Data center load is reshaping how U.S. utilities plan, price, and build generation

Three utility decisions made in the final days of July 2026 show just how fast data center load has moved from a planning footnote to a capital-allocation priority across the U.S. grid. Exelon trimmed 40% out of its speculative project queue. FirstEnergy reported a 50% quarterly surge in data center contracts while committing $2.7 billion to new generation. And Texas regulators handed down what may become the national template for co-locating AI computing loads directly on renewable generation sites. Each outcome carries a distinct set of implications for enterprise operators who procure power, manage facilities, or evaluate infrastructure vendors.

Exelon cuts the noise from its data center pipeline

Exelon's 'high probability' data center load figure now sits at 11 GW, down 40% from a higher earlier estimate, according to Utility Dive. CFO Jeanne Jones attributed the reduction to a deliberate screening process: the company used transmission service agreements to separate committed projects from speculative ones that were simply holding queue positions without realistic build-out plans.

That distinction matters operationally. Utilities that accept every interconnection request at face value risk over-building transmission and generation capacity that never gets used, while underserving load that materializes elsewhere. Exelon's approach effectively turns the interconnection queue into a qualification filter, requiring applicants to execute concrete transmission agreements before they count toward planning targets.

For procurement teams at large enterprises, the implication is direct: utilities are no longer passive recipients of load forecasts. Operators planning new data center deployments in Exelon service territory should expect more rigorous interconnection scrutiny and tighter timelines before capacity is formally allocated.

FirstEnergy's West Virginia bet: $2.7 billion, one customer, and a ratepayer surcharge

FirstEnergy reported a 50% jump in data center contracts in the second quarter of 2026, according to Utility Dive reporter Ethan Howland. The growth is concentrated enough to drive a single, massive capital program: Mon Power, a FirstEnergy subsidiary serving West Virginia, is building $2.7 billion in new generation capacity, with the project oriented primarily around a data center customer.

To finance that build, Mon Power plans to assess a surcharge on its broader customer base. That arrangement is already drawing scrutiny. Spreading the capital cost of infrastructure built for a single hyperscale tenant across all ratepayers is a model that regulators in multiple states are watching closely, and it sets a precedent that could affect how utilities structure large-load service agreements elsewhere.

For enterprise energy managers, this signals something worth tracking: as utilities pursue data center growth aggressively, the capital recovery mechanisms they choose will increasingly affect what other large commercial and industrial customers pay. Operators with facilities in regions seeing heavy data center investment should model the potential rate impact of utility infrastructure surcharges in their energy cost forecasts.

Texas sets the first template for AI co-location at a wind farm

Texas regulators approved an AI data center's co-location directly adjacent to a wind farm, a first-of-its-kind arrangement that Utility Dive reporter Marlene Wilden described as a potential early template for other large behind-the-meter loads. The approval came with explicit conditions: the facility must curtail load rapidly during grid emergencies, and its participation in demand response programs is restricted.

The curtailment requirement is the most operationally significant element. Co-location at a generation site has been pitched as a way to bypass constrained transmission and secure cleaner, lower-cost power. But the Texas order makes clear that grid reliability claims come first: when system conditions tighten, the data center must reduce consumption on a fast timeline, regardless of the cost to the operator.

The restrictions on demand response participation are also notable. Typically, large flexible loads earn revenue by bidding into demand response markets. Limiting that avenue for co-located facilities means the financial case for wind-adjacent data centers in Texas looks somewhat different than early proponents projected. Facilities teams and energy procurement officers evaluating co-location strategies should model both the curtailment frequency risk and the lost demand response revenue when underwriting these projects.

What this means for enterprise operators

Read together, these three developments reflect a grid under real stress from the weight of hyperscale demand, and utilities responding with tools that range from queue management to billion-dollar capital programs to regulatory frameworks that are being written in real time. The Exelon, FirstEnergy, and Texas cases each reveal a different lever utilities will pull as they try to manage load growth without compromising reliability.

For operations and infrastructure teams, the near-term signal is clear. Interconnection is getting harder to secure without firm commitments. Rate structures in data-center-heavy regions may shift in ways that affect all large commercial customers. And co-location at renewable sites, while increasingly approved, comes with curtailment obligations that require robust operational contingency planning.

The next regulatory test is whether the Texas co-location model spreads to other restructured markets, particularly PJM territory, where Exelon and FirstEnergy both operate and where data center load growth is most concentrated. Mon Power's $2.7 billion West Virginia build is scheduled to come online in the coming years; how regulators respond to its surcharge mechanism will be an early indicator of how broadly that cost-recovery model can travel.