Trump administration pushes Big Tech toward voluntary AI data center energy pact

The Trump administration is seeking public commitments from some of the world’s largest technology companies to a new agreement governing the rapid expansion of artificial intelligence (AI) data centers.

A draft agreement obtained by POLITICO outlines commitments designed to ensure that energy-intensive data centers do not drive up household electricity prices, strain water resources, or compromise grid reliability. The proposal also requires companies driving this surge in demand to bear the full cost of building the necessary new infrastructure.

The proposal, which remains in draft form and is subject to change, is framed as a “voluntary agreement” between President Donald Trump and major US tech firms and data center developers. The accord could bind AI giants such as OpenAI, Microsoft, Google, Amazon, Meta, and others to a broad set of energy, water, and community principles.

While the administration intends to announce the initiative during a high-profile White House event, the program has not yet been formally unveiled. It remains unclear which companies have already agreed to the terms or have been invited to participate.

This initiative represents one of the most ambitious efforts to shape AI infrastructure without imposing direct regulation. It comes just one month after the White House issued an unprecedented call to the Mid-Atlantic power grid operator to lower electricity prices. Concerns are mounting that the massive energy consumption of data centers could push prices even higher, potentially becoming a political liability for an administration that has otherwise championed rapid and unfettered data center development.

The agreement is also seen as a way to highlight efforts to mitigate the influence of these corporations ahead of the midterm elections.

“As President Trump announced weeks ago, leading tech companies are working with the President to ‘pay their fair share’ for energy consumption as they build out data centers,” White House spokesperson Taylor Rogers said in a statement. “More to come soon!”

A White House official noted that the draft is “outdated and no longer accurate” but did not specify which sections had been revised.

The decentralized nature of the nation’s power grid means that for the proposed agreement to be enforceable, grid operators, state regulators, and utility companies must agree to establish rules or draft contracts that reflect these principles.

This push emerges as utility providers, regulators, and members of Congress warn that the explosive growth of data centers—warehouse-sized buildings housing the powerful chips and servers required for AI development—could overwhelm regional power systems and inflate electricity bills for consumers already struggling with the cost of living.

Energy Secretary Chris Wright expressed awareness of these concerns on POLITICO’s energy podcast:

“People are skeptical. They think, ‘Oh my god, this is going to make the situation worse and drive up my energy prices.’ I understand their concerns. We are in a dialogue with all hyperscale developers to ensure they are not just a long-term force for lowering grid prices, but also a short-term force for halting current price increases.”

Major tech companies, known as “hyperscalers,” are building increasingly large data centers to process advanced AI computations. At the core of the agreement is the requirement that AI data center developers pay 100% of the costs for the new electricity generation needed to serve their facilities.

The agreement also stipulates that companies sign long-term electricity contracts to ensure other customers are not left with the bill if a data center fails. Similarly, tech companies would commit to paying the full cost of any current or future transmission upgrades required to connect new data centers to the grid.

In parallel, tech companies would agree to cooperate with federal, state, and local regulators to set power and transmission rates that are “neutral in every way” and ideally lower residential electricity prices in the regions where they operate. To prevent companies from outsourcing these impacts, the principles would apply not only to data centers they own but also to capacity they lease or operate from third parties.

Electricity costs are already on the rise, having outpaced the rate of inflation last year. Utilities have requested record-breaking price hikes, and government data predicts that costs will continue to climb in the coming years.

Microsoft recently made a similar set of commitments, stating it would pay more for the electricity serving its data centers, cover additional infrastructure costs, and reduce water consumption. Microsoft also announced it would no longer accept local tax breaks—a measure not included in the White House draft agreement.

In a post on Truth Social last month, Trump praised Microsoft’s announcement and indicated he was working with other tech companies to ensure “Americans do not pay the bill for their electricity consumption.”

The federal government predicts that energy demand from data centers could triple between 2025 and 2028. This surge is believed to be driving up prices in the power grid covering parts of 13 Mid-Atlantic and Midwestern states.

A 2025 Bloomberg analysis found that electricity prices rose in areas immediately surrounding data centers, while a separate 2025 report from the Harvard Law School Environmental and Energy Law Program found that consumers were bearing the infrastructure costs of serving these facilities.

However, the White House and industry allies argue that data centers are not the culprits and could actually be a significant force for lowering electricity prices. A report released last week by the Edison Electric Institute, a trade group representing investor-owned utilities, argued that costs have not increased in most areas where data centers are located.

According to the report, well-designed data center tariffs and agreements that place more responsibility on Big Tech for new energy generation and infrastructure could help lower consumer costs. However, this requires state utility regulators to draft tariff agreements and electricity contracts that fully account for the costs incurred by tech companies.

Secretary Wright highlighted two states experiencing the largest spikes in electricity demand due to data center developments without corresponding price increases. North Dakota has seen an approximately 35% increase in electricity demand over the last five years, and Georgia has implemented price freezes.

“Nominal electricity prices in these states have not increased. The real price of electricity has dropped significantly during this five-year period,” Wright said. “You will see more explanations. You have likely heard of Google’s deal in Georgia, which froze electricity prices for three years. Later this year, you will hear about agreements where major data center investments are announced in proportion to a decrease in electricity prices.”

Other companies maintain they are already covering their own costs. Meta, for instance, stated it covers all its energy costs and pointed to a study it commissioned last year showing that the clean energy projects it supports provide additional generation without increasing costs for taxpayers.

The draft also integrates data centers more directly into grid reliability planning. Signatories would commit to using non-critical backup generation at new and existing facilities, in coordination with grid operators, to support stability and reliability during emergencies.

Companies would also agree to voluntarily allow the curtailment of new data center loads when necessary to ensure reliable electricity for American households—a growing concern for grid operators facing rising peak demand and extreme weather events.

The concept of grid flexibility and backup power is gaining traction in policy circles. Last year, Texas lawmakers passed a landmark bill requiring large power users, such as data centers, to reduce power or disconnect from the grid during emergencies. Other states and grid operators are exploring similar programs. During last month’s winter storm, Wright also called on grid operators to secure backup power from data centers.

Beyond energy, the agreement aims to address local opposition in rapidly growing data center hubs. Hyperscalers would commit to developing or securing sufficient water resources to support new facilities and ensure no negative impact on local water availability or quality.

The agreement also encourages companies to establish AI education awareness programs in surrounding communities and public schools and to adopt best practices to mitigate noise, traffic, and other disruptions affecting nearby residential areas.

The deal could be significant for companies seeking federal assistance to speed up grid connections—a major hurdle for AI infrastructure projects. In the draft, the federal government commits to supporting the expedited connection of new data centers to the bulk power system that transmits high-voltage electricity across regions.