The National Unified Computing Power Network represents a critical mechanism within the broader Digital China stack for delivering computing power at scale. A recent article by journalist Liu Tong in the Ministry of Industry and Information Technology-run newspaper People’s Posts and Telecommunications News frames artificial intelligence as a national requirement for accelerating the construction of New Type Infrastructure. China has spent years expanding computing power at scale. Now the emphasis is shifting to the national system needed to deliver, allocate, and use that computing power efficiently.

Author’s note: A translation of Liu Tong’s “Strengthen Computing Power New Type Infrastructure to Consolidate the Foundation of Digital China” follows below. My summary and computing power systems analysis of this article can be read on its companion page.

Strengthen Computing Power New Type Infrastructure to Consolidate the Foundation of Digital China

夯实算力新基建　筑牢数字中国底座

Author: Liu Tong | 刘彤 (People’s Posts and Telecommunications journalist | 本报记者)

Source: People’s Posts and Telecommunications News1 | 人民邮电, Page 2

Publication date: March 13, 2026

Translation: This is a lightly-edited ChatGPT AI translation.

In recent years, the explosive development of artificial intelligence has driven an exponential surge in demand for computing power. Computing power is now, to an unprecedented degree, driving industrial transformation and reinforcing the foundational base of national strategy. This year’s Government Work Report proposed to deepen and expand the “AI Plus (AI+)” initiative, implement New Type Infrastructure projects such as Ultra Large Scale Intelligent Computing Clusters and Compute-Energy Coordination (Computing Power-Electricity Coordination), strengthen National Unified Computing Power Monitoring and Scheduling, and support Public Cloud development. During the Two Sessions, deputies and members actively offered proposals and recommendations on the development of Intelligent Computing, Compute-Energy Coordination, and Computing Power Scheduling, collectively supporting the High Quality Development of our country’s computing power industry and consolidating the foundation for Digital China construction.

近年来，人工智能的爆发式发展带来算力需求的指数级跃升，算力正以前所未有的深度驱动产业变革、夯实国家战略底座。今年政府工作报告提出，深化拓展“人工智能+”，实施超大规模智算集群、算电协同等新基建工程，加强全国一体化算力监测调度，支持公共云发展。全国两会上，代表委员围绕智算发展、算电协同、算力调度等方面积极建言献策，共同助力我国算力产业高质量发展，筑牢数字中国建设根基。

Policy-Market Resonance2

Computing Power Development “Accelerates”

政策与市场共振
算力发展跑出“加速度”

Our country’s computing power industry has consistently adhered to top-level guidance and systematic planning.3 A series of major policies have been rapidly introduced, providing clear direction and injecting momentum into the industry’s development. The recommendations for the 15th Five-Year Plan, for the first time, incorporate the “National Unified Computing Power Network” into National-Level Infrastructure System and call for its moderately advanced construction;4 the State Council’s Opinions on Deepening the Implementation of the “AI+” Initiative explicitly call for strengthening the coordinated planning of intelligent computing power, enhancing interconnection and interoperability, and improving the matching of supply and demand;5 the Ministry of Industry and Information Technology (MIIT) has also intensively introduced a package of policies including the Notice on Organizing and Launching the Computing Power Infrastructure Enhancement Challenge Campaign, Action Plan for Computing Power Interconnection and Interoperability, and the Notice on Launching the Metropolitan-Area ‘Millisecond Computing Access’ Special Action, collectively pressing the “accelerator” for the development of the computing power industry.

我国算力产业发展始终坚持顶层引领、系统布局，一系列重磅政策密集落地，为产业发展指明方向、注入动能。 “十五五”规划建议首次将“全国一体化算力网”纳入国家级基础设施体系，要求适度超前建设；《国务院关于深入实施“人工智能+”行动的意见》明确提出，强化智能算力统筹，加强智能算力互联互通和供需匹配; 工业和信息化部密集出台《关于组织开展算力强基揭榜行动的通知》《算力互联互通行动计划》《关于开展城域“毫秒用算”专项行动的通知》等政策“组合拳”，为算力产业按下发展“加速键”。

Driven by the release of policy dividends and the joint efforts of stakeholders across the industry, our country’s computing power network construction has achieved leapfrog development.6 It is now accelerating its transition from “scale expansion” to “value release,” entering a new stage centered on “efficient service.”7 By the end of 2025, our country’s computing power market had reached 835.1 billion yuan, representing year-on-year growth of over 30 percent. The scale of intelligent computing power exceeded 1,590 EFLOPS. Both total computing power and intelligent computing power rank second globally, and a diversified, coordinated computing power landscape,8 integrating general-purpose computing, intelligent computing, and supercomputing,9 has now taken shape. Regarding the next stage of computing power network development, Lou Xiangping, a deputy to the National People’s Congress and General Manager of Shanghai Mobile, proposed focusing on key national strategic regions and encouraging the construction of E-class intelligent computing power super nodes,10 in order to strengthen foundational support for the training and inference of trillion-parameter large models. He also called for forward-looking deployment of a new supercomputing architecture that integrates “quantum + classical” computing, accelerating breakthroughs in key technologies and scenario validation, and proactively cultivating a next-generation computing power technology system to contribute China’s capabilities to the coordinated development of global computing power.

在政策红利释放和产业各方的共同推动下，我国算力网络建设实现跨越式发展，正加速从“规模扩张”走向“价值释放”，迈入以“高效服务”为核心的新阶段。截至2025年年底，我国算力市场规模高达8351亿元，同比增长超30％，智能算力规模超过1590EFLOPS，算力总规模与智能算力规模均位居全球第二，通算、智算、超算多元协同的算力格局已然成型。 于算力网络下一步发展，全国人大代表、上海移动总经理楼向平建议，聚焦国家重大战略区域，鼓励构建E级智能算力超节点，强化对万亿参数级大模型训练与推理的底层支撑能力。 前瞻布局“量子+经典”混合的超算新架构，加速关键技术攻关与场景验证，超前培育下一代算力技术体系，为加快全球算力协同发展贡献中国力量。

Confronting New Challenges in Coordination

Resolving the ‘Growing Pains’ of Industry Development

直面协同新挑战
破解产业发展“成长的烦恼”

The rapid development of the computing power industry has also been accompanied by a series of pressing challenges that must be addressed, such as the coordination of computing power and electricity, bottlenecks in the efficiency of computing power scheduling, and the absence of market-based pricing mechanisms, which have become the focus of proposals and recommendations by deputies and members.

算力产业的飞速发展也伴随着一系列亟待破解的难题，如算力与电力的协同、算力调度的效率瓶颈以及市场化价格机制的缺失等，成为代表委员们建言献策的焦点。

“The end of AI is computing power, and the end of computing power is electricity.” This widely cited industry phrase succinctly captures the deep dependence of computing power on energy. Computing power is a typical high energy–intensive industry. A mid-sized intelligent computing center with 5,000 racks consumes as much electricity annually as approximately 100,000 households. However, the coordinated development of computing power and electricity faces multiple challenges. First is spatial misalignment.11 The eastern region accounts for more than 60 percent of national computing power demand, yet holds less than 20 percent of energy resources; as a result, supporting power infrastructure for intelligent computing centers in the east must be urgently strengthened. Second is cost pressure. Electricity expenses now account for more than half of the operating costs of intelligent computing centers, constraining the broad-based development of the artificial intelligence industry. Finally is the energy transition. The eastern region must accelerate the development of New Type Energy Systems such as coastal nuclear power and deep-sea offshore wind to support the massive future demand for green computing power.

“AI的尽头是算力，算力的尽头是电力。”这句业内的流行语精准地道出了算力对能源的巨大依赖。 算力是典型的高载能产业，一座拥有5000个机柜的中型智算中心，其年耗电量堪比10万户家庭的用电总和。然而，算力与电力的协同发展面临多重挑战。 首先是空间错位。东部地区算力需求占全国60％以上，但能源资源占比不足20％，东部地区智算中心供电配套亟待加强。 其次是成本压力。电费支出已占智算中心运营成本的一半以上，制约了人工智能产业的普惠发展。最后是能源结构转型。东部地区亟须加快沿海核电、深远海风电等新型能源体系建设，以支撑未来巨量的绿色算力需求。

In response to these challenges, Yang Jianyu, a deputy to the National People’s Congress and General Manager of Zhejiang Mobile, proposed: “On the one hand, guide ‘computing power to follow electricity,’ continuously shifting high-load, energy-intensive AI training demand to western regions. On the other hand, encourage ‘electricity to follow computing power,’ advancing the construction of cross-provincial and cross-regional power transmission channels, as well as supporting energy storage capacity.” He also called for the implementation of preferential electricity pricing policies for intelligent computing centers,12 and for exploring “direct compute–energy connections”13 between computing centers and power generation enterprises including renewable energy, hydropower, and nuclear power providers. Policy support should be provided in areas such as transmission and distribution pricing and project approvals, in order to effectively reduce energy costs and “invigorate” the development of our country’s artificial intelligence industry.

针对这一问题，全国人大代表、浙江移动总经理杨剑宇建议：“一方面引导‘算力跟着电力走’，持续推动高负载、高能耗的人工智能训练需求向西部地区转移。另一方面鼓励‘电力跟着算力建’，落实跨省跨区输电通道、储能配套等能力建设。” 同时，他还呼吁落实对智算中心的电力优惠政策，探索算力中心与绿电、水电、核电等发电企业开展“算电直连”，在输配电价、项目审批等方面给予政策扶持，切实降低用能成本，为我国人工智能产业“舒筋活血”。

In addition to the challenges of compute–energy coordination, the interconnection, interoperability, and efficient scheduling of computing power resources also face practical constraints. Uneven resource distribution and a diversified technical ecosystem have resulted in persistent pain points in cross-regional, cross-entity, and cross-architecture coordination, namely that computing resources are “hard to find, difficult to dispatch, and poorly utilized.” In response, the Ministry of Industry and Information Technology (MIIT) recently issued a notice clarifying the advancement of a “1+M+N” National Computing Power Interconnection Node System,14 aimed at activating public computing resources15 and enabling computing power to be accessed “anytime, anywhere, on demand.”16 To date, the system has connected 578 resource pools from 155 enterprises across 31 provinces (autonomous regions and municipalities), encompassing 316 EFLOPS of intelligent computing resources. With an average of nearly 300 scheduling operations per month, it is gradually addressing the challenges of matching supply and demand and reducing the cost of computing power usage.

除了算电协同难题，算力资源的互联互通与高效调度也面临现实挑战。资源分布不均、技术体系多元，导致跨区域、跨主体、跨架构的算力协同存在“找不着、调不动、用不好”的痛点。为此，工业和信息化部近日发布通知，明确推进“1+M+N”国家算力互联互通节点体系建设，旨在盘活公共算力资源，实现算力“随时、随地、按需”调用。截至目前，已接入31个省（区、市）的155家企业的578个资源池、316EFLOPS智算资源，月均调度近300次，逐步解决算力供需匹配和用算成本问题。

The distinctive characteristics of the compute economy (literally computing power economy) have also introduced new governance challenges. Unlike the digital economy, where greater scale drives marginal costs toward zero, the compute economy17 exhibits the opposite pattern: greater investment is associated with rigid marginal costs. “Our country has yet to establish a market-based pricing mechanism in the computing power sector that encompasses price formation, transaction matching, and pricing indices. Addressing gaps in the price formation mechanism has become an urgent priority,” said Zhang Yunquan, a member of the Chinese People’s Political Consultative Conference and a researcher at the Chinese Academy of Sciences Institute of Computing Technology. To this end, he proposed establishing a pricing standard for intelligent computing power based on ‘yuan per million Tokens,’ defining standardized service specifications and reference price ranges across different model sizes, levels of precision, and response times, and using pricing mechanisms to incentivize computing centers to improve operational efficiency. He also recommended the creation of an independent, third-party national computing power exchange,18 separate from both supply and demand sides, to build a unified national market for computing power trading.

算力经济的独特规律也带来了新的治理课题。与数字经济时代“规模越大、边际成本趋零”不同，算力经济呈现出“投入越大、边际成本刚性”的特征。“我国算力领域尚未形成覆盖价格形成、交易撮合与价格指数的市场化定价机制，补齐价格形成机制已成当务之急。 ”全国政协委员、中国科学院计算技术研究所研究员张云泉表示。为此，他建议制定以“元／百万Token”为基础的智能算力计价标准，明确不同模型规模、精度、响应时间下的标准服务规格和参考价格区间，以价格机制激励算力中心提升运营效率。 设立独立于算力供需双方的第三方全国性算力交易所，建设全国统一的算力交易市场。

Looking ahead, advancing the High Quality Development of our country’s computing power industry will require adherence to systems thinking and a multi-pronged approach: continuously deepening compute–energy coordination, introducing green energy to build a low-carbon computing power supply system; leveraging the development of computing power interconnection nodes to improve scheduling efficiency and reduce the cost of computing power usage across the economy; and accelerating the improvement of market-based price formation mechanisms while exploring the establishment of a society-wide computing power trading market. As the National Unified Computing Power Network becomes increasingly dense and fully developed, computing power will truly become as accessible and efficiently circulated as water and electricity, providing a continuous and powerful source of momentum for the construction of Digital China.

展望未来，推动我国算力产业高质量发展，需坚持系统思维、多措并举：持续深化算电协同，引入绿色能源打造低碳算力供给体系；依托算力互联互通节点建设提升调度效率，降低全社会用算成本；同时加快完善市场化价格形成机制，探索建立全社会算力交易市场。随着全国一体化算力网不断织密完善，算力将真正实现像水、电一样便捷取用、高效流通，为数字中国建设注入源源不断的澎湃动力。

Footnotes

  1. 人民邮电 is also abbreviated as Posts & Telecom. ↩︎
  2. In this context resonance (共振) implies synchronization between state policy and market forces. ↩︎
  3. Top-level guidance and systemic planning (顶层引领、系统布局) signals central design and system-wide coordination. ↩︎
  4. Moderately advanced construction (适度超前建设) is a key policy phrase to build ahead of demand, but not excessively. ↩︎
  5. Coordinated planning, interconnection and interoperability, and supply-demand matching (智能算力统筹, 互联互通, 供需匹配) point to a governed system, not just infrastructure. ↩︎
  6. Leapfrog development (跨越式发展) is an important concept in development discourse indicating not just rapid, but stage-skipping, development. ↩︎
  7. A new stage centered on efficient service (高效服务为核心的新阶段) signals a transition to utilization and service delivery, not just infrastructure. ↩︎
  8. A diversified, coordinated computing power landscape (多元协同的算力格局) points to system integration not just coexistence. ↩︎
  9. General purpose computing, intelligent computing, and super computing (通算、智算、超算) highlight the tri-layer structure of China’s compute ecosystem. ↩︎
  10. E-class likely refers to Exa-scale (10¹⁸) computing. Super nodes” suggests highly concentrated compute hubs within the national network. ↩︎
  11. Spatial misalignment (空间错位) points to a systems mismatch between demand geography and resource geography. ↩︎
  12. Preferential electricity pricing policies (电价优惠政策) indicate state intervention in cost structures. ↩︎
  13. Direct computer-energy connection (算电直连) is a key emerging concept that describes bypassing traditional grid pricing structures to link compute centers directly to generation sources. ↩︎
  14. “1+M+N” refers to a formal system architecture model, likely 1 (national hub/core) + M (regional hubs) + N (distributed nodes/resources). ↩︎
  15. Activating (literally revitalizing) public computing resources (盘活公共算力资源) indicates improving utilization of existing capacity not just building more. ↩︎
  16. Anytime, anywhere, on demand (随时、随地、按需) is standard cloud computing/service language. Similarly, resource pools (资源池) is a standard term for cloud/compute infrastructure. ↩︎
  17. The compute economy (算力经济), literally computing power economy, is referenced here as a distinct economic domain. ↩︎
  18. An independent, third-party national computing power exchange (独立。。。第三方全国性算力交易所) points to an institutional innovation similar to an energy or commodities exchange. ↩︎

Feedback & Corrections: Your questions, comments, additions, or corrections are always welcome! Please contact me at digitalchinawinsthefuture@gmail.com

AI Transparency: I use AI tools to support my editing, research, and translation process. Learn more on my AI Transparency Page.