Guizhou – a tier-two city located in the southern region of China – has been branded as the poster-child of China’s big data revolution, but in reality this reputation is slightly misleading. Guizhou’s data centers have been underutilized because it does not serve any major economic zones in the region.
In July 2021, the Ministry of Industry and Information Technology (MIIT) announced a 3-year plan to push for improving data center network interconnectivity, storage optimization, computing capacity, and energy efficiency. The data center problem lies at the heart of China’s broader digital ambitions. Failure to properly orchestrate the construction of this fundamental data infrastructure could slow down the upwards trend of China’s digital economy trajectory – regulators are determined to get it right.
The perennial problem lies in misalignment between supply and demand, which appears to be a natural consequence of top-down implementation of industrial policy. The Open Data Center Committee of China reports that metropolitan areas enjoyed a high demand (80% utility) for low-latency computing, regional data centers sit idle (below 30% utility) because of low computing and storage demand in rural regions. Tech regulators are faced with prescriptive constraints – given that data centers ingest incredible amounts of energy, Beijing wishes for “green data centers'' optimized for more efficient energy usage, capacity and performance.
Current Challenges
As the digital economy develops and matures in China, data centers are expected to facilitate higher performance in handling real-time data requests for digital transactions. So with about 783 million mobile users, it is no surprise that investment in data centers is expected to more than quadruple from 300 billion yuan ($46.6 billion) to 1.4 trillion yuan ($216 billion) between 2020-2023, according to the China Academy of Information Communication Technology.
The cautionary tale of this big data ambition lies in the local-central government funding tug-of-war. The appeal for the Chinese local government to over invest in data centers is similar to that of real estate land-based financing, thus raising prospects of a possible data center bubble. Although China’s Government Guidance Funds (GGF) – a form of public-private hybrid capitalization method for emerging technologies development – are no silver bullet against increasingly intense tech competition from the U.S., their effectiveness at creating demand-driven data centers has been suboptimal, at least in the short run.
Implications for players of the big data market
Made up of complex hardware systems of data cabinets, routers, storage systems, and servers, data centers reside at the upstream part of China’s big data value chain. Major telecom players like China Mobile, China Telecom, and China Unicom operate most of the data centers. Together, they hold about 60% of the market. Midstream players are third-party hardware security providers, and downstream players are major commercial cloud companies like Alibaba and Tencent.
Because of the discounted fixed-cost of land and electricity cost provided by the government incentives, big data operators have swarmed to remote regions to build their data centers. For example, all three telecom players occupied 800,000 meters squared and collectively constructed approximately 180 thousand data cabinets in Guizhou in 2016. They also made comparable investments in Inner Mongolia.
To offset the lower fixed cost, the downside of operating data centers in regional areas include longer distance and less network connection to their downstream internet companies. Internet companies are incentivized to use low latency (or minimal delay) computing and closer proximity to support operations that require near real-time access to rapidly changing data. This explains why underutilization rates in regional data centers are higher than rates among urban data centers.
The 3-year plan will push for more sustainable energy sources to power data centers. As regulators are aware of the huge consumption of energy in low-latency computing data centers, they are steady in pushing for renewable energy sources in computing and 5G operators. It is estimated that in 2035 China's digital infrastructure will emit 310 million tonnes of carbon, and data center computing emissions are estimated to double. What still remains a challenge is that regulators will have to find a mix of pricing and environmental-economic instruments to encourage tech companies to store and compute data sustainably without decelerating digital economic growth.
What’s to come
The 3-year-plan provides a solution to the underutilization issue by optimizing location to the functionality of the data center. In other words, location matters to building data centers with different purposes. Economic zones in the Yangtze River Delta, Beijing-Tianjin-Hebei Region, Greater Bay Area, and Chengdu-Chongqing Economic Circle will most likely build data centers for low-latency computing used for real-time public health monitoring, online gaming & streaming, data markets trading, and e-payment financial sectors – all of which directly contribute to economic growth. Data centers in Guizhou, Inner Mongolia, Gansu, and Ningxia will handle less demanding requirements for computing, storing non-transactional and more static data.
In good Chinese industrial policy fashion, the roll out of this 3-year plan will be coordinated across China in the forms of pilot zones via handpicked technology firms and SOEs. The plan demands 50 pilot projects to be created by 2023, focusing on building interconnectivity among mega-data hubs across the country with cloud technology. It’s expected that regulators will guide the establishment of a national computing power service evaluation system that tracks the supply and demand of computing power resources nationwide. Furthermore, regulators also hint at expanding these big data pilot zones along the Belt Road economic zones, effectively “strengthening coordination with China’s sea and land cables and other international communications infrastructure.”
Despite the hype over the new fleet of data centers, the execution and investment remains the biggest uncertainty. China’s recent data security laws and crackdown on tech already raised challenges for data processors and cloud operators. As regulators continue to put these laws into effect, it’s forecasted that foreign investors of China’s big data industry will continue to face red tape and remain financially unsatisfied. But when push comes to shove, the regulators will have to choose between prioritizing data innovation-security matters or the environment. It is likely they will choose the former.
