If you are trying to persuade people to welcome an enormous industrial development into their community, accusing your critics of working for China seems like a peculiar place to start.

Yet that is where Kevin O’Leary took the argument over Stratos, his proposed AI data center and power-generation campus in Utah’s Box Elder County. On Fox Business in May, the investor known as Mr. Wonderful suggested that opposition to the project reflected Chinese interference. He named people and organizations in connection with that supposed campaign.

Weeks later, he acknowledged that he had no evidence that the named critics were funded by China or the Chinese Communist Party. Fox issued its own clarification and apology. A defamation lawsuit followed. O’Leary and Fox have sought its dismissal; the allegations remain unresolved.

The episode opens The Verge’s Oct 1, 2026 investigation, “The People of Utah vs. Kevin O’Leary”. Its reporting, based on interviews and public records, takes us well beyond another argument about whether America needs more data centers. It examines how this particular project was assembled, advanced and presented to people who would have to live with it.

Those people wanted answers about water, heat, land, government authority and what their community would receive. Somehow, that became a conversation about their loyalty.

China is doing an awful lot of work in the American AI sales pitch these days.

The Warning Was About Incentives

Back in May, I wrote “The AI Industrial Complex”, using Stratos to examine the coalition forming around AI infrastructure. The original proposal contemplated approximately 40,000 acres and potential power requirements reaching nine gigawatts. Those numbers demanded a conversation about regional infrastructure, rather than an ordinary commercial development.

I reached for Dwight Eisenhower because of the incentives. Technology companies want compute. Energy suppliers want customers. Investors want returns, and politicians want development announcements. National security officials want America to maintain its technological advantages. Each has a reason to support expansion, and together they create considerable momentum.

My concern was that strategic importance could become a way to avoid difficult questions about a particular project. The Verge’s investigation shows how concrete that concern has become. The people living in Hansel Valley did not necessarily share the timetable, assumptions or enthusiasm of the people promoting its transformation into Wonder Valley.

They deserved a conversation before they became a problem to be overcome.

Everybody Was Racing

One of the most revealing findings in The Verge’s account concerns the county commissioners whose approval O’Leary celebrated. According to the investigation, they were officially told about the project only five weeks before they were expected to vote. Before that, they had heard rumors.

The reporting describes pressure to move quickly, including a developer’s warning that financing could fall apart if county consent did not arrive by a deadline. Meanwhile, residents were trying to understand a project whose scale would be difficult to absorb even with ample time and complete information.

There is a commercial explanation for the urgency. Developers are competing to assemble land, power access and permissions that will attract a small pool of enormous customers. In that market, speed has value. A site with a credible path to energization can be worth far more than the same acreage without one.

But the developer’s deadline is not automatically the community’s obligation.

A family asking what happens to its well, a rancher concerned about neighboring land or a commissioner trying to understand the agreement is doing necessary work. That work takes time. Calling the project essential to national security does not complete it.

The approval process became part of the controversy itself. The Verge describes a contentious meeting, public anger and commissioners moving to another room to finish their action. People who already felt excluded were given another reason to believe their participation was inconvenient.

Once that belief takes hold, every subsequent assurance becomes harder to sell.

What Is Actually Moving Forward?

We should be careful about declaring Stratos finished, either as a success or a failure.

The Verge reported that, as of late September, MIDA had left its signature off the interlocal agreement and paused work amid litigation over the referendum. It also reported that none of the allotted land had been sold and no tenant had been announced. Those are dated findings, not a declaration that the project can never proceed.

An Oct. 2, 2026 KSL report published by Deseret News gives the developer’s response. O’Leary Digital CEO Paul Palandjian says predevelopment, engineering and studies continue. MIDA’s paused activity and the company’s continuing design work can coexist. Neither establishes that construction has begun.

Even the public specifications require care. The developer’s current FAQ describes approximately 13,000 acres, up to two gigawatts of first-phase on-site generation supporting roughly one gigawatt of IT load, and a potential long-term capacity of approximately 7.5 gigawatts. The Oct 2, 2026 report instead cites a first-phase generation limit of 1.5 gigawatts.

Those descriptions need reconciliation. They also need to be read correctly. Electricity generation and computing load are different measures. A project area is not a building footprint. A maximum buildout is not installed capacity.

Assembling a development opportunity can be a legitimate business. But acquiring land rights, attracting a tenant, securing financing, obtaining permits and delivering operating infrastructure are separate accomplishments. Public discussion should identify which ones have actually occurred.

For all the enormous numbers attached to this proposal, the relevant question remains remarkably ordinary: What commitments stand behind it?

Can the People Vote?

The referendum fight gets to something deeper than project management.

The Box Elder Accountability Referendum group, or BEAR, sought to put the county’s resolutions before voters. The county attorney rejected the applications, classifying the resolutions as administrative actions implementing existing law. Under the county’s interpretation, they are not eligible for referendum.

BEAR disputes that conclusion. As The Salt Lake Tribune reported in September, the case also involves language in the MIDA statute making local consent “irrevocable.” Project backers argue that voters cannot undo consent the county itself cannot reverse. Opponents argue that the statute cannot extinguish their constitutional referendum rights.

The courts will decide those questions. The editorial question is why a decision with such substantial consequences appears to leave residents fighting over whether they can participate at all.

MIDA is Utah’s Military Installation Development Authority, a state-created entity with significant development powers. Its involvement supplies a framework for advancing Stratos. It does not eliminate the need to explain that framework to the public or justify the authority being exercised.

When consent may be irreversible, the process of obtaining it deserves more scrutiny. Residents should understand what their representatives are committing to while there is still an opportunity to influence the decision.

China Cannot Answer the Water Question

The developers deserve a fair hearing on their engineering and economic case.

They propose on-site power generation, which could reduce dependence on Utah’s public electrical grid. They say the cooling approach will limit water consumption. In the Oct 2, 2026 report, Palandjian projects average consumption below 900,000 gallons daily after recycling at full buildout and promises annual public reporting of withdrawals.

Those claims should be evaluated as proposed designs and projections. Consumption and withdrawals are different measures. Actual impacts will depend on what is permitted, financed, built and operated.

Generating power on site may address an important constraint. It also introduces questions about fuel supply, emissions, heat, noise and water. Closed-loop cooling still requires an account of how heat leaves the system. The public needs accessible engineering, clear operating limits and enforceable commitments.

The economic case deserves the same treatment. Construction employment matters, but construction ends. Permanent jobs, public revenue and continuing obligations need their own accounting. A very large investment can be valuable without delivering the kind of broad local prosperity implied by the announcement.

I have been writing about the danger of a new AI “missile gap,” in which fear of falling behind China becomes the default justification for an enormous buildout. The competition is consequential. It still cannot tell us whether this campus, on this land, under these terms, is a sound decision.

And it certainly cannot tell us that the people asking those questions are working for the other side.

Earn the Bargain

Stratos may eventually become a substantial operating campus. It may emerge smaller, later or differently configured. The legal disputes may change its path. None of that requires us to pretend the outcome is settled today.

What is already visible is the cost of treating public participation as something to manage after the important decisions have been made.

Communities can support AI, American competitiveness and economic development while questioning the bargain placed before them. Developers should be able to demonstrate consistent specifications, credible commercial milestones and benefits proportionate to the resources involved. Governments should make the approval process understandable, and protections should survive the ribbon-cutting.

There is a business reason to do this well. Distrust can become litigation, political resistance and delay. The pursuit of speed can end up consuming the very time it was supposed to save.

The AI industrial complex has powerful participants. Its host communities are participants, too, even when they are absent from the financing documents.

The industry wants them to believe in its future. It owes them a credible account of what happens to theirs.