Data centers drove half of US electricity demand growth as time-to-power becomes the defining constraint

Data centers now account for roughly 50% of all US electricity demand growth, according to Fortune — a concentration nearly three times the approximately 17% share those facilities represent globally. The divergence underscores how thoroughly the buildout of artificial intelligence infrastructure has reshaped America's power consumption curve in a compressed period.

Time-to-power displaces construction speed as the binding limit

For years, the headline challenge in data center development was how quickly a building could be erected and fitted with compute hardware. That calculus has shifted: Fortune reports that while a modern facility can be brought from groundbreak to operational in under three years, securing utility-grade power for that same facility routinely takes five to more than ten years.

The mismatch has given rise to a new term of art across the sector — time-to-power — now cited by developers, hyperscalers, and colocation providers as the single variable most likely to delay a project. A structure sitting ready but without power is, in operational terms, an idle capital expense.

Grid timelines push operators toward on-site and alternative generation

Faced with utility interconnection queues that stretch well beyond a decade in some markets, operators are reassessing how they source electricity at the point of demand. Three alternatives have moved from experimental to mainstream consideration: on-site natural gas generation, small modular and conventional nuclear agreements, and purpose-built microgrids.

On-site gas offers the fastest deployment pathway and dispatchable capacity, though it carries long-term carbon exposure as sustainability commitments tighten. Nuclear arrangements — both deals with existing plant operators and forward contracts tied to new reactor development — provide the carbon-free baseload profile hyperscalers prize, but introduce their own long lead times.

Microgrids occupy a middle ground, combining distributed generation, storage, and control software to create an islanded power environment that can operate independently of, or in parallel with, the public grid. Their appeal is resilience as much as speed: a campus that can self-sustain removes the utility bottleneck without requiring a decade-long interconnection negotiation.

What this means for the broader power industry

The demand signal emanating from data centers is reshaping investment priorities across the US energy sector. Transmission developers, independent power producers, and equipment manufacturers supplying switchgear, transformers, and backup generation are all recalibrating capacity plans around the sustained appetite for large, reliable power blocks.

Transformer lead times — already stressed by years of underinvestment in domestic manufacturing — have become a downstream symptom of the same constraint. Facilities unable to source critical grid equipment face delays compounding the interconnection backlog, a feedback loop that further extends effective time-to-power for new campuses.

For real estate, finance, and technology professionals evaluating data center assets, Fortune's figures suggest that power procurement strategy has become at least as material to project underwriting as land cost or network proximity. Sites with existing utility capacity or permitted on-site generation are commanding a scarcity premium that would have seemed implausible five years ago.