Federal and State Policymakers Target AI Data Centers as Electricity Costs and Grid Reliability Concerns Mount

The Trump administration is expected to call on major U.S. technology companies and data center developers to voluntarily commit to a compact designed to ensure power-needy data centers do not raise household electricity prices or undermine grid reliability.[1] The initiative comes amid a nationwide surge in energy demand, driven largely by the rapid proliferation of data centers that power the artificial intelligence (AI) boom.[2] Although the compact would be voluntary and details on monitoring or enforcement remain limited, it signals a clear expectation from federal policymakers that large technology firms “pay their own way” for the incremental costs their facilities impose on the grid.

The Trump administration’s draft pact would ask participating companies to voluntarily commit to, among other things:

• Paying 100% of the cost of new power generation required to serve their facilities.
• Funding any current or future transmission upgrades necessary to interconnect new data centers to the grid.
• Entering into long-term electricity contracts with utility companies to prevent cost-shifting to individual consumers.
• Using noncritical backup generation to support grid stability during emergencies.
• Allowing data center loads to be curtailed when necessary to ensure grid reliability.

Taken together, these commitments could materially increase capital and operating costs for large data centers and reshape how those costs are allocated among project sponsors, contractors, utilities, and end customers.

Importantly, these commitments would apply not only to data centers that participating companies own, but also to independently owned facilities from which participating companies lease capacity.[3] For context, many companies lease capacity — i.e., rent space, power, and cooling infrastructure from existing data center providers — to run AI workloads without building their own facilities or to secure capacity in power-constrained regions, a practice known in the industry as “colocation.”[4] The draft pact, however, suggests that participating companies could not avoid the pact commitments by leasing capacity from third-party providers.

If finalized in this form, the compact could drive material changes to colocation and wholesale data center agreements. Hyperscale tenants may push for provisions requiring landlords to: (1) demonstrate compliance with any applicable compact commitments; (2) indemnify tenants for noncompliance; and (3) share detailed information regarding interconnection costs, backup generation, and participation in curtailment programs. In turn, landlords may seek pass‑through mechanisms to allocate these incremental costs and risks back to tenants.

While the administration has not released any lists identifying specific companies that have agreed to the pact or been invited to participate, U.S. Secretary of Energy Christopher Wright has announced that the administration is “in dialogues with all of the hyperscale developers.”[5]

Concerns over rising electricity costs and grid reliability are also brewing at the state level. Indeed, more than 40 states have enacted or are considering data center-related legislation.[6] However, the mechanisms for addressing these challenges vary by state. For example, in June 2025, Texas lawmakers passed Senate Bill 6 (SB 6) to establish a regulatory framework to manage the rapid growth of large load customers (customers with demand of 75 MW or more).[7] Among other things, SB 6 requires large load customers to bear the costs of interconnecting to the grid. Notably, SB 6 also authorizes the Public Utility Commission of Texas to order emergency load reductions or disconnections during grid stress events. Similarly, Oregon passed the POWER Act last year, which requires large energy users (those using 20 MW or more) to buy power from state-regulated utilities for a minimum of 10 years and to pay for any additional infrastructure necessary to support their operations.[8]

By contrast, other states have adopted a study-first approach to data center regulation. For instance, California and New Jersey have enacted legislation that requires their respective utility commissions to study the impacts of data center electricity demand on retail customers and to identify policy alternatives to mitigate those impacts.[9] In New Jersey, one suggested policy alternative includes “the use of a special tariff to be applied to data centers within the State[.]”[10] These are just a few examples of recent state initiatives, but they underscore the importance for stakeholders to closely monitor the evolving regulatory landscape in the states in which they operate.

Attempts at legislating restrictions on power consumption by data centers are percolating at the federal level too. On February 11, U.S. Senators Josh Hawley and Richard Blumenthal introduced the Guaranteeing Rate Insulation from Data Centers Act (GRID Act).[11] The bipartisan bill seeks to ensure that data center electricity consumption does not increase individual consumers’ utility rates and that residential customers receive priority access to the grid. The GRID Act would also require new data centers with demand of 20 MW or more to obtain power from sources other than the electric grid, with a 10-year off-ramp for existing data centers to find alternative power sources.

Takeaways

Between the Trump administration’s draft voluntary compact and the pursuit of legislative initiatives to manage data center power consumption at both the state and federal levels, new risks may be emerging for data center owners, operators, and financiers.

While the full effects of these initiatives will not be understood for some time, it is not too early for stakeholders to start evaluating their potential risks and developing risk mitigation measures for their current and future projects. Project participants may want to consider, among other things:

• Who will bear the potential costs of new power generation and transmission upgrades necessary to interconnect data centers to the grid, and whether traditional cost‑sharing assumptions remain valid in light of emerging laws and policies.
• How additional infrastructure costs and curtailment risks are allocated in both construction contracts and capacity lease agreements, including flow‑down to service‑level agreements with end customers.
• Strategies to manage the risk of data center load reductions or disconnections, such as behind-the-meter solutions (e.g., natural gas, renewables, or nuclear power), and how those solutions interact with emerging state requirements and potential federal mandates.
• How evolving regulatory requirements may affect financing terms, including lenders’ views on curtailment risk and mandatory off-grid sourcing.
• Where to build future data centers amid policy divergence among states, including whether to prioritize jurisdictions that have adopted clear, long‑term tariff structures over those still in early “study” phases.

Of course, each project brings with it unique circumstances, challenges, and risks, which may only be compounded as project stakeholders navigate an uncertain policy environment. Troutman Pepper Locke attorneys continue to monitor these developments and are well-positioned to advise clients on emerging market trends, evolving federal and state policy requirements, and practical strategies to structure projects and contracts in this rapidly changing landscape.

As AI continues to drive unprecedented demand for data center capacity, developers, owners, operators, and capital providers must revisit their project delivery, contracting, permitting, and stakeholder strategies. Traditional risk allocation and deal structures are antiquating under energy constraints, interconnection delays, supply chain pressure, and evolving federal and state regulatory frameworks around air, water, and large load power usage.

Troutman Pepper Locke construction, energy, environmental, and real estate attorneys are hosting a three-part webinar series to explore how these dynamics are reshaping the data center landscape in 2026 and beyond. Each session will provide practical perspectives on how market participants are reallocating risk, structuring contracts, and positioning projects to remain bankable and scalable in a rapidly changing environment. Click here to register and to learn more.

[1] Sophia Cai et al., White House Eyes Data Center Agreements Amid Energy Price Spikes, Politico (Feb. 9, 2026, at 15:20 ET), https://www.politico.com/news/2026/02/09/trump-administration-eyes-data-center-agreements-amid-energy-price-spikes-00772024.

[2] See generally “Navigating Contractual Considerations in the AI Data Center Construction Boom” by Ryan Graham, Jamey Collidge, Jason Spang, and Christian Pirri; “EPA May Redefine ‘Begin Actual Construction’ in Permit Reform Intended to Expedite Construction of Emissions-Generating Developments” by Mack McGuffey, Jamey Collidge, Darby Koput, Christian Pirri, and Melissa Horne. This article frequently refers to data centers, which in essence, are warehouse-sized buildings that house powerful chips and severs for the development of AI technology. Id. However, it should not be overlooked that the issues addressed in this article are in large part equally applicable to other energy-intensive industries.

[3] Cai, supra note 1.

[4] Build or Lease? Inside the Billion-Dollar Dilemma Reshaping AI Infrastructure, Glob. Data Ctr. Hub (Oct. 15, 2025), https://www.globaldatacenterhub.com/p/build-or-lease-inside-the-billion.

[5] Politico Energy, Inside Energy Secretary Wright’s Playbook for Energy Dominance, at 22:43–23:14 (Apple Music, Feb. 9, 2026).

[6] State Data Center Policy 101, MultiState (last updated Dec. 4, 2025), https://www.multistate.us/resources/state-data-center-policy-101.

[7] S.B. 6, 89th Leg., Reg. Sess. (Tex. 2025) (amending or establishing, inter alia, Public Utility Regulatory Act (PURA) §§ 35.004, 37.0561, 39.002, 39.169 and 39.170). While SB 6 defines a “large load customer” as having a demand of 75 MW or greater, the Public Utility Commission of Texas, when implementing SB 6, has the ability to set a lower threshold. PURA § 37.0561(c).

[8] H.B. 3546, 83rd Leg. Assemb., Reg. Sess. (Or. 2025).

[9] 2025 Cal. Legis. Serv. Ch. 647 (S.B. 57); 2025 NJ Sess. Law Serv. Ch. 98 (Assemb. No. 5466).

[10] 2025 NJ Sess. Law Serv. Ch. 98 (Assemb. No. 5466).

[11] Guaranteeing Rate Insulation from Data Centers Act, 119th Cong. (2d Sess. Feb. 11, 2026) (proposed bill).