
A young IT engineer inspects data center servers. Stock photo
Written by Wenxuan (Shane) Sun, Business Development & Program Director at New Energy Nexus China
As artificial intelligence (AI) tech progresses, data centers and intelligent computing facilities are becoming a new class of energy-intensive infrastructure. Globally, data centers consumed around 415 TWh of electricity in 2024, about 1.5% of global electricity demand, and the International Energy Agency (IEA) projects that this could roughly double by 2030. The United States and China are expected to account for the majority of this growth.
This creates a critical question for the clean energy transition: will AI-driven computing become another source of grid stress, or can it become part of the solution?
For NEX China, this is the starting point of our work on computing-power coordination, or suan-dian xietong (算电协同) in Chinese, the coordination between computing demand and power system operation.
As we begin a new project to explore the challenges and opportunities in this space, we wanted to cover the basics and what it means for entrepreneurs.
What is computing-power coordination?
Computing-power coordination refers to the practice of aligning computing workloads with the availability, location, timing, and constraints of electricity supply.
In simple terms, not all computing tasks need to happen in the same place or at the same moment. Some workloads are highly time-sensitive, such as real-time financial transactions, autonomous driving, or emergency response systems. But others, including AI model training, batch data processing, rendering, simulation, and certain industrial AI tasks, may have more flexibility. They can potentially be shifted in time, shifted across locations, or adjusted according to grid conditions.
This matters because electricity systems are increasingly shaped by two simultaneous trends. On the supply side, more solar and wind power are entering the grid, but their output is variable. On the demand side, data centers and AI computing loads are growing rapidly, often becoming large, concentrated electricity consumers. This raises a question: when can computing loads function as flexible resources for the grid, rather than only as fixed demand?
The answer is not automatic. Computing loads are not inherently flexible. They only become useful to the power system when the right technical, commercial, and institutional conditions are in place. These include dispatch authority, service-level agreements, measurement methods, settlement rules, and clear responsibility among grid operators, data center operators, computing platforms, and energy users.
Why does this matter for the energy industry?
The energy sector is entering a new phase in which flexibility is as important as capacity.
Historically, power systems were designed around predictable demand and controllable generation. Today, the system must integrate variable renewable energy, electrified transport, distributed solar, batteries, industrial electrification, and now fast-growing digital infrastructure. AI data centers add a new layer of complexity: according to the IEA, AI-focused data center electricity consumption grew by 50% in 2025, while total data center electricity demand grew by 17%.
The challenge is not only the total amount of electricity consumed. It is also where, when, and how that demand appears. Data centers are large, concentrated, and often developed faster than energy infrastructure can be planned and built. The IEA notes that this mismatch between fast-moving data center development and slower-moving energy investment can create risks for grid planning, electricity prices, and system reliability.
Computing-power coordination offers a different lens. Instead of asking only how to supply more electricity to data centers, it asks whether some computing demand can be shaped to support the grid. A few examples:
- A data center could increase computing activity when local solar output is high and reduce or shift non-urgent workloads when the distribution grid is constrained.
- AI training tasks could be scheduled in regions and time windows with abundant renewable energy.
- Computing platforms could offer differentiated service levels, where users pay less for flexible computing tasks that can be delayed or relocated.
- Data centers with batteries, advanced energy management systems, and workload orchestration could participate in demand response or other flexibility markets.
This does not mean turning data centers into power plants; it means recognizing that digital infrastructure and energy infrastructure are becoming interdependent. The next generation of clean energy innovation will not only be about producing greener electrons, but also about designing smarter demand.
Why China?
China is one of the most important places to explore this question because it sits at the intersection of three global trends: rapid growth in AI infrastructure, massive deployment of renewable energy, and real-world grid integration challenges.
China’s total computing power scale already ranks second globally, and by the end of 2023, the country had more than 8.1 million data center racks in use. China’s government has also set clear green data center targets, including lowering the average data center PUE to below 1.5 by 2025 and increasing data center renewable energy utilization by 10% annually.
But China is not only building large data center clusters. It is also facing a very practical distributed energy challenge. County-scale rooftop solar programs and distributed renewables have expanded rapidly in many regions, creating new stress on local distribution grids. In these contexts, renewable generation is often location-bound, while computing loads remain largely inflexible.
This makes China a valuable “stress test” environment. Lessons from China can not be copied directly to Europe, Southeast Asia, or other markets, but they can help answer questions that many systems will soon face: How should grids coordinate with new digital loads? What kinds of computing demand are truly flexible? How should flexibility be measured and rewarded? Where do technical possibilities break down because institutions, contracts, or market rules are not ready?
What can entrepreneurs do?
Entrepreneurs play an important role because computing power coordination is not a single technology. It is an emerging system innovation field that requires new tools, platforms, services, and business models.
Entrepreneurs can develop workload orchestration tools that classify computing tasks by urgency, location sensitivity, carbon intensity, and grid impact. They can build energy-aware AI infrastructure. This may include software that links AI training schedules with renewable energy availability, electricity prices, grid congestion signals, or carbon intensity data. They can also develop measurement and verification systems. Without credible measurement, it is impossible to prove that computing loads have provided real grid value. In addition, there is space for new commercial models: flexible computing contracts, green computing products, demand response aggregation for data centers, carbon-aware cloud services, and location-aware computing marketplaces.
NEX China’s approach is therefore deliberately hypothesis-driven and simulation-based, focusing on decision-useful learning before large-scale deployment. Follow us to stay updated on our findings here.
Wenxuan (Shane) Sun has extensive experience in China’s wind and renewable energy markets, both within the industry and through market consulting.
























