US Data Centers Face Massive Power Shortage by 2028

The evidence provided by the sources creates a strong and consistent case that the rapid growth in electricity demand from data centers, driven by AI, is highly likely to cause or contribute to power shortages in the United States by 2028. The assessment is based on a convergence of forecasts and reports from multiple, authoritative sectors including finance, industry, utilities, and academia.A highly specific and impactful piece of evidence comes from a Morgan Stanley forecast, which predicts a 45-gigawatt power deficit in the U.S. by 2028, directly attributing this shortage to the AI data center boom. This provides a quantitative measure for the potential shortfall.This forecast is supported by evidence from other sectors. An Accenture report, cited by energy infrastructure company Schneider Electric, projects that data centers could consume over 7% of total U.S. electricity by 2028. A separate, albeit unnamed, report cited in 'Domestic Preparedness' raises that estimate to 12%. While the numbers vary, they both point to a dramatic and rapid increase in demand that will strain the national grid.Evidence from the utility sector itself corroborates these concerns. A forecast from Dominion Energy, a major utility in the data center hub of Virginia, expects a doubling in its overall electricity demand, driven by data centers. Reports of grid connection delays in the same region further illustrate that a functional shortage, where supply cannot meet new demand in a timely manner, is already occurring.The scale of the problem is highlighted anecdotally by a report in POWER Magazine about plans for new nuclear and natural gas generation to power a single massive data center, demonstrating that existing infrastructure is insufficient for this new class of demand. Finally, citations of the International Energy Agency's (IEA) 'Electricity 2024' report in academic journals confirm that this is a recognized global issue, adding significant weight to the national-level concerns.There are no contradictory sources provided; all evidence points in the same direction. The term 'shortage' is supported both by quantitative deficit forecasts and by qualitative evidence of grid strain and an inability to meet new demand. The 2028 timeframe is explicitly mentioned in several of the key forecasts. Therefore, the statement is assessed as likely true with high confidence.