The AI boom is creating a new infrastructure paradox: the same technology driving a surge in data center electricity demand is increasingly being used to help utilities manage the resulting strain. A new survey from National Grid Partners finds that utilities are deploying AI to handle rising interconnection demand, improve grid reliability and control operating costs as AI data centers put additional pressure on power infrastructure.
National Grid Partners, the corporate venture capital and innovation arm of National Grid, says 78% of surveyed utility innovation leaders are deploying or operationalizing at least one AI application to manage data center interconnection demand. At the same time, 74% said AI-driven data center load growth is already affecting grid reliability.
The findings come as utilities confront a substantially different electricity-demand environment. The International Energy Agency expects global data center electricity consumption to roughly double to about 950 TWh by 2030, with AI among the primary drivers. In the U.S., data centers are projected to account for almost half of electricity-demand growth through the end of the decade.
For utilities, the challenge is not simply generating more electricity. Connecting large data centers requires grid planning, capacity analysis, demand forecasting and increasingly dynamic management of power loads. AI can potentially support those processes by analyzing infrastructure data, forecasting demand and automating operational decisions.
National Grid Partners’ 2026 Utility Innovation Survey, conducted among 134 U.S. utility innovation leaders, illustrates how quickly priorities are changing. Reliability ranked among the top three organizational priorities for 73% of respondents, compared with 43% in 2025. Meanwhile, the proportion placing net-zero goals among their top three priorities fell from 54% to 16%.
The survey also highlights a less visible obstacle: getting AI projects from experimentation into production. Eighty-four percent of respondents said their organizations now take more than a year to move innovation projects from pilot to full deployment. Most innovation spending remains focused on incremental improvements, with 59% allocated to that category compared with 16% for transformational initiatives.
That implementation gap matters as AI infrastructure expands. Gartner projects global data center electricity consumption will reach 565 TWh in 2026, a 26% increase from 2025, with AI-optimized servers accounting for 31% of data center power consumption.
National Grid is also backing technologies designed to make the grid more responsive. Its portfolio company Emerald AI participated in a pilot with National Grid and NVIDIA that tested software capable of reducing data center energy consumption during simulated grid stress while protecting critical workloads.
The company has also joined the AI Energy Management Alliance alongside technology companies including Google and NVIDIA, AI companies such as Anthropic, and several utilities. The consortium is focused on flexible data centers that can automatically reduce electricity consumption during periods of peak grid demand.
National Grid Partners said its latest investments include Terragrit, which develops software for simulating physical operational changes, and LineVision, a grid intelligence platform designed to help utilities increase network capacity and resilience.
The broader shift is significant for enterprise AI infrastructure. As data center operators, cloud providers and utilities compete for limited power capacity, software that makes compute workloads more flexible could become an increasingly important part of the AI infrastructure stack. The next stage of AI scaling may therefore depend not only on GPUs, cloud platforms and data centers, but also on how intelligently those systems interact with the electricity grid.
Market Landscape
AI is turning electricity availability into an increasingly important constraint on data center expansion. The IEA estimates data center electricity consumption reached roughly 485 TWh in 2025 and projects it to approach 950 TWh by 2030. AI-focused data center consumption is growing considerably faster than overall data center demand.
For utilities and hyperscalers, this creates demand for AI energy management, flexible data center infrastructure, predictive grid analytics and automated load management. The competitive landscape extends beyond traditional utility software toward AI platforms, data center operators, semiconductor companies and cloud providers.
The emerging model is increasingly two-sided: AI consumes more electricity, while AI-based optimization can help utilities and data centers use that electricity more intelligently.
Top Insights
- National Grid’s survey shows 78% of utility innovation leaders are operationalizing AI to manage data center interconnection demand as electricity loads rise.
- Grid reliability has become a leading utility priority, reflecting growing pressure from AI data centers and large-scale computing infrastructure.
- Long deployment cycles remain a barrier, with 84% of surveyed utilities taking more than a year to move innovation projects into production.
- Flexible data centers could become an important AI infrastructure layer by dynamically reducing workloads during periods of grid stress.
- Utility-startup collaboration is expanding as energy companies seek AI technologies for grid intelligence, capacity optimization, simulation and operational resilience.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
