OPEIU President Urges Unions to Back an AI Data-Center Moratorium

Tyler Turner, president of the Office and Professional Employees International Union, is asking labor to stop treating the AI data-center boom as an uncomplicated source of union construction work. In a newly published opinion essay, he calls for a moratorium on new AI data-center construction until protections for workers, communities and the environment are in place.

Turner’s argument turns on a clash within organized labor. He credits the North American Building Trades Unions with securing well-paid temporary work through agreements with BlackRock, Meta and OpenAI, but argues that the deals support facilities meant to power systems that may displace workers across industries. He says data centers create very few long-term jobs once construction ends.

A construction-job bargain with a larger labor cost

That distinction is central to the case. Building a data center requires a large, time-limited workforce; operating one requires far fewer people, according to Turner. He argues that unions should weigh the immediate benefit to construction members against what he describes as the industry’s longer-term aim: using AI to reduce companies’ labor costs.

Turner is not arguing that workers should receive none of AI’s gains. He instead calls for a worker-centered framework that could include higher wages, shorter workweeks or the removal of dangerous tasks. But his proposed sequence is blunt: pause new construction first, then negotiate the terms under which AI infrastructure proceeds.

Our data fuels their algorithms, our taxes fund their development and our communities host their hardware.

Tyler Turner, president of OPEIU

A pause as negotiating leverage

Turner backs the Artificial Intelligence Data Center Moratorium Act, introduced in Congress by Bernie Sanders and Alexandria Ocasio-Cortez. He also points to New York’s halt on issuing data-center permits for up to a year while the state develops a regulatory framework, calling that approach a practical but insufficient first step.

The mechanism Turner favors is less a permanent rejection of data centers than a change in bargaining power. A construction pause, he argues, would give lawmakers time to assess impacts and give workers a chance to demand conditions before projects lock in public subsidies, power needs and local commitments.

The economic evidence Turner puts forward

His essay uses individual projects and workforce accounts to challenge the standard jobs-and-investment pitch. Turner cites a Rockland County data center that reportedly received $77 million in tax breaks and created one permanent job. He also cites a Virginia study finding that data centers produced less than 1% of the permanent jobs per dollar invested than other industries.

The employment concern is personal to the union he leads. Turner says an OPEIU membership survey found reports of AI-related layoffs, surveillance and workers being punished for errors generated by AI models. Those reports do not establish that every data-center project produces those outcomes, but they explain why Turner treats the facilities as part of a labor question, not just a construction question.

Who pays for the computing buildout

Turner pairs the job argument with an energy argument. He says 335 AI data centers in Texas are already consuming enough electricity to raise rates for other customers, with at least 248 more facilities planned. He argues that the scale of demand makes community consent and sustainable construction conditions essential rather than optional.

The essay puts a difficult choice to unions that have often fought for good construction jobs: should they help build the infrastructure now and seek safeguards afterward, or withhold support until those safeguards are secured? Turner’s answer is clear. Whether other labor groups accept that sequencing may determine how unified labor is as AI infrastructure expands.