Bill Gates delivered a blunt message to the tech giants racing to build ever-larger data centers for artificial intelligence. Communities won’t tolerate projects that drive up electricity rates for ordinary homes. The Microsoft co-founder and Breakthrough Energy founder spoke as opposition mounts across the U.S. and power demand projections climb to unprecedented levels.
His warning lands at a moment when data centers already strain grids from Virginia to Texas. Residential bills have jumped. Some homeowners saw monthly charges rise from around $100 to $281 in a single month. “It’s just so far beyond any bill that I’ve ever had,” one Virginia resident told Consumer Reports.
The Scale of AI’s Electricity Appetite
Data center electricity consumption in the United States hit records in 2025. The Energy Information Administration expects power use to set new highs again in 2026 and 2027. AI-driven facilities account for much of the surge. A single large AI training cluster can consume as much electricity as 100,000 households. The biggest ones under construction draw 20 times that amount.
Global data center demand more than doubled projections in recent forecasts. The International Energy Agency sees data centers reaching 945 terawatt-hours by 2030. That’s roughly Japan’s entire annual consumption. In the U.S., data centers could represent nearly half of all electricity demand growth through the decade’s end. Goldman Sachs forecasts U.S. data center power demand will more than double to 66 gigawatts by 2027 from 31 gigawatts in 2025. Reuters reported the EIA’s latest outlook just days ago.
Yet the costs don’t stay isolated inside server farms. Utilities often spread infrastructure upgrades across all ratepayers. Wholesale electricity prices near data center clusters have soared 267 percent in five years. PJM Interconnection, which serves 13 states and more than 67 million people, saw capacity prices explode from $30 to $270 per megawatt-day. That ninefold jump tied directly to data center demand, according to its independent market monitor. Bills for some households climbed 7.1 percent in 2025, outpacing inflation. In certain states the increase topped 20 percent.
John Steinbach lives near Manassas, Virginia. His experience mirrors complaints heard from Ohio to Pennsylvania. Data centers cluster in these regions because of land, fiber and existing power lines. But the concentrated load overwhelms local infrastructure. “They’re building them like it’s ‘Field of Dreams’—build it and the electricity will come—but we don’t see how that’s going to happen,” Steinbach said.
Gates takes a different view on the long-term outlook. He acknowledges data centers could lift global electricity consumption by 2 to 6 percent. The real test, he argues, lies in whether AI can deliver efficiency gains and new clean energy technologies that more than offset its own appetite. “The question is will AI accelerate a more than 6 percent reduction? And the answer is: certainly,” Gates said at a Breakthrough Energy summit, as reported by Forbes.
His comments carry weight. Gates backed TerraPower, a nuclear startup now partnering with data center operators to deploy advanced reactors. Small modular reactors and next-generation nuclear designs feature prominently in industry plans because they offer always-on, carbon-free power without the intermittency of wind or solar. Microsoft, Google, Amazon and others have signed massive power purchase agreements. Some pledge to cover their full energy costs and avoid seeking tax breaks that shift burdens to taxpayers.
But promises haven’t quieted local anger. In early 2026, Microsoft outlined steps to pay its own electricity bills, cut water consumption, create jobs and stop pursuing special tax treatment. Similar commitments came from Anthropic and others. President Donald Trump hosted tech executives at the White House and extracted pledges to build dedicated generation and sell excess power back to the grid. AP News covered the March meeting.
Still, the numbers keep growing. Bloom Energy predicts U.S. data centers’ combined demand will nearly double from 80 gigawatts to 150 gigawatts between 2025 and 2028. That’s the equivalent of adding Spain’s entire electricity needs in three years. Deloitte estimates AI data center power demand could surge 30-fold by 2035, reaching 123 gigawatts. Business Insider’s analysis of permits filed through 2025 shows potential annual consumption between 224 and 359 terawatt-hours, a 50 percent jump over the prior year.
So far, many new facilities rely on natural gas plants approved in Texas and Pennsylvania. Utilities extend the life of coal units in some regions. Microsoft’s own North American data centers still pull at least half their power from coal and gas grids despite heavy renewable investments, according to a Stand.earth report. The pattern raises questions about how fast the industry can decarbonize while scaling at this pace.
Gates doesn’t downplay the short-term friction. He notes a coming frenzy. “There will be a frenzy. Some companies will burn billions on data centers whose electricity is too expensive, and won’t earn it back before the next generation of chips arrives,” he observed recently. The remark, shared widely on X, underscores the risk that capital expenditure races ahead of viable power solutions.
Opposition has slowed projects. Between 30 and 50 percent of planned 2026 data center builds face delays or cancellation due to power shortages, equipment backlogs and supply chain issues. Northern Virginia, already home to the world’s largest concentration of data centers, faces record demand amid constrained supply. States like Texas passed legislation to reform interconnection, cost allocation and grid planning.
The tension pits two powerful forces against each other. On one side sits the promise of AI transforming industries, accelerating scientific discovery and potentially optimizing energy systems themselves. On the other lies the immediate reality of higher bills, grid congestion and public pushback. Households shouldn’t subsidize Big Tech’s expansion, Gates insists. Tech companies possess the balance sheets to fund their own generation, negotiate direct power deals and site facilities where abundant clean energy already exists.
Whether they follow through will shape electricity markets for the next decade. Early signs show movement. Some hyperscalers explore co-locating with nuclear plants or building private grids. Others invest in geothermal, advanced batteries and hydrogen. The pace of these alternatives will determine if AI’s power hunger becomes a catalyst for energy innovation or a source of prolonged consumer pain.
One thing seems clear. The days of quietly absorbing massive new loads onto the existing grid have ended. Bill Gates has put the industry on notice. Pay your own way. Or face growing resistance that could stall the very AI ambitions driving the boom.
WebProNews is an iEntry Publication