Power has not replaced land in North Carolina. For the largest loads, it has joined it.

Site-selection consultants now rate electric power availability at scale as important for 98.5% of projects, and a reading follows easily: that grid connection has overtaken land entitlement as the bottleneck for large industrial projects. We put that reading to the North Carolina record.

On the public record, it has not. Not for most projects, and not yet.

The claim is narrower than it sounds. “Power” covers three different things, local grid headroom, generation for peak demand and a regulator's confidence in a forecast, and each binds different projects on a different clock.

We traced 30 large North Carolina projects from 2020 to October 2026, across data centres, semiconductors, EV and battery plants, aerospace and life sciences. For each we asked one question: is a sponsor, utility or regulator on record, or credible reporting, attributing a change in location, scale or timing to power? We also read the large-load settlement filed with the state's Utilities Commission on 6 October 2026, and set its terms against three neighbouring states and the Tennessee Valley Authority.

Eleven of the 30 record a change. None has power on record as the reason. Where a reason is stated, data-centre proposals ended at local entitlement and manufacturing projects at capital and market conditions. The finding covers the public record, and it has a limit we could not close: North Carolina publishes the projects it wins, not the ones it loses. We hold it at medium confidence.

The picture changes above a line the state's own pending tariff draws: 150 MW, or 50 MW running at 80% load factor or above. There, power is a co-primary constraint. One developer's SEC filing records a utility schedule of 24, 40 and then 99 MW within four years, offered as “commercially reasonable efforts”. The state air regulator has permitted 57 utility-run engines to bridge a data-centre campus until it is connected to the grid. Even at that scale power is coupled with land rather than replacing it: at least 11 counties and 17 towns had adopted data-centre moratoria by August 2026.

What North Carolina does well

Its pending large-load terms sit within the regional range on every dimension we could verify, including a 75% minimum bill and minimum terms of 10 or 15 years, and they are set out in a filed public tariff, which is more than most of the comparators offer. For a sponsor doing diligence, that openness is an advantage.

What is missing is a date. Nothing on the public record obliges the utility to serve a large load by a date, and no North Carolina utility publishes site- or substation-level capacity. On 18 September 2026 the Commission declined to approve new generation against a load forecast it found “insufficiently reliable”. The most visible power constraint in the state is regulatory as much as physical.

Two things surprised us

Our view, for what it is worth

We went in expecting to confirm that power had overtaken land, and we no longer think that is the useful question. For most projects, water, wastewater, local entitlement and the sponsor's own capital deserve the same scrutiny as power. For the largest loads, power belongs before any acreage is committed, and the only evidence worth relying on is a dated, written position from whoever holds the decision. A certification badge is not that. Neither is a published site list. Four questions seem worth asking before land is committed:

These are questions, not advice, and the decision stays with the reader. Four public decisions will change the terms, beginning with the Commission's order on the settlement, expected by mid-November 2026.

If the public record misses something you know, we would like to hear it.