OpenAI's Michigan data center draws thousands of electricians
AI companies are competing for skilled trade workers to build and run the data centers their models depend on, and the pull is strong enough to bend local labor markets. In Detroit, Tyler Shelton, a 29 year old electrician in training, spends most days fixing cables in manholes, but his company has been sending more and more of its workers to a large data center site about an hour away that recently broke ground. He could soon join them: OpenAI's project in Saline Township is, according to the state of Michigan, the largest single investment in the state's history, and it needs hundreds of electricians working 10 hour days, seven days a week. Shelton says the overtime pay is tempting but he also wants a life outside work and hopes to buy a house within a year. At the Detroit Electrical Industry Training Center, where roughly 850 apprentices train, the data center boom is a constant topic; the megaprojects are drawing skilled trade workers to remote job sites with some of the highest pay and bonuses the industry has seen. AI companies are spending hundreds of millions of dollars to line up enough people to build and staff these facilities. The piece also notes a broader labor shift since ChatGPT's release: commercial construction hiring has risen even as residential construction hiring has fallen.
Key facts
- OpenAI's data center project in Saline Township, Michigan is, per the state, the biggest single investment in Michigan's history.
- The project requires hundreds of electricians working 10 hour days, seven days a week.
- About 850 apprentices train at the Detroit Electrical Industry Training Center, where the data center boom dominates conversation.
- AI companies are spending hundreds of millions of dollars to secure enough skilled trade workers to build and run data centers.
- Since ChatGPT's release, commercial construction hiring has risen while residential construction hiring has dipped.
Why it matters
The AI buildout has a physical bottleneck that gets less attention than chips or power: the electricians and carpenters who actually construct data centers. OpenAI's Saline Township project shows the scale, described by the state of Michigan as its largest single investment ever, and it needs hundreds of electricians on extended schedules just to get built.
Who it affects
It affects trade apprentices and electricians like Tyler Shelton, whose employer is losing workers to data center sites; training centers such as Detroit's, with about 850 apprentices, where the boom shapes career choices; and the AI companies themselves, which have to spend heavily to staff construction rather than just design and software work.
How to use it
For trade workers or contractors, data center megaprojects currently offer some of the highest pay and bonuses in the industry, though often at the cost of long hours (10 hour days, seven days a week on the Michigan project) and relocation to remote sites. For anyone tracking labor markets, the shift shows up in the data: commercial construction hiring up, residential hiring down since ChatGPT's launch.
How solid is it
The account rests on named, on the record sourcing: a specific apprentice interviewed at his training center, a state government characterization of the Saline Township project's scale, and a construction hiring comparison since ChatGPT's release. It is a single feature rather than an industry wide survey, so it illustrates a trend through one region and one company's project rather than quantifying it nationally.
Risks and caveats
The piece centers on one Michigan project and one training center, so it may not generalize evenly across regions or companies. It does not give exact wage figures or the total number of workers OpenAI's project will employ, describing the scale in comparative terms (biggest investment in the state, hundreds of electricians) rather than precise counts.
“Everybody wants that money.”
— Tyler Shelton, electrician apprentice in Detroit