THe AI Backlash that toppled utah’s senate president

Yesterday, Utah Senate President Stuart Adams lost his reelection bid in his first-ever Republican primary challenge. 

That sentence should get the attention of every company building data centers, energy infrastructure, advanced manufacturing, or anything else that requires a physical footprint and public approval.

Adams was not a marginal figure. He was one of the most powerful elected officials in one of the most pro-growth states in America. He had served in the Utah Legislature for more than two decades and led the Senate since 2018. Yet in a Republican primary, he was defeated after a campaign in which his support for a major data center project in Box Elder County became a defining issue

The project, backed by Shark Tank star Kevin O’Leary and advanced through the Military Installation Development Authority, which Adams chairs, came to symbolize much more than a single land-use decision. It became a test of whether voters trusted state and local leaders to manage the physical buildout of the artificial intelligence economy.

The lesson is that data center construction — as the most visible physical manifestation of artificial intelligence — is now politically salient enough to help take down some of the most powerful leaders in state government.

That is new. And it should change how companies think about site selection, public affairs, and political risk. For most voters, artificial intelligence is still abstract. It is a chatbot. A headline. A vague concern about jobs. A sense that technology is moving faster than public institutions, local communities, and regulatory processes can absorb.

But a data center makes AI tangible.

It is the building. The water use. The power demand. The transmission line. The tax treatment. The special district. The closed-door meeting. The fear that the benefits will be global, while the costs will be local.

That's why data centers have become a proxy for AI anxiety. They give the public’s unease about artificial intelligence a physical address.

And the public is uneasy.

A 2026 Pew Research Center survey found that Americans are far more likely to believe AI will have a negative impact on society than a positive one over the next 20 years: 40 percent negative versus just 16 percent positive. More Americans also expect AI to hurt them personally than help them personally: 31 percent negative versus 23 percent positive. And despite the constant media attention around Claude, ChatGPT and other tools, 51 percent of U.S. adults still do not use AI chatbots at all.

That is the context in which data centers are being proposed.

The AI industry often talks as if adoption is inevitable and public support will eventually follow. But adoption and trust are not the same thing. A person can use AI at work and still oppose the data center proposed near their home. A voter can understand that AI is powerful and still believe the buildout is moving too fast, consuming too much, and benefiting too few. Disregard these AI adapter-opponents as hypocrites at your peril. 

The industry has made this harder for itself. Unusually for a major technological breakthrough, the most prominent AI leaders have spent the last several years warning the public about the very technology they are racing to build. Anthropic CEO Dario Amodei has warned that AI could wipe out half of entry-level white-collar jobs within one to five years and cause a severe labor-market shock. Other AI leaders have warned about misinformation, national-security threats, model autonomy, and even existential risk.

If the public mostly hears that AI may destroy jobs, undermine trust, concentrate power, and outpace government oversight, then the first large AI facility proposed in their community will not be received as a neutral infrastructure project. It will be received as the local embodiment of a national anxiety.

That is what happened in Utah.

The Stratos data center project became more than a land-use dispute. It became a political symbol. Opponents raised concerns about water, energy, transparency, governance, local control, and whether state-level entities were moving too quickly around local objections. Adams, because of his support for the project and his role with the Military Installation Development Authority, became the face of that controversy.

Yesterday, voters showed that the controversy was not just noise. It had electoral force.

That is the part every company should internalize: Traditional site-selection analysis has focused on land, power, taxes, permitting timelines, workforce, logistics, and proximity to customers. Those factors still matter. But they are incomplete.

Political risk is now a core site-selection variable. 

A project can look perfect in a spreadsheet and still be a political disaster. The land can be available. The tax structure can be favorable. The permitting pathway can appear manageable. The utility interconnection can pencil out. But if the project becomes a symbol of elite indifference, resource scarcity, or technological disruption, the risk profile changes overnight. A one-time win may unfold as a long-term liability.

The question is no longer simply: Can we get this approved? The better question is: Can the political champions survive having approved it?

That is a very different standard.

It requires industry coordination -- aligning companies to think earlier and more seriously about public trust. Who will be asked to defend the project? What local concerns will attach to it? Who benefits visibly? Who bears the costs? Is the jobs number credible? Is the water story defensible? Are the energy impacts understandable? Does the process look transparent? Are local leaders empowered or merely informed?

These are not soft questions. They are business questions.

A company that misreads the politics of a site can lose years. It can trigger litigation, legislative scrutiny, utility backlash, county-level opposition, and reputational damage. It can also make its political allies vulnerable, which is often when support begins to disappear. For on-the-ground developers, the immediate incentive is often to secure approval as quickly as possible. But if that approval creates political exposure for local allies, the short-term win can become a long-term ecosystem risk. CEOs and industry leaders need to step back and assess the implications before the project becomes a symbol. 

Adams' defeat is a warning sign for the AI economy. The same dynamic can apply to energy projects, semiconductor plants, and any other innovation that requires the public to accept visible change in the physical world.

America needs to build. We need the infrastructure of technological leadership. We deserve to access the life-changing benefits of AI. We need energy, compute, manufacturing, connectivity, and the physical systems that allow innovation to scale.

But companies cannot assume that the future sells itself. The future has to be explained. It has to be localized. It has to be defended -- transparently -- in terms that matter to the people who will live with it.

Data centers may power the next generation of artificial intelligence. But they are also becoming the place where voters express their doubts about that future. In Utah, those doubts were strong enough to help end the career of one of the state’s most powerful political leaders.

For innovators, that is the takeaway. AI policies are likely to be largely determined by long-term negotiations in Washington, D.C. But those deliberations will be heavily influenced by repeated experiences at county commissions, zoning boards, water authorities, state legislatures and city halls.

The AI companies and infrastructure industries that engage for the long term and build meaningful local trust will have an advantage. The companies that do not will continue to create liabilities for themselves and their allies — advancing projects that look inevitable on paper but become politically radioactive on the ground.