Grid access is now the gate for data center capital

Data center development across the Americas has reached a record 37.7 gigawatts under construction, and the variable that now sorts the map is whether a project can get a wire. In a report relayed by Connect CRE, Cushman & Wakefield puts power availability, grid infrastructure and regulatory certainty among the factors that decide where that growth can occur, and therefore where private real estate capital can be deployed at all.

John McWilliams, the firm's head of data center insights, doesn't treat the shortage as distant. “Everyone knows the demand is insatiable,” he said. “The challenge today is finding the power and infrastructure to accommodate that development.” Power availability, grid infrastructure and regulatory certainty, he added, “are becoming defining competitive advantages for markets across the Americas,” and that, in his telling, is beginning to redraw the development map.

The 37.7 gigawatts is a construction figure, but the geography beneath it is a queue that splits two ways. Northern Virginia, Atlanta, Chicago and Dallas keep attracting significant investment on the strength of connectivity, customer concentration and mature infrastructure ecosystems — the latency-sensitive, enterprise-facing work that does not relocate. West Texas, Pennsylvania, Alberta and Cheyenne are pulling a different kind of project: some of the region's largest proposed developments, sited where the power and the land are available rather than where the fiber runs thickest.

That split is where the underwriting separates. An established hub sells certainty: a tenant signs because the campus, the substation and the permitting regime have been proven at scale, and because its peers are next door. A power-available market sells scale, and right now it is selling scale on proposals — the operative word in the Cushman & Wakefield finding, and doing real work, because the region's largest proposed developments are landing in markets that have not carried this load before. Each is a wager on an interconnection calendar as much as on the dirt underneath it.

The bottleneck has moved to the substation, as this publication argued this week, and a pipeline nearly six times the installed base is being underwritten as a leasing story when the binding constraint is interconnection timing. Cushman & Wakefield has now drawn that argument as a map, which is the more useful artifact for a site-selection committee, because a map shows where the queue sits.

The two lists are different products, which makes the redirection awkward for allocators who have spent a decade buying the same four markets. Connectivity and customer concentration are advantages a market accrues over decades and cannot easily copy; power availability is a function of geology, grid capacity and permitting posture, and it can belong to a market that has never hosted a hyperscale campus. Capital that treats the second as a lesser version of the first will misprice both halves of the map.

Outside digital infrastructure, the year's defining fact has been a construction freeze — the drop in apartment, office and industrial starts that has returned leverage to owners who already hold product. Data centers are running the other way, and their delivery depends on a grid they do not control.

The equity side of the trade is where the mispricing most likely sits, because what prices a data center today is the energy, land and regulatory calendar, and the debt markets have begun charging for that calendar while equity largely has not. Of the Cushman & Wakefield trio, regulatory certainty is the leg least controllable from a sponsor's side, and a record construction pipeline is where that gap should surface first — rather than in headline rents, in the spread between what a sponsor pays for power-available land today and what the energization schedule costs tomorrow.

What a buyer is actually acquiring has changed with it: in Northern Virginia, Atlanta, Chicago and Dallas, incremental capacity is contested, so the premium sits on assets that are already energized or fully permitted; new greenfield in those markets is a queue-position play with a schedule attached. In West Texas, Pennsylvania, Alberta and Cheyenne, the scarce good is land with a credible path to interconnection, and the asset being priced is the interconnection agreement, not the address.

Both are defensible trades at the right basis: paying up for power-available land with a credible position in the queue is one, and paying connectivity prices for greenfield in a hub where the energization date belongs to somebody else's capital plan is the other. The distinction is easy to lose on a site plan, because the parcels look alike and so do the pro formas, right up until the schedule slips.

What to watch is conversion: the projects in West Texas, Pennsylvania, Alberta and Cheyenne are proposals, while the 37.7 gigawatts is what is under construction. How quickly the first list moves toward something like the second, and which of those markets does the moving, is the cleanest available read on how much of the redrawn map becomes steel and how much stays a slide. It is also the test that belongs beside every data center memo crossing an investment committee this quarter.