AI’s Next Constraint May Be Getting Things Built
Reuters reported Saturday that Defiance, Ohio put a six-month hold on data center approvals. No company has announced one there.
After the county's economic development office fielded inquiries from tech companies, residents gathered signatures for a November 3 ballot measure to ban all but the smallest data centers. City officials imposed the moratorium.
Defiance isn't isolated. The Ohio Capital Journal reports that voters in 18 Ohio communities will decide data center measures on the same day. Some restrict very large facilities, some rework zoning, and some deal with specific developments.
Nothing has been proposed yet in Defiance, but the conditions under which something could get built are already being negotiated before a shovel is in the ground.
Most of the AI conversation is still about models, chips, compute and software. At scale, the build-out is increasingly physical. Data centers need electricity, which can mean new generation, transmission, substations and grid investment. They also need land, cooling, construction crews and skilled trades.
People who have spent their careers in energy, infrastructure, mining or other capital-intensive industries know what happens next. A project can have committed capital, workable engineering and a sound business case and still run into issues that materially change the schedule.
I've seen how much harder that becomes when stakeholder questions surface after the design has narrowed the available choices.
Governments carry priorities that outlast the project. Regulators have obligations that extend beyond it. Utilities weigh reliability and cost against one customer's timeline. Communities want to know what they're hosting, what it will consume, what it changes and what they get back.
That work belongs inside project design and delivery, not in the communications plan that follows it. By the time a concern becomes public, the project may have fewer options available to respond.
The technology industry moves quickly in the digital world. Physical projects move differently. You're dealing with regulators, governments, utilities and communities while the engineering is still moving underneath you.
Defiance isn't evidence that data centers can't be built, or that communities will oppose them everywhere. But when 18 communities in one state put some form of data-center question before voters, it's hard to treat those issues as peripheral to the build-out.
Done well, much of this looks unremarkable. An approval lands when the plan said it would. An issue surfaces while the design can still respond. A concern gets worked through before it defines the project.
I've spent much of my career working through exactly these issues on major projects.
For those who've delivered major energy or infrastructure projects: what are technology companies most likely to underestimate?
Comments