For decades, America's electric grid quietly did its job. Lights came on. Factories hummed. Rates stayed manageable. Nobody thought much about it. Now, almost overnight, that quiet confidence has evaporated. NCEA’s new report, penned by Dr. Jonathan Lesser and Patrick J. McCormick III, reveals just how unprepared the nation’s electricity infrastructure is for what’s coming.

The Demand Is Real. The Infrastructure Isn't Ready.

After two decades of near-flat electricity demand — growing at less than 0.6% per year — America is suddenly staring down a surge unlike anything since the mid-20th century. AI data centers are the biggest driver, but reshoring manufacturing, electric vehicles, and electrified heating are all pulling in the same direction.

The problem is that the grid was optimized for a world that no longer exists. Years of policy focused on managing and reducing demand has left utilities, transmission operators, and regulators poorly positioned for today’s landscape. You can't build a grid for surging demand when the entire regulatory framework was designed to slow consumption.

A System Built for Another Era

The U.S. electric system is the product of a century of layered decisions — monopoly franchises, cost-of-service regulation, regional power pools, federal restructuring mandates, state renewable energy goals. Each layer made sense at the time. Together, they've created a system that is structurally slow to adapt.

Regional Transmission Organizations (RTOs), the entities responsible for operating the grid across most of the country, have authority over wholesale markets and transmission, but they cannot order new generation to be built, and they cannot stop existing plants from retiring. When demand jumps faster than the market can respond, the RTOs can watch the gap widen but lack the tools to close it.

Meanwhile, the capacity markets designed to signal the need for new generation have increasingly been shaped by politics rather than economic logic. Intermittent wind and solar receive capacity credits that don't reflect their real availability during peak demand. The result has been aa tenfold spike in capacity market prices in PJM — the nation's largest RTO — that signals genuine scarcity without producing the new dispatchable generation the grid actually needs.

The Math Problem Nobody Wants to Solve

Electricity demand and supply must be balanced every second of every day. There is no warehouse where you store Monday's extra electrons for use on Tuesday. This physical reality means adequate, dispatchable generation, i.e. power that can be called upon on demand, not just when the wind blows, is non-negotiable.

Existing reliability standards, NERC concluded in a March 2026 white paper, "are inadequate for the reliable integration of emerging large loads." That's a regulatory body acknowledging, in plain language, that the rulebook has failed to keep up with reality. Data centers are being built and planned at a pace that outstrips the infrastructure being built to serve them.

The Private Grid Option

Faced with regulatory gridlock, some of the largest electricity consumers in the world are simply deciding to opt out of the public system.

In March 2026, seven major tech companies, Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI, signed the Ratepayer Protection Pledge and committed to build, buy, or bring their own generation rather than rely on the public grid for their new power needs. President Trump endorsed it wholeheartedly: "They can build their own plant, they're going to produce their own electricity."

A fully private grid, generating plants, battery backup, advanced controls, all physically separate from the public utility system, removes large industrial loads from cost pools shared with residential ratepayers, sidesteps slow interconnection queues, and can be deployed faster than a utility-scale project navigating a decade of permitting. Between 20% and 50% of planned new data center capacity is expected to pursue some form of this approach.

The Tradeoffs Are Real

Private grids aren't without costs, however. A data center operator running its own generation is, in effect, running its own vertically integrated utility; a function most tech companies are not built to perform. Reliability would require either expensive battery storage or redundant dispatchable capacity, since there will be no larger interconnected system to draw from as backup. Natural gas is the practical fuel of choice at data-center scale but requires pipeline infrastructure that some states have actively resisted expanding.

The Political Temptation to Slow Things Down

In many places, the response to rising electricity demand has led to blaming data centers for higher rates and a push for moratoriums on new construction. This misreads the evidence. Rate increases have many causes, from grid modernization to renewable mandates to aging infrastructure. More fundamentally, restricting data-center construction doesn't eliminate the underlying demand for AI and cloud computing. It relocates it. Europe has already demonstrated this dynamic. Attempting to solve a supply problem by attacking demand doesn't fix the grid. It just moves the jobs and the investment elsewhere.

The Path Forward

America has the industrial capacity, the engineering talent, and the economic incentive to build the grid its economy requires. What's been missing is the regulatory will to reform the approval processes for new generation, new transmission, and new interconnections, and respond swiftly to this opportunity.

Whether the answer is expanding public grid investment, private generation by large loads, or hybrid models that blend both, one conclusion from this report is clear: business as usual isn't a viable option. The grid is at a crossroads, and the choices made in the next few years will determine whether America's electricity infrastructure becomes an economic asset or a constraint in the decades ahead.

‍