The Grid Doesn't Care About Politics — Only Physics

Global energy demand is accelerating, and electricity is at the center of it. Industrial expansion, the electrification of transport, and — most dramatically — the explosion of data centers powering AI have created a step-change in load growth that many planners still treat like a blip. This isn’t tomorrow’s problem — it’s already reshaping where we build, how we price, and who pays.

Start with the data: the International Energy Agency projects electricity use by data centers will more than double by 2030 — rising from roughly 460 TWh in 2024 to ~1,000 TWh by decade’s end — and that’s before the next wave of AI inference becomes table stakes across every sector. In the United States, the era of flat load is over; EIA now forecasts just over 2% annual power-demand growth in 2025–2026, with the sharpest increases where AI campuses and new manufacturing are concentrating — Texas and Northern Virginia. In parallel, retail prices and sales are rising across sectors, with national average revenues up around 5% year over year as of mid-2025. Translation—demand is up, prices are up, and the curve is getting steeper, not flatter.

Yet we’re retiring dispatchable capacity faster than we’re replacing it. Coal exits are accelerating, and nuclear — our largest source of zero-carbon, 24/7 power — has been essentially stagnant for decades. The important (but singular) exception is Vogtle Units 3 and 4, which came online in 2023–2024. One plant — welcome as it is — doesn’t backstop a national surge in 24/7 demand.

Meanwhile, demand from AI and digital infrastructure is not abstract; it’s local, lumpy, and grid disruptive. In 2024, a single protection-equipment failure triggered scores of data centers to auto-disconnect in Northern Virginia — suddenly removing gigawatts of load and forcing operators to ramp generation down to avoid wider instability. “Data Center Alley” now relies on massive fleets of diesel backup because the grid and transmission build-out aren’t keeping pace. Texas and mid-Atlantic operators have warned that fast additions of very large loads are tightening reliability margins even as extreme weather pushes peaks higher.

Transmission is the quiet giant in this story. The Department of Energy’s needs studies are blunt: we face a pressing requirement for both within-region and interregional lines this decade and beyond. Added transfer capacity between regions delivers the largest benefits; in plain English, we need far more high-voltage highways to move power from where it’s generated to where it’s consumed. For perspective, between 2011 and 2020 the U.S. energized only a few thousand transmission circuit-miles per year—mostly one-off reliability fixes, not the backbone build we now require. And new high-voltage lines routinely take 7–10 years from concept to energization — longer for multi-state projects — so every year we delay is another year we accept congestion, curtailment, and higher delivered costs.

Who pays? On average, U.S. residential customers pay roughly 17–18¢/kWh, compared with ~13–14¢/kWh for commercial and ~9¢/kWh for industrial users (mid-2025 national averages were ~17.5¢, ~13.6¢, and ~8.9¢, respectively). In Virginia, residential hovered around the mid-15¢ range vs ~9–10¢ for commercial; in Texas, ~15¢ vs ~8½¢. When utilities invest in generation, transmission, and distribution, those costs are typically rate-based and recovered across customer classes according to regulatory allocation rules — so households do feel the build-out in their monthly bills. Said plainly — the residential class often sees increases sooner and more visibly, even when the new load is primarily commercial.

Data-center growth is now visible in rates. Rapid additions of hyperscale load can force upgrades to substations, feeders, and transmission that are recovered in rates. Utilities in fast-growth regions are proposing special tariffs for large, 24/7 digital loads amid concerns that discounts could shift costs onto households and small businesses.

Permitting is the rate-limiter. High-voltage transmission lines can take 7–10 years to plan and build, and the critical path is routing, environmental reviews, land acquisition, and multi-agency approvals — plus litigation. Even with federal streamlining, most big lines still span many budget cycles, delaying congestion relief and keeping delivered-power costs higher than they need to be. Until cost-allocation frameworks catch up — and even after, for many local upgrades — residential and small business customers will keep shouldering a meaningful share unless policy directs large, 24/7 loads to fund dedicated facilities and network reinforcements.

Transportation electrification was supposed to be the easy part. Instead, consumer acceptance has cooled just as mandates got complicated. In the U.S., states are revisiting or exiting EV rules and reverting to federal standards, reflecting infrastructure and affordability concerns. Abroad, some governments have delayed internal-combustion phase-outs — policy whiplash that underscores the gap between aspiration and grid readiness. Reversals and delays aren’t repudiations of clean goals; they’re acknowledgments that the infrastructure sequencing isn’t there yet.

Layer on ambitious decarbonization targets and you get real constraints on what, where, and how quickly we can build. That’s not an argument against decarbonization; it’s an argument for honest sequencing. Fast-growing loads (AI/data, onshore industry, EV charging) require firm capacity, long-distance transmission, and local distribution upgrades. Batteries help — but they don’t substitute for backbone transmission or around-the-clock generation. Even data-center operators admit their Plan B is still diesel backup, precisely because local grids and interconnections remain the bottleneck.

There’s also a geography we rarely discuss. Urban cores often have denser grids and shorter interconnection queues; rural counties — where land is available for renewables, storage, and campuses — frequently lack high-capacity lines to move power in or out. The result is friction: data centers default to diesel; factories wait on substations; communities see rising bills before they see local benefits. Until we build the “middle mile” of transmission into rural America, we’ll keep stranding resources in the wrong places at the wrong times.

Add up the math and it doesn’t work — yet. Load is rising faster than dispatchable, low-carbon capacity; transmission approvals and construction timelines lag demand by years; interconnection queues are swollen; and policy zigzags inject uncertainty exactly when we need aligned incentives. The grid doesn’t care who’s in office; it cares about physics, megawatts, and miles of steel in the ground.

That’s why an “all-of-the-above” approach isn’t ideology — it’s common sense. Keep every safe nuclear electron online while we build more; accelerate transmission and distribution with corridor and permitting reform; add flexible thermal capacity (with carbon management where practical) to firm the system; deploy storage where it pencils; and expand renewables where they best match load and transmission. Pair that with clear tariff design so ultra-large, 24/7 loads fund the dedicated upgrades they require — protecting households from unfair cost-shifts.

The legacy we leave our children and grandchildren won’t be measured by slogans but by whether the lights stay on, the air gets cleaner, and the economy keeps creating opportunity. Meeting today’s demand surge — responsibly and affordably — requires urgency and pragmatism in equal measure. We all have a reason to care, because the digital future our kids are building runs on electrons — and the bill shows up at home.