Are Utility Aspirations for Data Centers Realistic?

The electric utility industry is riding a wave of euphoria over data center demand. Load growth projections that would have seemed delusional five years ago are now being presented to Wall Street with straight faces. But physics, policy, and politics don't care about investor narratives — and all three are starting to push back hard.

The Gap Between Ambition and Reality

For the better part of a decade, utilities watched their load growth flatline. Energy efficiency improvements, industrial offshoring, and stagnant population growth in many service territories kept demand stubbornly flat. Then came the AI boom, hyperscaler expansion, and a sudden explosion in data center construction — and utility executives who had been defending low-growth assumptions to regulators suddenly found themselves projecting 15%, 20%, even 30% load increases over the next decade.

The numbers are staggering on paper: some regional grids are fielding interconnection requests that would double their current peak demand. PJM Interconnection, which serves roughly 65 million people across 13 states, has seen its interconnection queue balloon to over 290 GW of requested capacity — a figure that dwarfs what's actually been built. That's not a pipeline. That's a wish list.

The appeal is obvious. Data centers represent anchor load — large, predictable, credit-worthy customers that run 24/7. For utilities accustomed to fighting over marginal residential kWh, landing a 100 MW hyperscaler campus feels like striking oil. The problem is that striking oil and actually getting it to market are two very different things.

The Constraints Nobody Wants to Talk About in Earnings Calls

Physics First

Transmission infrastructure is not a software problem. You can't patch a substation with a firmware update. When a data center developer drops 500 MW of demand into a region built to serve suburban housing and a few light industrial parks, the grid doesn't magically accommodate it.

Transformer procurement timelines are running 2-4 years in many cases. High-voltage transmission lines can take a decade to permit, site, and build. The fundamental bottleneck isn't capital or even political will — it's the sheer physical complexity of moving electrons at scale through aging infrastructure that was never designed for this kind of concentrated, always-on load.

Interconnection queues reflect this reality. Most of those 290 GW of requests in PJM's queue will never get built — not because demand disappears, but because the grid can't physically absorb them fast enough. Developers who jumped into Tier 1 markets like Northern Virginia (which already hosts the world's largest concentration of data center capacity) are now quietly scouting secondary markets precisely because the grid has hit a wall. Loudoun County, Virginia — "Data Center Alley" — has effectively reached saturation from an interconnection standpoint.

Policy as a Wildcard

Regulatory frameworks governing electric utilities were not designed with 500 MW single-customer loads in mind. Rate cases, cost-allocation rules, and interconnection standards were built for a world of distributed, moderate demand. That mismatch is creating friction at every level.

The question of who pays for grid upgrades is particularly contentious. When a hyperscaler's data center campus requires $400 million in transmission upgrades, regulators face an uncomfortable choice: socialize those costs across all ratepayers or require the data center developer to bear them directly. Neither answer is politically clean. Residential ratepayers don't love subsidizing Amazon's infrastructure. But requiring developers to self-fund upgrades changes the economics of projects dramatically and can kill deals outright.

Some states are beginning to respond with data-center-specific utility frameworks, but legislation moves slowly. In the meantime, utilities are navigating a regulatory environment that was essentially designed for a different century.

The Political Dimension — Which Way Is the Wind Blowing?

Politics adds another layer of complexity that doesn't show up in load forecasting models. Data centers are simultaneously beloved and resented.

Beloved because they bring tax revenue, construction jobs, and the perception of tech-sector credibility to communities. Resented because they consume enormous amounts of power and water while employing relatively few permanent workers. A 300 MW data center campus might employ 30-50 full-time staff. That math doesn't sit well with local officials who are also fielding complaints from manufacturers or agricultural users about grid congestion and rising electricity rates.

The political calculus is shifting in ways that utility executives are only beginning to account for. Several states are exploring or have already implemented moratoriums, impact fees, or enhanced review processes for large data center interconnection requests. Virginia — the epicenter of data center development — passed legislation in 2023 requiring the State Corporation Commission to scrutinize data center rate structures more carefully. That's a significant signal from the state that essentially built the modern data center industry.

At the federal level, FERC (the Federal Energy Regulatory Commission) has been grappling with how to reform interconnection processes that have become functionally broken under the weight of clean energy and data center requests. New queue reform rules are in effect, but their real-world impact on timelines won't be clear for years.

Meanwhile, the intersection of data center growth with clean energy mandates creates its own political tension. Utilities committed to carbon reduction goals are now being asked to build or contract for significant amounts of new gas generation to serve data center load that can't wait for renewables-plus-storage to mature at scale. That contradiction is not lost on state legislators, environmental advocates, or the utilities themselves.

What Comes Next — And What Actually Works

None of this means the data center buildout stops. Demand is real, capital is available, and the underlying drivers — AI compute, cloud migration, digital infrastructure — aren't going away. But the shape of the growth will look different than the most optimistic projections suggest.

Geographic diversification is already happening. Developers who once insisted on Northern Virginia or Silicon Valley are seriously evaluating markets like the Midwest, Southeast, and Texas — places with available land, lower power costs, and grid capacity that hasn't been fully spoken for. This is good for those markets and arguably healthier for the grid overall, but it requires developers to accept longer construction timelines as local infrastructure catches up.

On-site generation — natural gas, nuclear microreactors, and in some cases large-scale fuel cells — is moving from a niche conversation to a serious operational consideration. Companies like Microsoft and Google are actively exploring small modular reactors (SMRs) precisely because the utility grid, in many markets, cannot realistically deliver the scale and reliability they need. If that trend accelerates, it fundamentally changes the relationship between hyperscalers and electric utilities — from dependence to partial bypass.

For utilities, the smarter play may be focusing less on chasing every data center megawatt and more on building the transmission and distribution infrastructure that allows them to serve diverse load types over the long term. Utilities that overextend capital commitments to serve specific data center customers — only to see those customers co-locate generation or relocate to better markets — face real stranded-cost exposure.

The insider reality is this: utility load forecasts tend to get revised down, not up, over time. The track record of large-scale demand projections in this industry is not encouraging. Utilities that build rate cases around optimistic data center assumptions and then fail to capture that load face regulatory consequences and investor skepticism. The better long-term position is disciplined infrastructure investment with flexible interconnection structures — not a land grab for every MW of data center demand that crosses the desk.

The electric utility industry's aspirations for data center-driven growth aren't wrong. They're just running faster than the grid, the regulators, and the political environment can accommodate. The developers, utilities, and policymakers who recognize that gap — and build strategies around it rather than pretending it doesn't exist — are the ones who will still be standing when the dust settles.

[INTERNAL LINK: data center demand]

[INTERNAL LINK: regulatory frameworks]

[INTERNAL LINK: clean energy mandates]

EDITOR NOTES:

• Consider cutting the paragraph discussing the appeal of data centers as it may feel redundant after the previous sections.
• Ensure the internal links are relevant and lead to appropriate content on the blog.