Skip to main content

Archived Article — The Daily Perspective is no longer active. This article was published on 6 March 2026 and is preserved as part of the archive. Read the farewell | Browse archive

Politics

America's Data Centre PR Problem Has a Substance Gap

Big Tech pledges to pay for AI power, but local communities and regulatory reality remain the real hurdle

America's Data Centre PR Problem Has a Substance Gap
Image: Wired
Key Points 5 min read
  • Seven major tech companies signed White House pledge to cover all power costs for AI data centres, but the agreement is voluntary and unenforceable at federal level.
  • Data centre construction in primary US markets fell to 5.99 gigawatts in 2025, first decline since 2020, due to permitting, zoning, and power hurdles.
  • Over $64 billion in data centre projects have been blocked or delayed by community opposition, particularly in Virginia and other states facing NIMBY backlash.
  • Local authorities, not the White House, control permitting decisions; power grid capacity is booked through 2030 in most markets, forcing reliance on onsite generation.

President Trump observed that data centres "need some P.R. help" because communities fear that local facilities will drive up electricity prices, and on that diagnosis, at least, he was correct. The real problem is that a White House pledge and a public relations campaign are solving an issue that sits well beyond executive reach.

On 5 March, seven leading technology firms including Amazon, Google, Meta, Microsoft, OpenAI, Oracle and xAI signed the Ratepayer Protection Pledge at the White House, agreeing to cover the costs of all new power generation required for their data centres. The commitments sound robust. Companies promise to build, bring, or buy the new generation resources and electricity needed to satisfy their new energy demands, paying the full cost of those resources whether by building, or buying from, new or otherwise additive power plants; where possible, they will also add more capacity that serves the broader public by increasing supply.

Yet the agreement does not carry any concrete, binding commitments. This is not merely a technical distinction. Legal experts note that the White House has the bully pulpit but no legal authority to impose new rate structures, and administration officials have admitted that enforcement of the pledge would have to come from states and state utility regulators. For a policy intended to address a nationwide challenge, this represents a profound structural gap.

The underlying economic pressure is real. Electricity prices are forecast to rise 6 percent through 2026 and another 3 percent in 2028 as data centre demand grows more rapidly than power supply. Public anxiety over this prospect is justified. What is less clear is whether voluntary corporate promises will meaningfully constrain the problem.

The Infrastructure Reality

The industry's construction data tells a more sobering story. There were 5,994 megawatts under construction at the end of 2025, down from 6,350 megawatts in 2024; many planned projects remain delayed due to ongoing permitting, zoning and power procurement hurdles. New data centre capacity under construction in primary US markets declined in the second half of 2025 as community opposition increasingly disrupted planning approvals. This marks the first decline since 2020. Simultaneously, overall vacancy rates fell to a record low of 1.4 percent by the end of 2025, with reduced inventory limiting large-scale deployments.

The crunch is instructive. Supply cannot keep pace with demand even as developers scramble to secure sites. Grid-power capacity for existing projects is largely booked through 2030 in most markets, driving new projects to incorporate on-site power like natural gas generators, wind turbines, hydrogen fuel cells and solar panels with battery energy storage. The irony is that despite massive investment and political will, the foundational problem is one of local politics and utility infrastructure constraints. The White House pledge does nothing to resolve either.

Consider the economics. Rental rates continued to rise due to the limited availability of powered land and capacity constraints in key markets, with rent growth expected to continue outpacing inflation for the next two to five years. Developers are willing to build onsite power and assume infrastructure costs in principle because scarcity has made those investments profitable. Yet the bottleneck is not financing. It is permission.

The Community Question

Local opposition has emerged as perhaps the most durable constraint on data centre expansion. 18 billion dollars' worth of data centre projects were blocked, and another 46 billion dollars of projects were delayed over the last two years in the face of opposition from residents and activist groups. There are at least 142 activist groups across 24 states organising to block data centre construction and expansion.

The objections are not ideologically monolithic. Opposition to data centre development cuts across political lines; Republican officials often raise concerns about tax incentives and energy grid strain, while Democrats tend to focus on environmental impacts and resource consumption. Common concerns include higher utility bills, water consumption, noise, impact on property value, and green space preservation.

Data centre challenges arise primarily at the local level, as most permitting decisions are made by local authorities; consequently, even a supportive White House has limited control over delays arising at the local level. This is precisely the challenge that an executive order or a corporate pledge cannot surmount. The Ratepayer Protection Pledge may assuage some rational anxiety about power costs. It does nothing about the township zoning board.

Developers themselves recognise the challenge. Even before the pledge, tech companies have announced plans to supply their own power; Amazon has announced a deal to co-locate a data centre next to Talen Energy's Susquehanna nuclear plant in Pennsylvania, Microsoft has signed a long-term deal to buy power from the former Three Mile Island nuclear plant also in Pennsylvania, and Google has announced plans to build data centres in Minnesota alongside new power generation through a partnership with Intersect Power in Texas and Xcel Energy. They are already internalising costs. The pledge merely formalises what market pressure has already begun to impose.

A Pragmatic Middle Ground

The energy affordability problem is genuine and deserves policy attention. Rising electricity costs erode household budgets and threaten industrial competitiveness. If data centres are to contribute meaningfully to the power grid burden, insisting they fund their own infrastructure and pay utilities fairly is economically rational. The pledge, in that sense, reflects sound fiscal principle.

Yet the pledge alone is insufficient. Several states are starting to adopt regulations and pass laws that would ensure consumers don't pick up the tab for data centre power, but there is no such federal regulation in place. Effective policy would combine corporate accountability with state-level regulation. It would also require honest conversation with communities about what data centres deliver and what costs they impose, rather than assuming public relations can substitute for genuine local engagement.

The irony of Trump's moment is that it reveals both the value of market-driven solutions and their limits. Markets can incentivise companies to internalise costs. What they cannot do is move electricity through a grid booked until 2030 or persuade a town council to accept a facility that transforms the local landscape. Those problems require regulatory clarity, adequate transmission infrastructure, and genuine community consent. The pledge is a start. It is not the finish.

Sources (7)
Oliver Pemberton
Oliver Pemberton

Oliver Pemberton is an AI editorial persona created by The Daily Perspective. Covering European politics, the UK economy, and transatlantic affairs with the dual perspective of an Australian abroad. As an AI persona, articles are generated using artificial intelligence with editorial quality controls.