As AI Booms, Data Centers May Create Electricity Scarcity Among Users
By Ken Silverstein
In the sunbaked outskirts of El Paso, Texas, construction crews are laying the foundations for Meta’s “AI‑optimized” 1.2-million-square-foot data center, designed to house thousands of servers and support large AI workloads. It will draw nearly 1 gigawatt of electricity—enough to power 200,000 homes—and have a water footprint that highlights the paradox of modern tech ambition: utilities must meet this new demand while managing environmental costs.
The Meta project illustrates a looming challenge for the United States and the tech industry: as data centers expand, hyper-scalers will strain energy grids, water resources, and climate commitments, requiring careful planning, oversight, and policies—all to ensure there’s enough power to serve industrial and residential consumers.
“This came out of nowhere. Pre ChatGPT, we weren’t seeing this kind of load growth. It is an entire supply chain issue, involving utilities, industry, the workforce, and engineers, who don’t just fall out of trees because you want them,” says Tom Falcone, president of the Large Public Power Council, during a virtual press event hosted by the United States Energy Association, where I was a panelist.
Across the U.S., AI and cloud computing are rapidly increasing electricity demand. In 2024, data centers used roughly 4% of national electricity—about the same as New York and Chicago combined. By 2030, that share could rise to 9–12%, adding hundreds of terawatt-hours of demand—enough to power tens of millions of homes. The energy footprint of AI is no longer niche; it is becoming a central concern for the power system.
A mix of energy sources is required. Tech companies are signing renewable contracts and experimenting with storage. But AI servers run 24/7, so even the greenest data center needs backup power from natural gas, nuclear, or other firm sources. In regions with clustered AI facilities, local grids face pressure and potential bottlenecks.
AI’s appetite for electricity will compete with increasing needs from industry, homes, and transportation. Utilities may need tens of billions of dollars in new generation and grid upgrades. If fossil fuels supply this power, it could also undermine climate goals. Experts stress coordinated planning with renewables, firm generation, and smart grid management to ensure AI growth does not compromise reliability or affordability.
“I think everybody understands the criticality of getting this done,” says Jim Robb, CEO of the North American Electric Reliability Corporation, at the press event. “But a lot of us are stuck in the way we’ve done things for 20 years, which isn’t going to work for the next 10.”
A City’s Worth Of Power
Even with commitments to renewables, challenges remain: high demand is testing local grids, and excessive electricity use may conflict with renewable targets. Policymakers and utility planners need to balance technological advancements, energy reliability, and environmental goals.
Data centers are ramping up electricity use just as manufacturers, commercial users, and residential demand also rise. This perfect storm could push wholesale prices higher, particularly in regions where the grid is already maxed out.
Batteries and storage are increasingly important, warehousing excess renewable energy during the day and releasing it at night to smooth demand and prevent price spikes. Indeed, Texas has 12,000 megawatts of battery capacity, having added 4,000 MW in 2024. Spencer Hanes, vice president of EnerVenue, says we are “in the middle of a Renaissance”—one that will require a lot more storage, which will provide much-needed power to data centers.
Without thoughtful planning, the increasing demands of AI infrastructure, industry, and households could clash, raising costs and causing tension between economic growth, reliability, and sustainability.
Every new project is like adding a new city to the grid, says Ted Vatnsdal, executive director of the Midcontinent Independent System Operator. “That’s an enormous amount of energy that will be required.”
AI’s potential, though, is enormous. The El Paso facility brings $1.5 billion in investment, 1,800 construction jobs, and 100 permanent positions, positioning the United States as a leader in AI infrastructure, attracting talent and innovation.
Companies are taking steps to reduce environmental impacts, including LEED-certified construction, liquid cooling, and 100% renewable energy contracts. Meta pairs solar and wind with battery storage to smooth fluctuations and reduce reliance on fossil fuels. Even in desert areas of the Southwest, which are criticized for water scarcity, careful planning can conserve water.
Avoiding Price Spikes
Data centers are exploring grid-friendly innovations: demand-response programs, shifting non-essential workloads to periods of excess renewable energy, and colocation with industrial facilities to reuse waste heat and share infrastructure.
Still, the scale of AI growth is unprecedented. Even the most energy-conscious designs cannot fully offset gigawatt-level power demand. Hence, it is a difficult task to balance AI with the needs of manufacturing, commercial, and residential sectors. The energy-smart steps are forward-thinking, but they do not eliminate pressure on grids or the potential for price and reliability tensions.
Shanthi Muthiah, electricity demand specialist for consulting firm ICF International, notes the PJM Interconnection predicts “significant shortfalls” by 2030: “The announcements for new loads are increasing continuously, and I think there could be some challenges in meeting demand.”
It’s a dilemma that will be with us for a while. AI data centers may be the catalyst. But manufacturers, commercial users, and electrified homes all contribute to the puzzle, which will test reliability. Meanwhile, regulators are also under the gun, pressured to control rates or ensure that those who consume the most power pay for it.
“People will start to see what works in Kansas and adopt it elsewhere,” says Charles Bayless, former head of Tucson Electric Power and Illinois Power Company. “But upfront, it’s going to be chaotic—and you are going to have rate shock.”
Meeting this increasing need requires renewables, dispatchable power supplies, and onsite generation with battery storage. And it may also mean bringing some nuclear units out of retirement or postponing the mothballing of others.
Under any set of circumstances, it necessitates thinking outside the box, says David Naylor, CEO, Rayburn Electric Cooperative: “We’ve got to be mindful of our past, but we’ve got to keep moving forward.”
The lesson is clear: AI’s promise and economic benefits are vast, but if we don’t coordinate planning and invest in generation, transmission, and storage, we will fall short. And that means creating conflicts over scarce resources among industrial, commercial, and residential users—a dynamic that may undermine not just reliability and affordability in the electricity sector but also AI’s role in the American economy.
