The five-year wait for grid connections has become unacceptable. So tech companies are building their own power plants, creating what amounts to a shadow electrical grid across the United States. The implications for energy policy, carbon emissions, and your electricity bill are enormous.
The Scale of the Problem
A single AI data center can consume as much electricity as 1,000 Walmart stores. An AI-powered search uses 10 times the energy of a standard Google query. Multiply that across billions of daily interactions, and you have an energy crisis that the existing grid was never designed to handle.
The U.S. power grid was largely built between the 1950s and 1970s. Today, approximately 70% of it is approaching end-of-life. Meanwhile, cumulative planned data center capacity in the U.S. totals roughly 296 gigawatts, with more than 70 projects exceeding 1 gigawatt of peak demand each.
PJM, the regional transmission organization serving much of the eastern United States, projects the grid will be six gigawatts short of its reliability requirements by 2027.
The Corporate Response
Rather than wait, tech companies are going around the utilities entirely.
Meta’s Socrates project in New Albany, Ohio will run on a pair of off-grid natural gas power plants, scheduled to go live later this year. Equinix already operates 19 U.S. facilities that rely on on-site power generation to some degree.
OpenAI and Oracle’s $500 billion Stargate facility in West Texas is deploying its own natural gas turbines. Elon Musk’s xAI is powering its Memphis data centers the same way.
Natural gas leads as the preferred energy source because it can be deployed in 18-24 months compared to five or more years for grid interconnection approval. The nonprofit Global Energy Monitor found that proposals for new natural gas facilities in the U.S. tripled in 2025 compared to the previous year, driven largely by data center demand.
The Cost to Everyone Else
This buildout isn’t free, even for people who never interact with AI. Transmission companies are spending billions to accommodate data center demand, and those costs filter down to residential customers.
The Energy Information Administration forecasts residential electricity prices will rise another 4% nationwide in 2026, following a 5% increase in 2025. Data center power demand is a significant driver of this trend.
The political backlash is already materializing. Both Bernie Sanders and Ron DeSantis have spoken out against the data center boom, a rare point of bipartisan agreement that suggests regulatory intervention may be coming.
Who Wins, Who Loses
The winners are obvious: natural gas suppliers, turbine manufacturers, and the tech companies wealthy enough to build their own power infrastructure. Nvidia and the other chip suppliers continue to benefit from insatiable AI demand.
The losers include everyone paying higher electricity bills, communities living near new natural gas plants, and the climate targets that these carbon-intensive facilities make harder to meet. The tech companies’ net-zero pledges look increasingly hollow when their AI ambitions require burning fossil fuels at scale.
Data center operators are betting that AI revenue will eventually justify the massive upfront capital costs. But if AI demand slows or regulation tightens, these companies could be stuck with expensive stranded assets, while residential ratepayers are left with the infrastructure costs.