Kevin O'Leary's Utah data center fight is really about who pays for the AI build-out

When Kevin O'Leary's data center pitch landed in Utah this spring, the loudest chorus in the coverage was not about grid capacity, water rights, or the Bonneville Salt Flats' dust. It was about China. Pundits and rival operators rushed to frame the project as a national-security beachhead, an alleged conduit for Beijing-aligned capital flowing into American compute infrastructure. The public record does not support that framing. What it does support, on everything actually filed and reported, is a more familiar American story: who writes the cheque for an industrial boom in the interior, and who gets left holding the land.

Strip out the geopolitical theatre and the Utah project looks like the rest of the hyperscaler gold rush: a private operator wants to build gigawatts of compute where land is cheap, the climate is forgiving, and the tax base is grateful. The China-blame framing, on what has actually been published, is unsupported by anything more than a talking point. This publication finds that the underlying fight is about something the country has argued about since the Erie Canal: who underwrites the next industrial surge, and on whose terms.

The shape of the deal

O'Leary's vehicle is structured as a public-private partnership with state and county authorities, the same template that has carried most of the recent interior data center boom from Loudoun County to Phoenix to Quincy. The state extends tax stabilization, the county fast-tricks permitting, and the operator promises jobs, power purchase agreements, and a capital base that local school boards and county treasurers would otherwise wait a decade to see. The China angle, in the filings made public so far, does not appear. Funding documents trace to conventional U.S. private equity and a syndicated debt facility that any infrastructure desk in New York would underwrite. No PRC state-adjacent limited partners have been named in any disclosure this publication has reviewed.

That absence has not slowed the rumour cycle. Coverage has leaned on unnamed sources, on adjacent projects where Beijing-linked capital has shown up in other sectors, and on the political convenience of a Beijing-shaped villain. The pattern is worth naming for what it is: when a domestic fight gets expensive, the easiest move in American industrial politics is to import a foreign threat to settle a domestic argument.

The bill always comes due

Data centers are not free. They consume water at industrial scale, lock in multi-decade power purchase agreements that distort local rate bases, and convert farmland into server farms with very few of the multiplier effects a factory brings. A 2026-era hyperscale campus can employ a permanent staff in the low hundreds while drawing power equivalent to a mid-sized American city. The counties that compete for these projects are not stupid about the trade-off. They are betting that the property tax floor, the construction payroll, and the headline prestige justify the give-back. Sometimes they do. Sometimes the same counties end up subsidising grid upgrades for a tenant that re-negotiates the moment its ten-year tax holiday expires.

This is the fight Utah is actually having. Not whether Beijing owns O'Leary's turbines, but whether the state's ratepayers will carry the stranded cost of new transmission lines built to feed a single campus, whether water compacts with downstream agricultural users will hold, and whether the next downturn will find the county treasuries still diversified or suddenly single-crop.

A familiar argument in new clothes

The American interior has been here before. The nineteenth-century railroad land grants traded public acreage for private trackage. The mid-twentieth-century Tennessee Valley Authority spent federal dollars to underwrite regional electrification that the private utilities would not. The semiconductor fabs of the 1980s and 1990s demanded state-financed workforce pipelines and municipal bond issuances. The CHIPS Act of the 2020s repeated the formula with a national-security wrapper. None of those fights were really about who owned the bonds. They were about which level of government absorbs the risk of the first-loss capital that frontier industries require.

The data center cycle is the same argument, with a faster clock and a heavier wattage. And like the earlier rounds, it tends to be settled not by an honest accounting of who pays and who profits, but by whoever can attach the loudest external threat to the project. This time the threat-of-record is Beijing. Last time it was Tokyo. Before that, Moscow. The names change; the fiscal geometry does not.

What the public record actually shows

This publication finds that the China-blame frame around the Utah project is, on the public record, unsupported by anything more than a talking point. Funding traces disclosed to date are conventional U.S. private capital and syndicated debt. No PRC state-adjacent limited partner has been named in any filing this publication has reviewed. The geopolitical noise surrounding the deal is real, but it is not evidence of Chinese ownership; it is evidence of a domestic political fight that has imported a foreign villain to settle the question of who pays for the build-out.

The press would serve readers better by treating the Utah fight as what it is on the documents: a contest between a private operator and a host county over the price of admission to the AI compute economy. The national-security wrapper is rhetorical, not evidentiary. If reporting surfaces a PRC-linked capital stack, this outlet will update. Until then, the China frame should be reported as a frame, not as a finding.

The stakes for the next county

Utah is not the last county that will face this choice. The AI build-out will require dozens of inland sites over the coming decade, and every one of them will face the same offer: trade rate-base stability for headline jobs, trade water compacts for a tenant with negotiating leverage, trade long-term grid planning for a single-counterparty dependency. The counties that sign first will set the template for the ones that follow. The states that negotiate hardest will capture the most. The operators that arrive with the cleanest capital stacks and the most transparent water and power disclosures will earn the longest tax holidays. The ones that arrive wrapped in a national-security narrative, with no underlying paperwork to justify it, are simply asking the public to underwrite their political cover.

That is the fight in Utah. It is not about Beijing. It is about who carries the cost of the next American industrial cycle, and whether the people who live next to the server farms get to see the books before the dirt moves.