US Government Drafts Voluntary Compact with Tech Giants on AI Data Center Energy
The United States government is actively pursuing a formal legal commitment from leading technology corporations regarding the substantial electricity consumption of data centers that power artificial intelligence services. According to a recent report citing two anonymous Trump administration officials, the initiative aims to secure public pledges from companies including OpenAI, Google, Amazon, and Meta to adhere to a new compact governing the rapid expansion of AI infrastructure.
Microsoft's Precedent and the White House Proposal
This development follows closely on the heels of Microsoft's own set of commitments, announced just weeks prior. Microsoft pledged to pay more for electricity serving its data centers, cover additional infrastructure costs, and reduce water consumption. Notably, Microsoft also stated it would no longer accept local tax breaks, a provision absent from the current White House draft compact. Last month, President Donald Trump confirmed via a Truth Social post that Microsoft was collaborating with other tech firms to ensure Americans do not bear the burden of their power consumption.
A draft of the compact, obtained by Politico, outlines commitments designed to ensure that energy-intensive data centers do not increase household electricity prices, strain water supplies, or compromise grid reliability. It emphasizes that the companies driving demand must also shoulder the costs of building new infrastructure. Framed as a voluntary agreement between President Trump and major U.S. tech companies and data center developers, the pact could bind entities like OpenAI, Microsoft, Google, Amazon, and Meta to a broad set of energy, water, and community principles.
Addressing Energy and Grid Concerns
The compact represents an ambitious effort to shape the environmental and economic footprint of AI infrastructure without imposing direct regulation. It emerges amid rising concerns that the enormous energy demand from data centers could drive up electricity prices, posing a political challenge for an administration that has supported their rapid development. The White House recently appealed to the Mid-Atlantic Energy Grid Operator to lower electricity prices, highlighting these worries.
In a statement to Politico, White House spokesperson Taylor Rogers said, "As President Trump announced weeks ago, top tech companies are working with the President to 'pick up the tab' for their power consumption as they build data centers. More to come soon!" However, another White House official noted that the draft "is outdated and no longer accurate," without specifying which parts have changed.
The decentralized nature of the U.S. electricity grid means that grid operators, state regulators, and utilities would need to agree on rules or craft contracts to implement aspects of the compact. Energy Secretary Chris Wright, in a recent interview, acknowledged public skepticism, stating, "People are sceptical. 'Oh my gosh, this is going to further add insult to injury and drive up my energy prices.' I understand their concerns. We are in dialogue with all the hyperscale developers about not only being a long-term force to drive down electricity prices on the grid, but to also be a short-term force to stop the existing price rises."
Core Requirements and Company Responses
At its core, the compact requires AI data center developers to pay 100% of the cost for new power generation needed for their facilities. It also calls for companies to sign long-term electricity contracts to prevent other customers from footing the bill if a data center fails. Additionally, companies would commit to covering the full cost of any transmission upgrades required to connect new data centers to the grid.
In parallel, tech companies would agree to work with regulators to establish power and transmission rates that, "in every manner possible," protect and ideally reduce residential electricity prices in areas where data centers operate. These principles would apply not only to owned data centers but also to leased or operated capacity. Electricity costs have been rising, even outpacing inflation over the past year.
Some companies, such as Meta, have stated they already cover their energy costs. Meta commissioned a study last year finding that its supported clean energy projects add generation without raising costs for ratepayers.
Grid Reliability and Community Impact
The draft also integrates data centers into grid reliability planning. Signatories would commit to using noncritical backup generation at facilities, in coordination with grid operators, to support stability during emergencies. Companies would voluntarily agree to curtail new data center load when necessary to ensure reliable power for households.
Beyond energy, the compact aims to address local opposition in data center regions. Hyperscalers would commit to being "water positive," developing or procuring sufficient water supplies to support new facilities without negatively impacting local water availability or quality. The agreement also calls for companies to establish AI awareness programs in surrounding communities and schools and adopt practices to mitigate noise, traffic, and other disruptions affecting residential neighborhoods.
Potential Benefits and Challenges
The pact could be significant for companies seeking federal assistance to accelerate grid interconnections, which are often a bottleneck for AI infrastructure projects. Under the draft, the U.S. government would commit to supporting the accelerated interconnection of new data centers to the bulk power system, which delivers high-voltage power across regions.
This initiative, planned for rollout at a White House event, has yet to be formally announced, and it remains unclear which companies have agreed or been invited to participate. As discussions continue, the compact highlights the growing intersection of technology, energy policy, and community relations in the era of artificial intelligence.
