AI Emerges As Dark Horse Of The Midterm Elections

Voters in communities targeted for massive datacenter development pushed back at the ballot box. Their message leads to a consideration for business: What happens to enterprise AI when the infrastructure needed to power it becomes politically harder and more expensive to build?

(Image: Getty)

Though artificial intelligence didn't literally appear as a candidate on the ballot during the 2026 midterm elections, its infrastructure did.

AI datacenters were the dark-horse candidate of the midterms, appearing again and again in fights over electricity bills, water, zoning, tax incentives and local control.

The results sent a resounding message.

According to Ballotpedia's datacenter ballot tracker, voters faced 41 datacenter-related measures across 10 states on Nov. 3, with 28 measures aiming to restrict development. Twenty-one of the 41 were in Ohio alone.

As results came in on Wednesday, voters had sided against datacenter development in every decided ballot measure except one, in Harwood, North Dakota, according to a CNBC analysis.

An NBC News exit poll found 56 percent of voters considered AI datacenters a crucial or major factor in their vote for Congress. Moreover, 76 percent said they would oppose construction of new datacenters in their community.

It seems the frenetic trajectory of AI's rise is colliding headfirst into public resistance to the massive physical infrastructure needed to power it.

That conflict poses a question for midmarket IT leaders to consider: What happens when AI compute demand keeps rising, while the infrastructure needed to support it becomes more expensive and more politically fraught to build?

That's where the datacenter issue shifts from a political issue to a business one. Organizations have dual stakes in the argument: They're electric utility customers potentially exposed to the costs of AI's enormous power demands, and they're AI consumers trying to keep up with stakeholders' demands that the technology will transform their business.

As the datacenter debate promises to rage on well past the midterms, IT leaders mapping out their AI 2027 plans may want to consider another factor: AI infrastructure expansion may not happen as quickly or as cheaply as the industry envisioned a year ago.

AI may live in the cloud, but the cloud needs physical infrastructure. And voters just made clear they intend to have a say in where.