Skip to main content
Open questionOpen question

Open question: is AI training load interruptible, and who gets to say so

Training runs can pause where inference cannot, which would make a large share of AI load flexible and change what the grid must build. No regulator in this atlas has defined an interruptible class for AI workloads or priced one, and no operator has published a commitment that a system operator could rely on.

The verdict

Open question

No settled answer exists. The ledger poses the question, links the live debate, and does not answer it.

Key facts

What the sources say

Record ID
AIA-2026-0035
Kind
Open question
Jurisdiction
Global
Last verified
Added
  • China's East Data West Computing sorts workloads by latency tolerance, which is the closest thing in this atlas to a flexibility rule, and it is a siting policy rather than a dispatch right.
  • Britain's readiness test and Ireland's generation condition both treat large load as firm demand.
  • No instrument found by this build defines an interruptible tariff class specific to AI training.
  • Flexibility claimed in a press release is not flexibility a system operator can dispatch.

What it changes

For an operator, a utility or a community

An operator that can genuinely curtail training has an asset it is currently giving away, and a tariff that recognised it would change project economics. A utility should ask for the curtailment commitment in writing, with notice periods and a penalty, before counting on it in a plan. A community told a campus will be a good grid citizen should ask what it is contractually obliged to do when the system is short.

Sources

What this record was verified against

  1. Carbon Brief, Explainer, how China is managing the rising energy demand from data centresSecondary · 16 April 2025
  2. International Energy Agency, Energy and AI, executive summaryPrimary · 10 April 2025

Related

Records that sit beside this one

Cite this record

Free to reuse under CC BY 4.0, with attribution. The record ID AIA-2026-0035 is permanent and is never reused.

In a sentence
According to the GAGE AI Infrastructure Atlas (as of 15 September 2026), open question: is ai training load interruptible, and who gets to say so.
APA
GAGE (Global Academy of Generative-AI Education). (2026). Open question: is AI training load interruptible, and who gets to say so. AI Infrastructure Atlas. Retrieved 15 September 2026, from https://www.gage.academy/tools/ai-infrastructure-atlas/records/AIA-2026-0035-is-ai-training-load-interruptible
MLA
"Open question: is AI training load interruptible, and who gets to say so." AI Infrastructure Atlas, GAGE (Global Academy of Generative-AI Education), 15 September 2026, https://www.gage.academy/tools/ai-infrastructure-atlas/records/AIA-2026-0035-is-ai-training-load-interruptible.
Chicago
GAGE (Global Academy of Generative-AI Education). "Open question: is AI training load interruptible, and who gets to say so." AI Infrastructure Atlas. Last modified 15 September 2026. https://www.gage.academy/tools/ai-infrastructure-atlas/records/AIA-2026-0035-is-ai-training-load-interruptible.
Permalink
https://www.gage.academy/tools/ai-infrastructure-atlas/records/AIA-2026-0035-is-ai-training-load-interruptible

Last updated . Every record re verified . The ledger is checked monthly, first Monday, and the same day for any regulator decision on large loads or water.

Back to the full ledger, or every record for Global and every open question record.

GAGE briefings tell you which AI regulation deadlines are coming, what they actually require of you, and when a program opens.

Powered by GAGE, free EU AI Act tools from the Global Academy of Generative-AI Education
CallEmailHelp