Integrating AI Data Centers with the Power Grid
National Academy of Engineering, The Bridge (LBNL authors) · Jessica Granderson, Ian M. Hoffman et al.
This LBNL article addresses demand flexibility for AI data centers facing grid constraints. It sets out four flexibility mechanisms, utility tariff approaches that encourage flexibility, and operator motivations, stressing that flexibility complements but does not replace new generation.
On this page
Overview
The article was published in The Bridge of the National Academy of Engineering and its authors are from Lawrence Berkeley National Laboratory (they include Jessica Granderson, Ian M. Hoffman and Natalie Mims Frick). It is dated 2026 and carries DOI 10.20357/B7X61S.
According to the abstract, there are four flexibility mechanisms: computational load flexibility, flexible use of core facility infrastructure, energy storage and onsite generation. They represent different ways in which a data center can adjust the power it draws from the grid over time.
On the utility side, the tools include voluntary interruptible service riders, mandated flexibility requirements and faster interconnection for flexible loads. On the operator side, the drivers are faster interconnection, regulatory compliance and reduced community opposition. The article also relays a projection that US data center energy use will reach two to three times its 2023 level by 2028.
The article's central caution is that flexibility complements but does not replace new generation. Flexibility is therefore not the sole answer to the capacity problem, but a tool that can speed up interconnection and ease pressure on the grid.
Key points
- Four mechanisms: computational load flexibility, flexible use of core facility infrastructure, energy storage and onsite generation.
- Utility tools: voluntary interruptible service riders, mandated flexibility requirements and faster interconnection for flexible loads.
- Operator drivers: faster interconnection, regulatory compliance and reduced community opposition.
- US data center energy use is projected to reach two to three times its 2023 level by 2028.
- Flexibility complements but does not replace new generation.
- Published in The Bridge (National Academy of Engineering), LBNL authors, 2026.
Using it in design
- Expose flexibility capability per design (curtailable MW, BESS duration, on-site generation) as outputs for interconnection applications.
- Treat a flexible-load commitment as an input to interconnection-timing negotiations.
- Do not assume flexibility is the only answer to a capacity shortfall; assess the need for new generation separately.
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
Review your power design in SoSo CAD
Inspect supply paths, equipment loads and UPS failure scenarios in one model.
Request a demo
Who should read it
Electrical designers planning grid interconnection and energy strategy, operators, and policy analysts.
Publication details
- Authors
- Jessica Granderson; Ian M. Hoffman; Billie Holecek; Eliot Crowe; Sarah Josephine Smith; Natalie Mims Frick
- Publisher
- National Academy of Engineering, The Bridge (LBNL authors)
- Year
- 2026
- Document language
- English
- Source type
- Expert / community
- Access
- Free to read
- Licence
- Copyright retained; citation only — the publisher’s terms apply to the original document
Original licence note
LBNL (contractor-operated DOE laboratory) resource; not US-Gov public domain. Cited only.
Citation
Jessica Granderson; Ian M. Hoffman; Billie Holecek; Eliot Crowe; Sarah Josephine Smith; Natalie Mims Frick (2026). Integrating AI Data Centers with the Power Grid. National Academy of Engineering, The Bridge (LBNL authors). https://doi.org/10.20357/B7X61S
Events on this subject
- Electricity Eurasia 2027
- ELECO 2026 Elektrik-Elektronik ve Biyomedikal Mühendisliği Konferansı
- ICCI 2027 – Uluslararası Enerji ve Çevre Fuarı ve Konferansı
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.