Meta warns political backlash could slow AI data centers — but says the buildout still pays

Updated November 5, 2026 10:55 pm

Meta says the political backlash now building around data centers could make it harder to win approval for the huge infrastructure its AI strategy depends on, even as the company believes those projects can still generate revenue and support its broader AI push. The warning came Wednesday from Meta president Dina Powell McCormick at the WSJ Tech conference, where she said post-midterm politics may make future buildouts more difficult and public skepticism more visible.

The company’s message is blunt: data centers are no longer just a technical necessity, but a political liability. Meta says it has operated data centers for 20 years, now has facilities in 23 states, and is pushing ahead with five planned projects at 1 gigawatt or more, including one in Texas. But the company also worries that campaigns against data centers could turn local objections into broader resistance to AI itself.

What Meta is warning about

Meta’s core concern is that the next phase of AI infrastructure expansion will face more scrutiny from voters and local communities, especially after Democratic midterm wins, according to TechCrunch’s account of Powell McCormick’s remarks. She linked opposition to data centers with wider unease about artificial intelligence, arguing that the physical symbol of the AI boom is often a giant facility that neighbors may not want nearby.

That framing matters because Meta’s AI plans depend on real-world buildouts: land, power, cooling, and local permits. If those become harder to secure, the company could face delays or higher costs just as it is trying to scale products and monetize its compute capacity.

“I think it’s going to be harder,”

Powell McCormick said, referring to data-center construction in the political environment after the midterms, according to TechCrunch.

Why it matters

The stakes go beyond Meta. If communities and elected officials become more resistant to data-center construction, the AI industry could face a bottleneck that affects everything from model training to product deployment. Data centers are the backbone of cloud computing and AI workloads, and they require heavy power consumption, cooling systems, land, and usually local approvals.

For readers, that can translate into visible neighborhood disputes over electricity demand, noise, tax incentives, land use, and the balance between promised jobs and actual local benefit. For the market, it means AI growth may increasingly depend not only on chips and models, but on politics, utilities, and zoning boards.

Meta is also signaling that it believes the public narrative around AI has become part of the permitting fight. The company says it may improve sentiment by better explaining the value of both the local infrastructure and the technology that runs on it. That is a broader public-relations challenge, not just an engineering one.

How does Meta think it can win local support?

Meta’s answer is to make the benefits tangible. Powell McCormick said communities are more likely to support projects when they see direct local gains, and she described the company’s preferred deal terms as including paying its own power costs, hiring locally, and directing tax gains to teachers and first responders.

That approach fits with a familiar pattern in data-center politics: companies try to frame large facilities as long-term economic assets rather than extractive industrial projects. Meta’s argument is that if residents see lower fiscal risk and more visible community upside, opposition may soften.

She also suggested Meta has not done enough to explain why these investments matter. In her words, the company has “not done as good a job as we needed to explain the benefits of these investments,” a sign that Meta sees communications as part of its infrastructure strategy, not an afterthought.

What Meta says it is trying to show

Meta says it wants to demonstrate value in two directions at once:

• local value, through jobs, taxes, and power arrangements that communities can understand;
• product value, through the AI tools that the infrastructure helps support.

That dual message is important because Powell McCormick framed Meta’s AI mission in broad public terms, including transformative uses in cancer-related and education-related applications. The company appears to be arguing that a data center is not just an industrial building; it is the physical foundation for widely useful software.

What the numbers say about Meta’s buildout

Meta provided several figures that show how large the strategy already is and how much more expansion it wants to pursue. The company said it has been building data centers for two decades, operates them across 23 states, and has five planned projects at 1 gigawatt or more, including one in Texas.

The scale is notable because Meta’s AI infrastructure strategy is not being described as a hypothetical future plan. It is already embedded in the company’s operations, and the company is still adding to it. That makes any political slowdown more consequential than if the buildout were just beginning.

Background: how we got here

Data centers sit at the center of the AI economy because training and running large models requires massive computing capacity. That means more servers, more networking gear, more electricity, more cooling, and more land than most communities are accustomed to hosting. The result is a growing tension between AI companies that need infrastructure quickly and local residents who may see the projects as disruptive or opaque.

Local opposition often focuses on practical concerns. Residents and officials may question power use, environmental impact, tax incentives, noise, or whether promised economic gains are worth the trade-offs. Those arguments can gain force when data centers become campaign issues, which is why Meta appears to be watching the post-midterm climate closely.

Meta, for its part, has been investing heavily in both AI infrastructure and consumer-facing AI products. The company’s strategy appears to rest on a simple proposition: if it can build enough compute, it can support more useful products, and if those products reach enough users, the infrastructure becomes economically defensible.

Muse and the monetization question

Meta says its consumer AI agent Muse now has more than 10 million users. That is a meaningful audience, but Meta is also signaling that it does not want the product’s success or failure to determine the entire business case for its infrastructure buildout.

Instead, the company says it sees another path: leased compute demand. In other words, even if Muse does not scale as hoped, Meta believes there is still a way to make money by supplying computing capacity to others. Powell McCormick said the company is “seeing a path to monetization,” underscoring that the infrastructure itself may be a revenue engine.

What TechCrunch reported from the WSJ Tech conference

According to TechCrunch, Powell McCormick made the remarks on Wednesday at the WSJ Tech conference. She said the political environment after the midterms would make data-center construction harder and suggested that community resistance to AI infrastructure and unease about AI more broadly are connected.

She also said Meta has spent 20 years building data centers and now operates them in 23 states, reinforcing that this is not a new bet but a large-scale extension of an existing one. The comments indicate that Meta is trying to reconcile two realities at once: the company needs more compute, but that compute is becoming more politically visible and more difficult to site.

Comparison: the two sides of Meta’s AI bet

What happens next

The immediate question is which of Meta’s planned projects could face the most resistance as the political mood changes. The fact sheet does not identify specific sites beyond the Texas project, but it does suggest that future permitting battles may be harder to win.

Watch for three things in particular:

1. whether any state or local governments change permitting or incentive terms;
2. whether Meta adjusts how it presents power, hiring, and tax arrangements to communities;
3. whether leased compute and Muse growth are enough to support the business case if construction slows.

Another open question is how much of Meta’s AI expansion truly depends on new builds versus leased capacity. If the company can lean more heavily on leased compute, it may have more flexibility. If not, the infrastructure push becomes more vulnerable to local politics.

Meta is clearly trying to get ahead of that possibility now. By framing data centers as both the visible face of AI and a source of local benefits, the company is making the case that the industry’s physical footprint should be treated as part of the public interest, not just corporate expansion.

For now, Meta’s message is that the backlash is real, the politics are tougher, and the buildout still has to continue.

Key facts

• Meta says it has been building data centers for 20 years and operates them in 23 states.
• The company says five planned projects, including one in Texas, are at 1 gigawatt or more.
• Meta says its consumer AI agent Muse has more than 10 million users.
• Dina Powell McCormick delivered the remarks Wednesday at the WSJ Tech conference.
• TechCrunch reported that Powell McCormick tied post-midterm politics to a harder path for data-center construction.