JPMorgan Says 2028 Could Be the Year Data-Center Build-Outs Hit Political Friction

JPMorgan said on November 5, 2026 that 2028 will be a problem for data-center build-outs, pointing to strong showings by Democratic gubernatorial candidates in the midterms. The remarks came from MarketWatch's report on the bank's analysts, and they followed a narrower warning from JPMorgan's Michael Cembalest the day before.

That earlier warning, reported on November 4, was more specific. According to Institutional Investor's interview with Cembalest, ballot measures curbing data centers won in the midterms, and newly elected governors are likely to impose restrictions on them.

The distinction matters. A ballot measure is a voter decision that can take legal force immediately in some cases. A gubernatorial restriction depends on what a governor can get through a legislature, a permitting agency, or a utility commission, and that process takes months or years. The timeline Cembalest and his colleagues draw points toward 2028 as the year the friction shows up in actual construction schedules.

The physical constraint sits underneath the politics. In an October 21 Reuters column, U.S. data center developers were said to face a 30% to 40% power shortfall. The same piece cited Cembalest on AI-related stocks accounting for 90% of capital spending growth, meaning the growth in money companies spend on buildings, chips, and equipment.

When nearly all of that incremental spending comes from a narrow group of companies, a bottleneck does not spread evenly across the economy. It lands on a handful of balance sheets, their suppliers, and the utilities that serve them.

The scale is large. In a January 2 CNBC piece, Cembalest pointed to expectations of roughly $2.5 trillion in cumulative data-center capital expenditure. The U.S. Energy Department has said data centers could consume up to 12% of total U.S. electricity by 2028, up from 4.4% in 2023, a projection first reported by CNBC in June 2025. Together, those figures describe a build-out that needs a great deal of new generation and transmission on a short clock. That is where a 30% to 40% shortfall and new state-level restrictions would collide.

Cembalest had already framed power as a central question for the year. His Eye on the Market Outlook 2026, titled Smothering Heights, identified four risks for 2026, including constraints on U.S. power generation. JPMorgan Asset Management published the same report as a PDF. His November 2026 Eye on the Market podcast, Now You Can See Everything, is the latest installment from the same team.

For ordinary households, the chain runs through electricity bills and local politics more than through stock prices. Who pays for new generation, how fast interconnection queues (the waiting lines of projects seeking grid connections) move, and whether a state accepts new load at all are decisions made by regulators and legislatures. The midterm results change who holds those levers. For investors, the open question is whether the 2028 timing is a forecast the market has already priced or one it has not. A projection three years out from a bank's research desk deserves the same skepticism as any long-range call.

In my view, the debate has shifted from whether data-center demand will be constrained to when and where. Think of the grid as a pipe: the demand can be enormous, but a narrow pipe backs water up. The $2.5 trillion figure assumes power arrives on schedule. The 12% share assumes host states keep letting facilities connect. Both assumptions now carry more political risk than they did before the midterms, and because 90% of spending growth sits with a small group, that risk is not diversified across the market. Those are the variables to watch, along with the specific measures and appointments that follow from the November results.