Powering the ballot: Why AI’s energy footprint is the ultimate midterm election issue

A high-stakes fight is unfolding in the suburbs of Phoenix that should be a wake-up call for the entire technology sector. 

Arizona Public Service, the state's largest utility, has proposed a 45% electricity-rate increase for "extra-large energy users" — primarily data centers — alongside a roughly 14% hike for residential households, about $20 more a month for the typical home.

The message from local communities, consumer advocates, and regulators is increasingly clear: American families, already facing a rising cost of living and triple-digit summer heat, are pushing back on subsidizing the power demands of the artificial intelligence build-out.

From Coast to Coast, the Grid is Under Strain

Phoenix isn't an isolated case. Across the country, the rapid expansion of artificial intelligence infrastructure is colliding with local communities, energy affordability, and grid planning.

Nevada-based utility NV Energy has told Liberty Utilities that it will stop supplying power to the California side of Lake Tahoe after May 2027 — leaving about 49,000 residents needing a new source. The utility's filings cite northern Nevada data centers and transmission constraints among the reasons. An isolated community with little market leverage is now competing for power against hyperscale industrial load near Reno.

Across the PJM Interconnection grid — which stretches from the East Coast into West Virginia — ratepayer advocates are fighting a multi-billion-dollar transmission build-out.

Whether in Arizona, Nevada, or West Virginia, the same question is emerging: Who pays for the infrastructure, and how can households avoid the burden?

The Midterm Battleground: Affordability

Residential electricity rates have moved from a routine expense to a bipartisan political flashpoint. As the midterm elections approach, politicians on both sides are tuning into voter frustration over rising bills.

In Arizona, the APS rate case landed in a midterm year in which two of the five commissioners are seeking reelection, and the state's attorney general has publicly challenged the increase. Candidates are recognizing that the cost of the digital build-out shows up directly on household bills.

The technology sector can't ignore this. In March 2026, the major hyperscalers signed the White House's Ratepayer Protection Pledge, a voluntary commitment that data-center owners — not households — will cover the cost of the power and grid upgrades their facilities require. But the pledge is non-binding and its enforceability is unclear.

That gap between promise and enforcement is precisely why efficiency, not just commitments, will determine whether the industry avoids a regulatory and consumer backlash at the ballot box.

Energy Moves From the Utility Bill to the Balance Sheet

The pressure is also reshaping how businesses themselves account for power. For years, energy was treated as a facilities issue. However, as AI workloads grow, power availability, and costs are drawing the attention of the C-Suite — tracked and managed like every other driver of growth.

In a Dec. 2025 survey of 300 executives by MIT Technology Review Insights, every single respondent expected the ability to measure and strategically manage power consumption to become an important business metric within two years. 

The cost pressure is already here: 68% reported energy-cost increases of 10% or more over the prior year tied to AI and data workloads, and 97% expect their AI-related energy use to keep climbing. More than half named rising costs the single biggest energy-related threat to their AI ambitions.

MIT Technology Review Insights calls the emerging discipline energy intelligence: developing a granular picture of how much power is consumed across operations, when, and to what end, then using that visibility to control costs and optimize operations.

Treating energy intelligence as a C-suite metric turns energy from an uncontrolled liability into a managed input. It is the bridge between the affordability backlash above and the efficiency playbook below.

The Path That Eases the Strain: Efficiency by Design

Much of the technology needed to keep growing AI capacity without overburdening ratepayers already exists. It requires a shift in mindset toward efficiency by design because on a strained grid, the cheapest watt is the one you never have to generate.

Crucially, the largest drivers of data-center power are AI compute and cooling; storage is a smaller slice. So, an honest efficiency strategy has to work across all three pillars:

1. Compute and workloads. Idle and over-provisioned servers draw power for no return. Architectures that scale capacity dynamically and maximize performance-per-watt during training and inference cut the single largest source of waste.

2. Cooling and facility design. Cooling can rival compute as a share of total draw. Modern approaches – liquid cooling, higher-density racks, smarter thermal management – reduce the overhead that traditional air-cooled halls carry.

3. Data storage and lifecycle. Flash-native, high-density storage platforms reduce both footprint and energy versus legacy disk arrays (up to 85%). Service-based models (Storage-as-a-Service) also allow non-disruptive upgrades, cutting the e-waste and capital cost of "rip-and-replace" cycles.

Winning the Efficiency Marathon

By prioritizing density and efficiency across compute, cooling, and storage, enterprises and hyperscalers can grow AI workloads largely within their existing facilities — reducing the need to trigger extra-large-user tariffs or push communities to finance new power plants.

Smart public policy can accelerate the shift. Federal grants, tax incentives, and IT procurement guidelines should be tied to measurable energy-reduction benchmarks, rewarding efficiency rather than raw expansion.

The race for AI leadership is a marathon of efficiency. By championing existing, energy-efficient standards, the industry can build the most advanced digital foundation in the world without the American ratepayer having to foot the bill.

Bill Wright is head of government affairs at Everpure.