Technical guidePDF2025

Request for Information on Artificial Intelligence Infrastructure on DOE Lands (Federal Register, April 7, 2025)

Original title DOE Request for Information on Artificial Intelligence Infrastructure on DOE Lands (90 FR, April 7, 2025)

U.S. Department of Energy / Federal Register · U.S. Department of Energy, Office of Policy

A US Department of Energy request for information (RFI) gauging industry interest in building AI data centres and energy infrastructure on DOE land. It lists sixteen candidate sites and seven categories of questions, and it is not a solicitation.

U.S. public domain — U.S. Government work; free to reuse

On this page

Overview

Published in the Federal Register of 7 April 2025 (Vol. 90, No. 65), the RFI says DOE wants AI infrastructure construction at select sites to begin by the end of 2025 and operation to start by the end of 2027. AI infrastructure covers data centres, their specialised IT equipment and cooling, and the energy supply: generation, transmission including substations, and storage. Responses were requested by 7 May 2025, and grid operators were also asked about existing infrastructure and co-location with new generation.

DOE points to its national laboratories, supercomputers and existing energy infrastructure as advantages. Land rights could be granted through a long-term ground lease or an easement. Realty terms could require non-federal parties to bear all responsibility for costs and liabilities of construction, operation and necessary infrastructure upgrades, including costs to transmission providers; in-kind contributions may count toward costs. The developer would be responsible for state and local electricity requirements, including interconnection.

Sixteen candidate sites are listed in no particular order: Idaho National Laboratory, the Paducah and Portsmouth Gaseous Diffusion Plants, Argonne, Brookhaven, Fermilab, NETL, NREL, Oak Ridge, PNNL, PPPL, Los Alamos, Sandia, Savannah River, Pantex and the Kansas City National Security Campus. The appendices give figures such as Paducah's design for up to 3 GW (MISO) and 30 million gal/day of water, Portsmouth's 2.2 GW (PJM) and 40 million gal/day, a 110-acre Argonne site for a 1,000 MW AI data park, and NREL Flatirons capacity for a 100 MW data centre.

The questions fall into seven categories: interest and site preference; site information for data-centre design and technology (including siting in hot humid places with little water, innovative cooling, high-efficiency power electronics, and data-centre waste heat in district heating); on-site energy (nuclear, enhanced geothermal, fuel cells, carbon capture, storage); off-site energy and transmission capacity, including whether flexible operation could speed interconnection; financial and contractual terms; benefits and collaboration; and economic opportunities. DOE stresses that the notice is for information and planning only and that no land-use decisions have been made.

Key points

  1. Target: construction at select DOE sites begins by the end of 2025, with operation by the end of 2027.
  2. Responses were due 7 May 2025; the RFI is not a solicitation.
  3. Land rights could be a long-term ground lease or an easement.
  4. Realty terms could make the non-federal party bear all construction, operation and infrastructure-upgrade costs and liabilities; in-kind contributions may count toward costs.
  5. Sixteen candidate sites are listed in no particular order and the list is not exhaustive.
  6. Paducah: designed for up to 3 GW (MISO) and 30 million gal/day of water; Portsmouth: 2.2 GW (PJM) and 40 million gal/day.
  7. Argonne: a 110-acre site for a 1,000 MW AI data park with an early operation target of 2028; Brookhaven: more than 90 acres; NREL Flatirons: capacity for a 100 MW data centre.
  8. The seven question categories are interest, design and technology, on-site energy, off-site energy and transmission, financial terms, benefits and collaboration, and economic opportunities.
  9. DOE explicitly asks whether flexible data-centre operation could enable faster interconnection.

Using it in design

  • For federal-site scenarios, keep water (million gal/day), grid market (MISO, PJM) and area (acres) as separate fields in site records.
  • Tie the cooling technology choice to the site's water constraint; the RFI explicitly asks about siting in hot, humid and water-scarce places.
  • Model flexible-load operation as a design option against delayed interconnection capacity.
  • Evaluate data-centre waste heat use in district heating or neighbouring facilities as its own siting criterion.
  • Include non-federal cost items (transmission, environmental review, infrastructure upgrades) in the capital budget.

Monthly newsletter · free

Keep your reading list current

Once a month: new library sources, changes to Turkish regulations and upcoming data centre events.

SoSo CAD

Bring standards pre-checks into your project

Review model-based TS EN 50600 and Tier pre-checks, missing inputs and items requiring engineering review in SoSo CAD.

Request a demo
Selected rule packs and findings in the SoSo CAD standards panel

Who should read it

Investors, developers and grid operators evaluating AI data-centre or associated energy projects on DOE land.

Document contents

  1. I. Background
  2. II. Purpose
  3. III. Sites Under Consideration
  4. IV. DOE Realty Agreement Terms and Conditions
  5. V. RFI Categories and Questions
  6. Category 1: Interest in Solicitation
  7. Category 2: Site Information and Considerations for Data Center Design and Technology
  8. Category 3: On-Site Energy Development
  9. Category 4: Off-Site Energy and Transmission Capacity
  10. Category 5: Financial and Contractual Considerations
  11. Category 6: Benefits and Collaboration Opportunities
  12. Category 7: Economic Opportunities and Considerations
  13. VI. Confidential Business Information
  14. VII. Disclaimer
  15. IX. Appendices: site summaries

Related standards

  • EO 14318
  • EO 14141

Read the document

doe-rfi-ai-infrastructure-doe-lands-2025.pdf

PDF · 5.9 MB · English

U.S. public domain — U.S. Government work; free to reuse

The document is provided unaltered because its licence permits redistribution.

Publication details

Authors
U.S. Department of Energy, Office of Policy
Publisher
U.S. Department of Energy / Federal Register
Year
2025
Document number
FR Doc. 2025-05936
Document language
English
Source type
Government / public
Access
Open licence — downloadable
Licence
U.S. public domain — U.S. Government work; free to reuse
Licence evidence
www.copyright.gov
Original licence note

Official Federal Register document (Office of the Federal Register / GPO govinfo); U.S. government work not subject to copyright (17 U.S.C. 105). The file is the Federal Register page range containing the notice, which also carries adjacent notices from other agencies.

Citation

U.S. Department of Energy, Office of Policy (2025). DOE Request for Information on Artificial Intelligence Infrastructure on DOE Lands (90 FR, April 7, 2025). U.S. Department of Energy / Federal Register. FR Doc. 2025-05936. https://www.govinfo.gov/content/pkg/FR-2025-04-07/pdf/2025-05936.pdf

All content and rights belong to the respective authors and publishers. SoSo CAD provides only the citation, its own summary and a link to the source, and is not responsible for the sources' content.