers, no data centers may be built or expanded in Beijing’s Dongcheng and Xicheng districts (essentially the city core).
In the future, data can be handled in the EDWC network according to its computational needs — particularly tolerance for delay.
Therefore, data centers for tasks that require little-to-no delay (such as self-driving car networks or industrial robots) can be built in node cities in the future. Data centers for tasks without strict delay requirements, such as servicing back-end processes, offline analysis, and cold data storage, can prioritize western node locations such as Guizhou, Inner Mongolia, Gansu, and Ningxia.
Additional opportunities
The EDWC project is shifting the paradigm for China’s data infrastructure, and it’s opening up paths for new patterns and projects. Some of these opportunities include cloud-based rendering for games and video production, as well as AI model training and image processing.
As the project progresses and China’s data infrastructure grows more mature, EDWC can be applied to an even broader range of scenarios.
Certain scientific applications, like modeling for lattice quantum chromodynamics (QCD) and molecular dynamics, and calculations for material and biological data, have begun exploring EDWC-based solutions.
Coming up next
The second installment in this two-part series will explore the following:
The impact of the EDWC project from a broader economic perspective
Challenges that the initiative will face over time
Further projections for the future of China’s digital landscape
English and Chinese versions of the full Party Congress report that Xi Jinping drew from while giving his speech are available here.
The full title of the report is Research Report on the Development of New Types of Computing Infrastructure in the Context of the National Eastern Data, Western Computing Project. It was published by the China Smart Computing Industry Alliance (中国智能计算产业联盟) with input from Tsinghua University, the State Information Center of China (a government think tank associated with the NDRC), and other organizations.
China’s construction of a “national cloud” (which I previously wrote about) is one example of this, as is the recent announcement by the Civil Aviation Administration of China (CAAC) seeking to increase the level of integration and organization in civil aviation data and its infrastructure.
For more specific data on data center energy usage, you can read this 2016 report sponsored by the U.S. Department of Energy.
For more context on China’s city brain initiatives, refer to this Cloudology article from August.