Smarter Energy Management as a Driver of Sustainable Business Growth
Loudoun County, Virginia didn't set out to become the world's data center capital — it just happened to sit at the right intersection of fiber infrastructure, available land, and proximity to the US capital. Now, the region's local utility is scrambling to keep the lights on as AI-driven demand rewrites every assumption about how much power a modern economy actually needs.
The electricity math behind AI's growth
The numbers are stark. US data centers consumed around 4% of national electricity in 2024. By 2028, that share could hit 12% — a tripling of demand within a single presidential term. A 100-megawatt facility, which would have seemed enormous a decade ago, now draws roughly the same power as 80,000 American homes. And the industry is already designing campuses at gigawatt scale, enough to rival the consumption of a mid-sized city.
Dulles International Airport's response to this pressure is telling: it's building the largest airport solar installation in the country, not as a sustainability gesture, but as a practical attempt to shore up a regional grid that's visibly straining. When airports start doubling as power infrastructure, the scale of the problem becomes hard to ignore.
What 300 executives are actually worried about
In December 2025, MIT Technology Review Insights surveyed 300 executives to gauge how organizations are thinking about energy consumption tied to AI and data infrastructure. The findings paint a picture of an industry that sees the problem clearly but is still working out how to act on it.
Every single respondent — 100% — expects the ability to measure and manage power consumption to become an important business metric within two years. That kind of unanimity is rare in executive surveys and signals a genuine shift in how energy is being framed: less as an operational footnote, more as a strategic variable.
The cost pressure is already real. 68% of executives reported energy cost increases of 10% or more over the past 12 months, directly attributable to AI and data workloads. Nearly all respondents (97%) expect those costs to keep climbing over the next 12 to 18 months. Half of those surveyed ranked rising costs as the single greatest energy-related risk to their digital and AI initiatives — ahead of reliability concerns or regulatory exposure.
On the response side, organizations are moving on multiple fronts simultaneously. Three in four leaders (74%) are optimizing existing infrastructure. 69% are partnering with energy-efficient cloud and storage providers. More than half are scheduling AI workloads more strategically (61%) and investing in more efficient hardware (56%).
Why visibility into consumption is the real bottleneck
The survey surfaces a problem that goes beyond cost: most enterprises simply don't have the data they need to manage energy intelligently. The gap is sharpest for companies relying on third-party cloud providers and managed services — which account for where 71% of rising consumption-based costs originate — yet energy metrics from those providers are frequently opaque or aggregated to the point of uselessness.
This is what the report frames as "energy intelligence" — not just reducing consumption, but understanding where, when, and why it's happening. Without that granularity, optimization efforts are essentially guesswork. You can't manage what you can't measure, and right now, a significant portion of enterprise AI infrastructure sits in a measurement blind spot.
The communities hosting these facilities are also paying attention. Loudoun County residents have grown increasingly vocal about the environmental and infrastructural burden of data center expansion. That community friction adds a reputational dimension to what might otherwise look like a pure cost story — and it's one more reason energy intelligence is moving from a nice-to-have to a board-level concern.
The full findings, including detailed breakdowns by industry and company size, are available in the complete report. Download it here.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review's editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.
