The Ripple Effect of China's National Data Infrastructure Project (Part 2)
Concluding our exploration of Eastern Data, Western Computing
This is the final part of a two-article deep dive into the Eastern Data, Western Computing project. You can find the first article here. For further background on the project, you can also read my earlier standalone article introducing it.
Last week, we began examining the Eastern Data, Western Computing project (EDWC), China’s massive initiative to build a national data and cloud computing infrastructure to make its future data-driven development more efficient and environmentally sustainable.
In this article, we’ll conclude our in-depth look at the project by exploring EDWC’s predicted effects on China’s economy, as well as several key obstacles the project needs to tackle over time.
Many of the insights in this week’s article are taken from the same hefty EDWC report1 discussed last week. As mentioned previously, this report was published by the China Smart Computing Industry Alliance (中国智能计算产业联盟) with input from several institutions and organizations, including Tsinghua University and the State Information Center of China, a Chinese government think tank.
Note: For the sake of brevity, the above report will simply be referred to as “the report” throughout this article.
EDWC and China’s economy
The digital economy
The national data and computing infrastructure being built by EDWC is not only crucial to China’s data economy (discussed in more detail last week). It also boosts the country’s growing digital economy.
The speed of the digital economy’s development, the breadth of its scope, and the depth of its influence are unprecedented. The digital economy is promoting profound transformations in methods of production, life, and governance. It is becoming a critical force in reorganizing essential global resources, reshaping the structure of the global economy, and transforming the face of global competition.
…In 2020, the proportion of value added from the core industries of China’s digital economy to its GDP was 7.8%… By 2025, the digital economy will stride toward a period of comprehensive expansion. The core industries of the digital economy will add a 10% proportion of value to the GDP… Step by step, China’s digital economy will grow both in competitiveness and influence.
— 14th Economic Five-Year Plan, State Council (2021)
The EDWC report highlights the significance of computing power for the digital economy, referring to it as an “important pillar” of the latter.
According to a 2018 report by the Institute of Computing Technology at the Chinese Academy of Sciences,2 computing power will become a representative index for measuring the development of a region's digital economy, as well as a primary means for transforming and updating its momentum. In addition, every 1 RMB invested in computing power drives the production of 3 to 4 RMB in China's economy. Similarly, every single-point increase in China's computing power development index accompanies a growth of about 129.3 billion RMB in its GDP.
The industry ecosystem
EDWC is projected to result in development along the industry value chain. It will spur investment and lead to the construction of large-scale and hyperscale data centers in central and western China, as well as around tier-one cities such as Beijing, Shanghai, Guangzhou, and Shenzhen.
Since the beginning of 2022, 25 new building and development projects have commenced among the 10 EDWC data center clusters. China’s data center rack count has increased by 200,000 to reach a total of 5,400,000. More than 190 billion yuan has been invested in these projects.
In the long term, these developments are building a more complete industry ecosystem. According to the EDWC report:
Once implemented, the Eastern Data, Western Computing project will accelerate the push for greater circulation between the upstream data center equipment manufacturing industry and downstream data essentials as well as the focused development of innovative data applications and new consumer industries. The western regions show promise for attracting labor-intensive industries such as data processing, data cleaning, and data content services.
In addition, the increasing scale of data and level of computing power will stimulate innovation in data factors and lead to a surge of new types of businesses, including data exchanges, computing operators, residual heat brokers, and environmental assessors for data centers.
Demand for technical talent
A common thread through many government messages over the past few weeks is the importance of fostering technical talent. During his address at the recent 20th Party Congress, Xi Jinping emphasized the importance of tech, talent, and education, calling talent China’s “top resource.” Similarly, both the State Council and Ministry of Industry and Information Technology (MIIT) have published documents this month with recommendations for upskilling China’s technical workforce.3
The EDWC initiative is one of the many factors contributing to this demand.
Naturally, an increased focus on technological growth and self-reliance requires a larger and higher-skilled workforce. However, this demand was known well before last month. In their 2020 Report on the Development of Online Learning Platforms for New Vocations, the Center for Employment Training and Technical Instruction (part of China’s Ministry of Human Resources and Social Security) stated that over the next five years, the big data and cloud computing industries would face a shortage of 1,500,000 workers.
As long as China is successful in its attempts to build a skilled technical workforce (and attract talent from abroad as needed), EDWC will provide them with plenty of opportunities. The EDWC report says as much:
EDWC will do more than help transform the imbalanced landscape of China’s data infrastructure. It will also help businesses better offer a range of services — including cloud storage, cloud computing, data tools, R&D platforms, and AI technology — as well as further reduce the costs of cloud migration and accelerate the digital transformation of even more traditional enterprises and small and mid-sized enterprises (SME). A reservoir of digital talent is essential for digital transformation. In its current state, China’s labor market faces a talent shortage.
Upcoming obstacles
The EDWC report also digs into some challenges the project will face as it moves forward. This takes up a sizeable chunk of the document, but to avoid publishing an article as long as the original report, I’ll highlight three areas that stood out to me the most: 1) environmental sustainability, 2) localizing IT, and 3) scaling and integration.
Handling computing facilities’ high energy use
As China’s data center infrastructure grows, so does its energy consumption.
EDWC hopes to innovate how China’s data centers consume power. Most of the country’s data centers are small and scattered; large and hyperscale data centers only make up around 12% of the total. Most of these centers rely on traditional cooling technology. The average PUE (power usage effectiveness)4 of China’s data centers is 1.49, with a significant amount lingering between 1.8 and 2.0. However, new liquid cooling methods can reduce PUE levels down to 1.2, or even as low as 1.0.
Of course, the issue of efficiency is more complicated than swapping out cooling systems. The solution involves many other factors, including using more efficient hardware and continuing to develop more energy-efficient technologies for data centers.
One major aspect of EDWC is its emphasis on leveraging renewable energy in China’s western regions. In this regard, the project seeks to cancel out some of its inevitable energy demand by making its western facilities draw from abundant western resources, particularly renewables like wind and solar, and use natural elements to reduce energy consumption — for example, through natural cooling.
These efforts dovetail with China’s continued efforts to hit its stated “dual carbon” goals: carbon peak by 2030 and carbon neutrality by 2060.
The industrialization of innovative IT
The report points to some notable recent trends in Chinese cloud usage:
2020
A hybrid concept known as “one cloud, many chips” (一云多芯) emerges, in which a single cloud system manages different server clusters using different types of chips.5
Progressive cloud native technologies undergo R&D.
2021
End-to-end domestic cloud technology is employed much more rapidly in areas including government, logistics, finance, transportation, and power. This provides China with a technological foundation that is very much in its power to shape and control.
Thanks to China’s domestic tech prowess, computing power has become a core driver for the development of China’s businesses. The report sees EDWC as riding the wave of China’s growing domestic tech industry.
This last point is an important one, particularly in light of China’s recent official emphasis on technological self-reliance, as well as the United States’ recent restriction of chip exports to China (both of which occurred after the report was published). Broadly speaking, these restrictions could limit the network’s potential in certain regards, such as impacting the network’s ability to scale and optimize its performance through AI-powered analysis or handle expensive AI calculations for clients throughout the network.
Scaling and integrating computing networks
Last year’s published plan for the EDWC project stipulates that initially two data center clusters should be built in the eastern hub node regions, with the western regions receiving one each. Roughly speaking, the geographical motivations behind this are as follows:
In the east: to optimize the supply structure of the region’s databases while expanding their space for growth in computing power, while also satisfying the implementational demands of each area’s developmental strategies.
In the west: to establish a stronghold for non-real-time computing power that can handle the demands directed to it from other regions.
As the EDWC network grows and matures, it will need to develop in terms of three axes that the report defines as follows:
Networking: Reducing delay while increasing bandwidth and reliability
Computational power: Increasing diversity, power-to-consumption ratio, and clustering
Integration: Producing integrated computing networks, developing “computational network brains” (such as city brains), and providing computing services.
The report additionally divides this growth into three steps: 1) coordinated computing networks, 2) integrated computing networks, and 3) an integrated cloud network.
While it’s hard to predict how the recent international developments discussed above will affect the growth of the network in aspects requiring AI, these trends will be reflected through China’s cloud industry — and likely covered in upcoming Cloudology articles.
Final words
To reiterate, the progress of the EDWC project will have a notable impact on China’s economy (data and digital economies included), across its industries (from tech giants to small businesses trying to digitize and modernize), and in how China administers itself (on local, provincial, and national levels alike).
While this will likely be the last EDWC deep dive for a while, I’ll make sure to cover any notable project developments in the future.
Full title: Research Report on the Development of New Types of Computing Infrastructure in the Context of the National Eastern Data, Western Computing Project《国家“东数西算”工程背景下新型算力基础设施发展研究报告》.
The report is titled Research Report on China’s Computing Power Development Index《中国算力发展指数研究报告》. I wasn’t able to track down the original document; instead, the data cited here was included in the EDWC report. If you find a link, feel free to send it my way.
State Council: 《关于加强新时代高技能人才队伍建设的意见》(Oct. 7, 2022)
MIIT: 《工业和信息化部关于加强和改进工业和信息化人才队伍建设的实施意见》(Oct. 18, 2022)
As a frame of reference, notably efficient levels tend to be around 1.1, with 1.2 PUE considered peak efficiency when Google and Microsoft hit around this level in 2008.
This Zhihu response provides a much more detailed explanation of the “one cloud, many chips” concept.