We are in the middle of the largest infrastructure buildout of our lifetimes, and most people still think we're talking about real estate.

We're not.

The asset class the industry used to call a "data center" — the windowless shell, valued per square foot, tucked along a fiber route — is gone. What's replacing it is energy infrastructure that happens to have a building on top of it.

That pivot, from real estate to energy, is the single most important thing site selectors, economic developers, and community leaders need to internalize right now. Because the next economic map isn't going to be drawn in square feet. It's going to be drawn in megawatts.

The Metric Has Flipped

For about fifteen years, this business was valued in dollars per square foot. You'd talk about the shell, the rack density, the fiber route, and the lease rate. Power was assumed. It was a line item on the utility bill.

That world is over.

Today the governing metric is dollars per megawatt. Power isn't a line item anymore — it's the entire pro forma. Every site tour I do now starts with a transmission map, not a plat.

We have moved, in about 24 months, from Office Parks to Industrial Power Campuses. A legacy colocation facility was 20 to 50 acres on a 138kV drop. A modern hyperscale campus is 500 to 1,000 acres at 345kV. The gigascale AI clusters now being scoped are 1,500 acres and up, a gigawatt-plus of demand, and 500kV service. This is heavy industry. It just doesn't look like it.

The Math Nobody Wants to Look At

Here's what's actually driving all of this.

U.S. load growth forecasts nearly doubled between 2023 and 2024 — AI workloads, transportation electrification, and reshored manufacturing all landing on the grid at the same time. Meanwhile, about eleven gigawatts of firm, dispatchable generation retires every year, and most of what's coming online to replace it is intermittent.

Run the numbers. We're shedding roughly 83 gigawatts of firm capacity this decade while adding close to 90 gigawatts of new load. To replace one gigawatt of dispatchable reliability with solar nameplate, you need four to five gigawatts of panels.

We are not building fast enough. A structural power shortage is expected to peak between 2028 and 2030, and that timeline isn't hypothetical. It's in the queue studies.

Which means growth isn't going where land is cheap. It's going where firm power exists.

Where the Growth Is Actually Going

The legacy markets are effectively capped. Northern Virginia and Silicon Valley have run out of room on the grid. Dallas and Atlanta are tightening fast.

The runway is a corridor across the Midwest and South with diverse generation, transmission headroom at 345kV and above, and a willingness to plan.

Call it the Power Belt.

Behind-the-meter generation will keep solving today's speed-to-market problem, but it's a bridge, not a destination. The long-term winners will invest in 345kV-and-above transmission corridors now, so when the temporary fixes hit their limits, a second wave of deployment has somewhere to land. Nuclear co-location, grid-enhancing technology that squeezes 30% more capacity from existing wires, and eventual small modular reactor integration will all find homes in the regions that did that groundwork.

These Are Not Bad Neighbors

I want to push back on something I hear in nearly every town hall.

The modern campus is one of the quietest industrial neighbors a community can attract. Closed-loop cooling uses less water in a year than an 18-hole golf course. Acoustic engineering keeps the property line under 55 decibels — quieter than the nearest highway. Traffic is a fraction of a logistics facility. No school-age children. Minimal sewer load. The campus typically pays for the substation that makes everyone else's power more reliable.

The cost-of-community-services math is hard to argue with — about twenty cents of services for every dollar of revenue, versus more than a dollar for residential.

That's the tax base of heavy industry with the traffic profile of a library.

The Right Historical Parallel

When I'm trying to help a community see where they sit in history, I don't reach for the dot-com era. I reach for the railroad and the interstate.

In the 1860s, towns that secured a depot thrived. Towns bypassed by the rail became ghost towns. In the 1950s, the Federal Highway Act redrew the map around I-80 and I-95, and those decisions defined fifty years of winners and losers.

Fiber and firm power are the asphalt and steel of this cycle. A region without them will be flyover country in the intelligence economy, no matter how nice the downtown looks.

What to Actually Do

The playbook isn't complicated. It's just rarely done in the right order.

First, identify. Map every 138kV-and-above line in the region. Overlay fiber. Catalog parcels above 500 acres. Know what you have before anyone asks.

Second, entitle. Build technology overlay districts. Pre-approve compatible uses. Get environmental studies done before the first site selector calls.

Stay in continuous conversation with your utility and transmission operator — not transactional, continuous. And lead the conversation with your residents. If you don't, the loudest voice on social media will lead it for you.

The Window

The infrastructure of intelligence is being built right now.

Whether any of it gets built in your region is a choice.

And the window to make that choice is measured in months, not years.