Data Centre Projects Face Mounting Pressure After US Midterm Elections

A Political Reckoning for Infrastructure

The 2026 United States midterm elections have introduced a new constraint for hyperscale computing infrastructure. Across dozens of jurisdictions, voters approved measures imposing restrictions or moratoriums on data centre construction. In states where the issue did not appear directly on ballots, candidates who campaigned on limiting such projects won decisive victories. Texas elected Gina Hinojosa as governor after she made data centre oversight a central campaign promise, unseating incumbent Greg Abbott in a state that has historically leaned Republican.

The electoral outcome presents a challenge for companies that have spent the past two years announcing ambitious infrastructure buildouts to support AI model training and inference workloads. What began as isolated community opposition has consolidated into a political force that crosses party lines, though Democratic candidates proved more effective at channelling voter concern into electoral success.

The Texas Reversal

Texas represents the starkest example of the political shift. The state has attracted significant investment from technology firms seeking favourable energy costs and regulatory environments. Elon Musk, whose companies Tesla, SpaceX, and xAI maintain major operations in Texas, contributed an estimated half a billion dollars to Republican candidates and causes during the midterm cycle. The investment yielded little return; Democrats captured both the governorship and majorities in the state legislature for the first time in decades.

Musk's xAI operates a large-scale training facility in Memphis, Tennessee, and has announced plans for expansion. The political environment in Texas, where much of his corporate infrastructure is concentrated, now includes leadership explicitly committed to increased scrutiny of data centre projects. Hinojosa has stated that her administration will prioritise environmental impact assessments and community consultation before approving new facilities.

The Approval Process Under Scrutiny

The election results reflect accumulated frustration with how hyperscalers have pursued project approvals. In multiple jurisdictions, companies approached local officials under non-disclosure agreements, limiting public discussion of environmental impact, energy consumption, and water usage. This approach proved particularly contentious in water-scarce regions of the Southwest, where data centres require substantial volumes for cooling systems.

The strategy of seeking approval through confidential negotiations, rather than transparent public processes, created a backlash that extended beyond environmental activists. Voters across the political spectrum expressed opposition to projects they perceived as imposed rather than negotiated. Ballot measures that would have expedited data centre permitting were defeated in nearly every jurisdiction where they appeared. Conversely, measures imposing restrictions or requiring additional environmental reviews passed with significant margins.

OpenAI's Policy Pivot

Within days of the election, OpenAI published what it termed a "compute campus policy blueprint." The document represents a marked departure from industry practice. The company committed to abandoning non-disclosure agreements in discussions with local governments, providing detailed project specifications before seeking approvals, and engaging in public consultation processes.

The document also introduced new terminology. OpenAI now refers to its facilities as "compute campuses" rather than data centres, an apparent effort to distance itself from a phrase that has acquired negative associations. The rebranding extends to architectural renderings that emphasise landscaping and community spaces, though the core function of the facilities remains unchanged.

The speed of the policy shift suggests that companies recognise the electoral results as a signal of regulatory tightening ahead. When faced with the prospect of formal restrictions, OpenAI opted to pre-emptively adjust its approach. Whether other hyperscalers follow remains unclear, though the political incentives now favour transparency.

Wall Street's Parallel Concerns

The regulatory pressure arrives alongside a shift in investor sentiment. Over the past six months, equity analysts have begun questioning the capital efficiency of hyperscaler infrastructure spending. Several major technology companies have seen share prices decline following earnings reports that detailed multi-billion-dollar data centre commitments without corresponding revenue growth.

The financial pressure and the political pressure point in the same direction, creating a dual incentive to slow infrastructure expansion. Yet the technical requirements of frontier AI development have not changed. Training runs for large language models and multimodal systems require concentrated compute resources that existing facilities cannot always provide. Companies face a tension between the scale their product roadmaps demand and the financial and political costs of building to that scale.

Regional Strategies and Constraints

The political landscape now varies significantly by jurisdiction. States with Democratic control are likely to impose stricter environmental reviews, water usage caps, and community consultation requirements. Republican-controlled states may continue to offer streamlined approvals, though even in those jurisdictions, local opposition has proven effective at delaying or blocking projects.

Some companies are exploring international expansion as an alternative. Jurisdictions in Southeast Asia, the Middle East, and parts of Europe have actively courted hyperscale investment, offering tax incentives and regulatory concessions. However, data sovereignty requirements, latency considerations, and export controls on advanced semiconductors limit the extent to which companies can shift infrastructure outside the United States.

The 2028 Horizon

The immediate effect of the midterm elections is clear: data centre projects in the United States will face heightened scrutiny and longer approval timelines. The longer-term question is whether the issue retains political salience through the 2028 election cycle. If opposition to data centres continues to mobilise voters, candidates in both parties may adopt restrictive positions, narrowing the strategic options available to hyperscalers.

For now, the industry is adjusting. OpenAI's policy document represents one approach, emphasising transparency and community engagement. Other companies may pursue different strategies: smaller, distributed facilities rather than large campuses; partnerships with existing industrial sites that already have power and water infrastructure; or renewed focus on efficiency improvements that reduce the resource intensity of compute workloads.

The political economy of AI infrastructure has entered a new phase. The assumption that demand for compute would automatically translate into approval for the facilities to supply it no longer holds. Companies that ignored community concern during the expansion phase are now recalibrating, facing an electorate that has demonstrated it will use available political tools to impose constraints.