Grid Expansion Lags Behind AI Data Center Power Demand; Supply-Demand Coordination Emerges as Key to Integrating Energy and Digital Economy

April 2026 — At CERAWeek, the energy conference held in Houston, USA, consensus emerged among the U.S. Federal Energy Regulatory Commission (FERC), leading domestic grid operators, and stakeholders from the energy and computing industries regarding grid strain caused by surging electricity loads from data centers. The core agreement: implementing flexible load curtailment at data centers during peak grid stress periods is the essential pathway to resolving the conflict between lagging grid expansion and explosive power demand from the computing sector. This in-depth industry discussion further underscores the decisive importance of stable power supply and high-quality electrical equipment for the coordinated development of the digital economy and the energy system.

In recent years, the rapid global expansion of the AI industry has driven continuous growth in data center construction, with electricity demand experiencing explosive growth and placing unprecedented strain on existing power systems. Stu Bresler, Chief Operating Officer of PJM Interconnection, stated unequivocally at the conference that without flexible demand management, interconnection queue wait times for new large-load users would exceed a decade, severely constraining the expansion pace of the computing industry.

As core infrastructure for the digital economy, the stable operation of data centers relies heavily on continuous and reliable power supply. Electricity assurance capabilities directly determine the continuity and security of computing services. The public grid, serving as the primary power source for data centers, directly influences both the safe and stable operation of the power system and the development potential of the digital economy through its hosting capacity and dispatch flexibility. Faced with the reality of lengthy grid infrastructure construction cycles versus rapidly growing computing power demand, demand-side flexibility is widely recognized by the industry as the critical solution.

Data from the Electric Power Research Institute (EPRI) indicates that severe supply stress on the public grid occurs for only a few dozen hours annually—less than 1% of total yearly hours—typically concentrated during extreme weather events. Concurrently, although data centers apply for grid interconnection based on their maximum potential load, they rarely operate at full capacity around the clock. Regulators therefore propose that if data centers could reduce load during these brief peak stress periods, the power system could significantly increase its capacity to absorb new demand without requiring massive new infrastructure. Research from Duke University further substantiates this approach: within the service territory of PJM, the largest U.S. grid operator, a 0.5% load reduction by users during approximately two peak days per year could unlock 18 GW of additional load interconnection capacity without physical grid expansion.

However, industry representatives at the conference widely acknowledged that implementing load curtailment measures hinges on clear and robust financial incentives. FERC Commissioner Judy Chang noted during the session that the industry must focus on demand-side management potential, integrating large data centers' load curtailment capabilities into the power dispatch framework. This would involve providing economic incentives for large power users that respond to grid signals and voluntarily reduce load, mirroring compensation mechanisms on the generation side—a framework that currently remains underdeveloped.

Industry insiders point out that appropriate incentive mechanisms will effectively encourage data centers to proactively upgrade their electrical infrastructure. High-quality, advanced power equipment is the core enabler for data centers to balance grid dispatch requirements with their own operational security. Whether high-reliability Uninterruptible Power Supply (UPS) systems, large-scale energy storage installations, intelligent load management systems, or efficient backup generator sets, such premium equipment not only ensures business continuity during load adjustments but also significantly enhances the flexibility and feasibility of data centers' participation in demand response, achieving a win-win scenario for both grid reliability and data center operations.

Divergent views on the practical feasibility of load curtailment were evident during the conference. Amanda Peterson Corio, Global Head of Data Center Energy at Google, noted that certain AI training workloads can be paused or rescheduled, offering some degree of flexibility. However, she emphasized that this flexibility is site-specific, dependent on factors including operating permits, on-site equipment configuration, and the nature of the computational workloads. David Porter, Vice President of Strategy at EPRI, added that "compute factories" focused on AI training possess greater flexibility, whereas data centers hosting enterprise-grade services have workloads with little to no adjustment margin.

Kerry Person, Vice President at Amazon Web Services (AWS), stated bluntly that the vast majority of AWS sites cannot curtail power usage on demand, as any interruption to the cloud services and enterprise workloads they host would directly impact a vast number of end-users. Currently, the only industry method to achieve grid load reduction while maintaining operational continuity is switching to on-site backup generators. However, existing U.S. environmental regulations stipulate that backup generators may only be used in emergency scenarios. Most backup generators are fossil-fuel-powered, and battery energy storage technology is not yet mature enough for utility-scale application requirements. This regulatory contradiction presents a major obstacle to implementing flexible demand management.

This industry dilemma further highlights the urgency of technological advancement in power equipment. Experts argue that the industry urgently needs breakthroughs in energy storage technology, alongside the development and deployment of more environmentally friendly, faster-responding, and scalable backup power and storage solutions. Simultaneously, upgrading intelligent power management systems is necessary to resolve the conflict between grid dispatch requirements and operational continuity through technological means, thereby providing a solid hardware foundation for data centers to participate in flexible power system scheduling.

Reportedly, practices related to flexible load management in U.S. data centers remain in the pilot phase. The DC Flex initiative launched by EPRI is conducting on-site demonstration tests at operational data centers to validate the safety boundaries of load curtailment and the feasibility of data centers providing ancillary services such as frequency support to the grid.

Industry executives also issued a warning: if regulators and the data center industry fail to reach a workable consensus on peak load curtailment rules, large-scale computing users may abandon lengthy grid interconnection approval processes and resort to building on-site generation facilities to meet their power needs. Such a shift would not only substantially increase overall industry operating costs but also lock in further fossil fuel consumption, contradicting global low-carbon energy transition goals and ultimately weakening the resilience of the public power system's development.

Currently, the deep integration of the digital economy and the energy system represents an inevitable global trend. The growing power demand from the computing industry and the safe and stable operation of the power system are not contradictory, but form a community that requires coordinated development. Stable and reliable power supply is the lifeline for the high-quality development of the digital economy. In an era where power supply carries such vital importance, Hengda's power and voltage regulation and other equipments enable customers to achieve energy conservation, carbon reduction, cost reduction and efficiency improvement. Technologically advanced and high-performance power equipment serves as both the core foundation ensuring the safe operation of data centers and the critical link connecting computing power demand with grid dispatch, thereby driving the low-carbon energy transition. Going forward, only through tripartite coordination — refined regulatory policies, implemented incentive mechanisms, and upgraded technological equipment — can true two-way empowerment be achieved between the development of the computing industry and the security and stability of the power system.