AI companies want more computing capacity, but voters in many communities have made building it harder. In the midterm elections, local measures imposing data center restrictions or moratoriums largely passed, while Texas elected Gina Hinojosa, a Democrat who campaigned on restricting data centers.
A local issue with national reach
Data centers were not on every ballot, and their role in the wider Democratic victory is difficult to separate from the many other reasons people voted. Even so, opposition to the projects has become a grassroots issue with appeal across party lines.
Some of that resistance stems from how companies have sought approval. Developers have sometimes required nondisclosure agreements from elected or appointed officials and withheld details about the scale of proposed projects, including in environmentally sensitive areas. In the Southwest, water scarcity adds to local concerns.
Voters in many places had a direct say. While some ballot measures concerned proposals to speed up construction, many would impose restrictions or moratoriums, and almost all of those passed. The outcome points to a tougher approval process for at least some future projects, though the effect will depend on the rules adopted in each place.
Texas adds to the political pressure
Hinojosa defeated Greg Abbott after making data center restrictions part of her campaign. Her election puts the issue in front of a new governor in a state where Elon Musk's companies are headquartered and where he is a major figure in the tech industry.
Musk also donated roughly half a billion dollars to Republicans, with the final total still to be counted, according to an estimate discussed in the report. The election results leave his companies facing a different political climate in Texas despite that spending. The specific policies Hinojosa will pursue remain to be seen.
OpenAI changes its approach
OpenAI has released what it calls a compute campus policy blueprint. The company says it will stop requiring nondisclosure agreements and share its plans, a shift toward greater openness about proposed projects.
That addresses one of the complaints behind local opposition, though the report does not establish how the policy will work in practice or whether it will change the outcome of future approvals. Sharing plans and securing permission are separate tasks.
The terminology has attracted attention too. OpenAI calls the sites compute campuses, a phrase that may sound friendlier than data centers. A new name, however, does not answer residents' questions about a project's scale or its demands on local resources.
Construction meets financial pressure
The election is one source of pressure on data center builders. The report also describes concern among investors about how much hyperscalers and AI companies are spending on data centers and computing capacity.
Those pressures may pull against the companies' ambitions. Building at the scale they want is likely to require more data centers somewhere, even as some communities vote to limit them and investors question the expense. OpenAI's policy offers one response to the approval problem. The harder question is whether communities and investors will find the projects themselves more persuasive than the name.