digital heights
The scale of the PRC’s digital infrastructure and data resources is beyond doubt. It is world no. 2 in total computing power and home to 60 percent of world 5G users. Total data output also ranks no. 2. Items underpinning these rankings are targeted to multiply further by 2025.
February 2023’s 'Digital China’ blueprint lays out a broad digitisation vision, encompassing the economy, governance, culture, society, and ecology. Nothing is to be without a digital component, from factories, farms, health services, to commerce and literature.
Labelled ‘2522’ (an acronym for ‘2 foundations, 5 integrations, 2 major capabilities, 2 environments’, as shown in the graphic below), the framework aims to set up digital infrastructure and a data resource system as the ‘two foundations’.
But simply building out the infrastructure and resources has little point absent a system to ensure their use. Hence, the plan provides a ‘ghost in the machine’, as the graphic below shows: a smart layout of ‘hard’ physical infrastructure (left), with a ‘soft’ data resource system serviced by regulatory and institutional mechanisms (right). With these conjoined, data will be efficiently allocated. In physical terms, it will flow through an advanced network of cables, 5G stations, and data centres; in abstract terms, (market) institutions and rules will guide the flow in a way that unleashes its value.
digital infrastructure
‘Digital infrastructure’ consists mainly of cloud computing centres, data centres and network technology connecting them. As ‘new infrastructure’, strong state support has contributed to rapid growth. But officials are now mindful of 'data centre hype’ and overcapacity, recalling bouts of ‘infrastructure mania’ that bear no repeating.
Regional balance is another bugbear. Only 20 percent of the country’s data centres are located in western regions, even though land and electricity costs are lower there. But computing power and data are needed primarily in the east, home to most high-tech firms. Carbon neutrality lurks in the background, as data centres account for 2 percent of total PRC electricity usage. Rather than more infrastructure, Digital China calls for ‘efficiency, complementarity, and coordination of computing power’.
Efficiently connecting and integrating resources are hence at its core. Beijing flagged plans for a ‘rational layout' and integration of data centres from 2020-25 (in 2021 adding computing centres).
Dubbed ‘Eastern Data, Western Computing’, the strategy aims to set up national digital nodes in major cities in western and eastern China, as shown on the map below (will be finalised on Monday).
- sited in the West, heavy computing centres will benefit from energy consumption quotas, lower energy costs, and cooler climate
- they will be linked to Eastern provinces where firms, not least in the AI sector, need computing resources
- sectors with lower latency requirements, e.g. AI model training or long-term data storage, should lead the move to the West
Building connectivity, however, hits a number of practical walls, e.g. a lack of interoperable cloud computing platforms, which is problematic when different types of cloud systems come together. Key technologies, such as multi-core optical fibres, are due breakthroughs.
The challenges are not purely technical: operating firms resist joining up; utilisation rates in private enterprise clouds hover around 5-10 percent. Some suggest building an open platform integrating existing systems. A national computing infrastructure research centre may be needed to make it happen.
joining up super powers
Connectivity is also an emerging focus in the supercomputer realm, a crucial part of digital infrastructure. In 2022 the PRC was credited with 162 of 500 ‘most powerful supercomputers in the world’. National supercomputing centres are spread around the country: Wuxi, Tianjin, Jinan, Shenzhen, Changsha, Guangzhou, Zhengzhou, Kunshan, Chongxin, Xi’an... Yet not well connected to users or each other, their functionality is suboptimal.
A new national consortium for the ‘supercomputing internet’, launched by MoST (Ministry of Science and Technology), will serve as a ‘high-speed road’ to Digital China, claims Mei Jianping 梅建平 MoST High and New Technology Bureau. It promises, says Pan Jingshan 潘景山 Jinan National Supercomputing Centre, to deliver computing power to end users anywhere and anytime, just like ordinary people and enterprises can now use electricity from the power grid, without a need to bother about where it was generated and whether it comes from a nuclear, solar, wind, or thermal power source.
The first supercomputing internet project was launched in Jinan in July 2022, now reaching all 16 of Shandong’s cities. Expanding this acknowledges Pan will not happen overnight.
unlocking the value of data
To unlock its value and potential, data must be shared. ‘Hard’ infrastructure physically enables sharing, but regulatory and institutional support are needed for the smooth and efficient allocation of data resources. This underlies the second of Digital China’s ‘two foundations’. Deemed to have a competitive advantage due to the quantum of available data, the PRC’s focus is now on using it.
Data has been designated a major ‘factor of production’, added along with technology to the traditional trio of labour, land, and capital. This, among other things, serves to demarcate the roles of state and market (doctrinal details are still being figured out). It is clear that intellectual property for data, and research on data pricing mechanisms, will be needed to realise plans for data marketisation.
Data trading remains small-scale. A case in point is the pioneering Guiyang Big Data Exchange Centre (GBDex), which opened in 2015: only some 23 percent of its products had ever been traded as of 2022. That with the highest volume had been traded only 31 times. Other exchanges are even weaker: in Hainan, only some 11 percent of products have seen trades; the most successful of these traded a mere 12 times.
Ninety-five percent of data trading occurs ‘off-site’, notes Yang Jianyu 杨剑宇 NPC delegate and China Mobile’s Zhejiang. Causes are legion, he adds, noting
- unclear data intellectual property right (IPR)
- unclear data pricing mechanisms
- lack of trust in data transfer mechanisms and tech
- flawed top-level marketisation design
- unwillingness to share data
- legal focus on data standardisation, security and privacy, yet lack of clarity on access and supervision of marketised data flows
Trade entails transferring ownership, whose definition is taken for granted, between entities, but data ownership remains in the grey zone. To bring certainty regarding data IPR, He Zhimin 何志敏 CNIPA (China National Intellectual Property Administration) urges clarification of data IPR authorisation processes, the scope of data IPR protection and the difference between data IP and public data holding rights.
data governance: from local to national
Localities are moving faster than the centre building the ‘soft’ data resource foundation. Shanghai, Guangzhou, Zhejiang, et al. now regulate data trade, setting terms for a future national legal framework, hopes Tang Hanlin 汤寒林 Guizhou Data Pay Network Technology founder.
Keen to ‘export’ its model to the national level, Guizhou offers its GBDex as a model, aiming to solve some of the key obstacles. In April 2023, for instance, it launched the first PRC automatic data trading price calculator, helping solve information asymmetries that hamper pricing agreements.
Yet national data sharing and trading regulation remain scarce. With the national data administration (NDA) promised at the March 2023 Two Sessions, there will at least be a national-level agency. Housing it at NDRC signals a role in economic development beyond regulating the data industry.
The NDA will, when operational, hold the reins of ‘Digital China’, deciding how to roll out the ‘national data strategy’ and its digital infrastructure. The establishment of the agency signals that a modern society and the ‘new development pattern’ are in need of digital foundations. The NDA’s build-out remains to be seen. The trajectory is clear: digital resources must flow to unleash value.
super computers
National Supercomputer Centre in Tianjin (NSCT) | 国家超级计算天津中心
Tianjin’s National Supercomputer Centre (NSCT) was the first national-level supercomputing centre, approved by MoST in 2009. It was jointly established by Tianjin’s National Defence Science and Technology University and Binhai New District. It has reputedly served some 1600 heavy computation users in biomedicine, aircraft flight simulation, weather modelling, petroleum exploration and more.
NSCT built and houses the Tianhe-1 (天河一号) supercomputer, ranked the world’s most powerful in 2010-11 with a speed of 2.5 petaFLOPS. Costing USD88 million to build, Tianhe-1 runs up some $20 million in annual power and operating expenses. Tianhe-2, its successor, is sited at Guangzhou’s National Supercomputer Centre; the most recent prototype Tianhe-3 is again a Tianjin project. Sunway TaihuLight (神威·太湖之光) was built at Wuxi’s National Supercomputing Centre and ranked the world's fastest supercomputer from 2016-18.
Tianhe-1 runs on foreign chips (Intel Xeon X5670 6-core processors and Nvidia M2050 GPUs), whereas those in Sunway TaihuLight and Tianhe–3 are domestically produced. Tianhe-3’s ‘FeiTeng’ chips were made at Tianjin Phytium Technology, and added to the US ‘entity list’ in 2021, along with another six firms and institutions in the same sector. Little has been divulged about Tianhe-3, even its rated speed; hence, its absence from TOP500, a list ranking world supercomputers. It is reputedly at least as fast as the current No.1, US-developed ‘Frontier’.
Gao Wen 高文 | Peking University Faculty of Information and Engineering Sciences director
Gao envisions computing power delivered to consumers via a unified, standardised system, just as electric power is supplied by a common grid regardless of its source. Yet the computing power internet faces challenges, he admits, given the plethora of current systems, with low interoperability. He argues that a computing infrastructure research centre is called for at the national level, creating an open platform linking systems. The main challenges are the unwillingness of firms to cooperate and share their resources, standards ensuring interoperability and advancing critical technologies like multi-core optical fibres, coherent optical communication, and wavelength division multiplexing.
A computer vision expert, Gao founded Peng Cheng Laboratory, a province-level lab in Shenzhen. He is a Chinese Academy of Engineering Academician. He was Chinese Academy of Sciences graduate School vice president, National Natural Science Foundation of China vice president 2013-18, and China Computer Federation president 2016-20. He has served on standard working groups at the US-based Institute of Electrical and Electronics Engineers (IEEE). Gao is a top-level policy advisor, lecturing, for example to a collective Politburo study session on AI. A Harbin Institute of Technology graduate, Gao holds a PhD in electrical engineering from the University of Tokyo.
Jing Yaping 景亚萍 | Guizhou Provincial Big Data Development Administration director
Jing joined Guizhou’s Provincial Big Data Development Administration as deputy director in 2016 when Guizhou was approved as the first National Big Data Comprehensive Experimental Zone. Now its director, she seeks to move Guizhou from 'test zone 1.0' to 'innovation zone 2.0', setting up 'three 100-billion industries' of data centres, smart terminals, and data applications. A keen supporter of the 'Eastern Data, Western Computing' strategy, Jing wants to position Guizhou strategically as one of the five Western computing nodes.
Eager to ‘export’ Guizhou's as a model elsewhere, at this year's CPPCC (Chinese People’s Political Consultative Conference) Jing proposed building a national-level data trading exchange akin to the local Guiyang Big Data Exchange Centre (GBDex). Whether Guizhou might provide a model or inspiration for the newly established national data administration is unclear. Jing made other proposals to the CPPCC for infrastructure and a national computing power network, all consistent with Digital China’s ‘dual foundations’.
Directing Guizhou’s Big Data Development Administration, Jing is concurrently a CPPCC member. She also chairs the provincial Women's Scitech Workers Association. She is deputy chair of the Guizhou committee of the Zhi Gong Party, one of the eight ‘satellite’ political parties, and a member of its science and technology committee. However, she also joined the CPC in 2005.
Holding a PhD in management, Jing formerly worked in Guizhou Institute of Finance and Economics, Guizhou Institute of Technology, and Guizhou Academy of Sciences.
context
27-28 Apr 2023: Fifth Digital China Summit held in Fuzhou
25 Apr 2023: Ding Xuexiang 丁薛祥 vice premier highlights the importance of the digital economy for the ‘Global Development Initiative’ at UN forum
17 Apr 2023: MoST kicks off national supercomputing internet consortium
08 Mar 2023: new national data administration announced at the ‘Two Sessions’
27 Feb 2023: overall layout plan for digital China construction issued
19 Dec 2022: State Council opinions pledge to build a data foundation system to better utilise data as a factor of production, including through systems for data property rights, data sharing, revenue distribution, data governance, data security
25 Jul 2022: inter-ministerial joint conference system for the digital economy set up
12 Jan 2022: 14th 5-year plan on the digital economy
06 Jan 2022: pilot plan for the market-based allocation of factors of production to expand regulatory context for market-based data trading
28 Dec 2021: 14th 5-year plan on smart manufacturing
27 Dec 2021: 14th 5-year plan on digitisation
14 Jul 2021: development of new generation data centres three-year action plan (2021-23) proposes a more rational layout of data centres, aka ‘Eastern Data, Western Computing’
06 Nov 2019: 4th plenum decides data is a factor of production
10 Aug 2018: 3-year action plan for digital economy (2018-20)
23 Apr 2018: The first Digital China Summit held in Fuzhou
29 Sep 2016: guiding opinions on internet+ governance
26 May 2016: big data 5-year plan
16 Nov 2015: State Council approves Guizhou as the first national big data pilot zone
05 Sep 2015: State Council action outline for promoting the development of big data