In collaboration withEverpure
Loudoun County, Virginia, once celebrated for its rural landscapes and close proximity to Washington, DC, has in recent years developed a contemporary image: The region boasts the world’s highest density of data centers.
A decade prior, these facilities were primarily responsible for powering emails and online shopping. Presently, with the explosive growth in demand for AI-driven solutions, local utility Dominion Energy is striving to keep up with escalating energy requirements. The urgency is so significant that Dulles International Airport is building the largest solar project at an airport in the nation, a prominent initiative aimed at enhancing the region’s energy portfolio.
Data center parks such as Loudoun’s are emerging nationwide to satisfy an unquenchable demand for AI technology. However, this expansion comes with a substantial price tag. In the United States, data centers accounted for approximately 4% of national electricity in 2024. Predictions indicate this could increase to 12% by 2028. To illustrate, a solitary 100-megawatt data center consumes about the same amount of electricity as 80,000 U.S. households. Current data centers are being designed for gigawatt capacity, sufficient to supply energy for a mid-sized city.
For corporate leaders, energy expenses tied to AI and data systems are rapidly becoming a financial concern and a potential growth hindrance. Responding to this demand requires a capability that many organizations are just starting to cultivate: energy intelligence. This emerging field involves comprehending where, when, and why energy is used, and leveraging that knowledge to enhance operations and manage expenses.
These initiatives aim to tackle both immediate financial challenges and longer-term reputational threats, as communities like Loudoun County grow increasingly wary of the energy requirements associated with local data center growth.
In December 2025, MIT Technology Review Insights surveyed 300 executives to gauge how organizations perceive energy intelligence today and which challenges they expect to face going forward.
Here are five key insights from our findings:
- Energy intelligence is increasingly recognized as a critical business objective. All executives surveyed believe that the capacity to measure and manage energy use strategically will become an essential business metric within the next two years.
- AI-related workloads are driving noticeable cost increments, and this trend is just beginning. Two-thirds of executives (68%) indicate their organizations have encountered energy cost hikes of 10% or more in the last year due to AI and data-driven tasks. Nearly all respondents (97%) foresee an uptick in their organizations’ AI-related energy usage over the upcoming 12-18 months.
- Rising costs represent the primary energy-related threat to AI innovation. Half of executives (51%) view escalating expenses as the foremost energy-related risk to their digital and AI projects. The majority of firms currently monitoring and striving to optimize data center energy use are driven by cost control motives.
- Organizations are taking action through infrastructure enhancement and energy-efficient alliances. To manage increasing energy needs, three out of four leaders (74%) are refining existing infrastructure, while 69% are collaborating with energy-efficient cloud and storage solutions. Over half are also adopting AI workload scheduling (61%) and investing in hardware upgrades (56%).
- Bridging the measurement gap is the next challenge. Most companies still lack the detailed data necessary for true energy intelligence. This issue is particularly significant for organizations that depend on third-party cloud providers and managed services for their computing and storage, where 71% attribute rising consumption-based costs, yet energy metrics are frequently unclear.
This material was generated by Insights, the custom content division of MIT Technology Review. It was not authored by the editorial team at MIT Technology Review. It was researched, crafted, and written by a team of writers, editors, analysts, and illustrators. This includes designing surveys and gathering data for these surveys. AI tools that may have been utilized were restricted to secondary production processes that underwent extensive human oversight.
