America's Largest Utility Just Raised Its Capital Plan to $78 Billion—and Data Centers Are Driving Almost All of It

American Electric Power's President and CEO Bill Fehrman expects the total amount of power the company will generate nationwide by 2030 will nearly double what AEP already uses, and that will largely be driven by more data centers in Ohio and Texas , Fehrman said during AEP's earnings call for the first quarter of 2026 on Tuesday .

According to Fehrman, AEP will generate 63 gigawatts nationally by 2030, or enough to power tens of millions of homes, while the company currently generates only 32 gigawatts of power across 11 states . Nearly 90% of this expected incremental contracted load is from data centers, including hyperscalers, and as a result, the utility raised its five-year capital investment plan to $78 billion, up from the prior $72 billion . The announcement comes as data centers accounted for around 50% of all electricity demand growth in the U.S. last year, according to the IEA .

The scale of investment reflects how AI infrastructure is reshaping the entire energy sector. Morgan Stanley Research forecasts U.S. data center demand could reach 74 GW by 2028, with a projected shortfall of about 49 GW in available power access . Large technology companies are likely to commit more than $1 trillion of spending in just the 2025-2026 period , according to Morgan Stanley.

Grid Watchdog Issues Highest-Level Alert After Data Centers Trip Offline

On May 4, 2026, NERC issued a rare Level 3 "Essential Actions" Alert—its highest-urgency notification—in response to a pattern of customer-initiated load reductions in which 1,000+ MW of computational load (data centers) dropped off the bulk power system (tripped offline) in seconds . Level 3 alerts, which are rare, call for grid operators to take specific steps in response to an immediate issue .

In April, NERC said it had analyzed numerous load-loss events across the East Coast and Texas that weren't predicted by planning studies, and these "customer-initiated large load reductions" occurred in seconds, leaving little or no room for real-time responses — and jeopardizing the grid's stability as a result . These were "customer-initiated" because protection circuits at data centers detected problems with grid-supplied power and automatically disconnected to protect their sensitive computing equipment from electrical damage .

The organization plans to register companies with computational loads of at least 20 megawatts — such as Amazon, Google, and Meta — and, eventually, subject them to reliability standards similar to those for power plants . Registered NERC entities must acknowledge receipt by May 11 and must respond by Aug. 3 .

Utilities Threaten to Exit Grid Operators Over Slow Connections

The frustration over grid interconnection delays is boiling over. American Electric Power Co., one of the largest US utilities, is threatening to remove itself from two of the country's biggest electricity grids over the length of time it takes to connect new AI data centers , Bloomberg reported on May 5.

AEP is particularly concerned about difficulties in connecting new power supplies to the grid in PJM, according to Fehrman, who said "The current state of PJM's performance and stakeholder approval process does not give me great confidence that these issues will be resolved anytime soon" . The tension highlights a fundamental mismatch: as AI workloads scale from pilots to production, electricity demand is rising faster than the US power grid – much of it built decades ago – was designed to handle .

Clift Pompee, VP of power and emissions at Compass Datacenters, noted that "2026 is a pivotal year for the future of the U.S. power grid" and that "Most of the grid was built between the 1950s and 1970s, and today, approximately 70% of the grid is approaching the end of its life cycle" .

Smart Grid Tech Offers a Faster Path Than New Construction

While utilities race to build new generation, some experts argue the solution isn't just more copper and concrete. Stanford research reveals advanced economy grids operate at 30% utilization, leaving vast capacity idle due to outdated coordination systems, and a 1% improvement in system flexibility could unlock 100 GW in the US alone, equivalent to $500 billion in avoided infrastructure .

GridCARE's partnership with AI infrastructure developers and Portland General Electric demonstrates the concept at scale—by deploying predictive AI models that forecast renewable output and demand hours in advance, coordinating batteries and backup systems strategically, and dynamically managing loads across the grid, PGE has accelerated hundreds of megawatts of computing capacity years ahead of original timelines, without building new generation or transmission .

The approach represents a counterintuitive strategy: as nations compete to power the AI revolution, the infrastructure bottleneck constraining technological leadership is being solved not through copper and construction, but through code, and flexible grid optimization could double effective capacity faster than any building programme .

Chip Efficiency Breakthroughs Could Cut AI Energy Use by 70%

On the hardware side, researchers are making progress on energy efficiency. A breakthrough in brain-inspired computing could make today's energy-hungry AI systems far more efficient—researchers have engineered a new nanoelectronic device using a modified form of hafnium oxide that mimics how neurons process and store information at the same time, and unlike conventional chips that waste energy moving data back and forth, this device operates with ultra-low power—potentially slashing energy use by up to 70% , according to research published by the University of Cambridge in April.

With AI data center workloads forecast to triple or quadruple annually between 2026 and 2030, chip- and system-level integration will be required to enable system performance in hyperscale data centers, and chiplets are addressing chip-level performance needs in AI data centers, delivering yield, bandwidth, and energy-efficiency benefits , according to Deloitte's 2026 semiconductor outlook.

Normal Computing has raised $50 million in a round led by Samsung Catalyst as the startup pursues a two-pronged bet on the future of AI hardware: using AI to help semiconductor companies design chips more efficiently, while also developing a new kind of processor aimed at reducing energy use, and CEO Faris Sbahi told Fortune the company's software platform is already being used by more than half of the top 10 semiconductor companies by revenue .

The Affordability Question Looms Over the Buildout

As utilities pour billions into grid expansion, ratepayers are starting to feel the impact. The U.S. Energy Information Administration reported in February, this year's average electric industry revenue in Ohio increased by 22% compared to a year ago . Electric and gas utilities requested more than $30 billion in rate increases last year, according to a January analysis by PowerLines, a consultancy, affecting 81 million Americans, and overall, power bills have risen 40% from 2021 .

Data centers' ravenous energy needs have nonetheless received the brunt of the blame, with polling suggesting most households connect data center expansion with rising electricity costs . Local opposition blocked or delayed at least 16 data centers last year, worth a combined total of $64 billion , Fortune reported.

According to market data, WTI crude traded at $71.50 per barrel on Sunday, up 0.6%, while Brent crude stood at $75.20 per barrel, up 0.5%. Natural gas prices at Henry Hub fell 2.4% to $3.25 per million BTU. The energy sector faces a complex moment: AI is driving unprecedented electricity demand growth while geopolitical tensions—including the ongoing Strait of Hormuz closure—continue to pressure fossil fuel markets.

The collision between AI's insatiable appetite for power and an aging grid built for a different era is forcing utilities, regulators, and tech companies to rethink decades of assumptions about how electricity infrastructure works. Whether through massive capital investment, smarter grid coordination, or breakthrough chip efficiency, 2026 is shaping up as the year the energy industry either adapts to AI—or becomes its bottleneck.