PJM will require new data centers above 50 MW to provide dedicated generation or accept priority curtailment during grid shortages, materially affecting planned AI-compute projects in its territory.

state: expiredheat: lowuncertainty: mediumconvergesscott: mediumai-infrastructure data-center-power grid-curtailment inference-economicsPJM Interconnection

What is this?

PJM Interconnection, the grid operator serving 13 states and Washington, D.C., has proposed a reliability regime for new large electrical loads of at least 50 MW, including AI data centers. Beginning in June 2027, qualifying facilities that do not bring dedicated generation or new capacity would be curtailed ahead of broader emergency load-management measures when supply is short. The supplied snippets differ on whether this is a finalized decision or still a proposal, so its approval status and exact obligations are not firmly established here.

Why it matters to Scott

PJM’s proposed curtailment regime makes interruption tolerance an infrastructure requirement for some large AI deployments, strengthening Scott’s existing case for batchable workloads, stateless workers, durable external state, and checkpointed resumption. It could affect how he designs or argues for resilient agent systems, although the supplied hits do not show him operating planned 50 MW-scale facilities and the radar does not yet track this specific PJM development.

ip:framework.the-lane-doctrineip:concept.batch-processingip:framework.long-running-agentsip:concept.durable-external-stateip:concept.checkpoint-disciplineip:concept.stateless-executionradar:concept.ai-infrastructureradar:concept.inference-economicsradar:openai-power-trading-hedgingradar:spotwarp-spot-gpu-failoverradar:async-bulkhead-llm-admission-control

queries asked of Scott's wikis

AI compute economics under power constraints
dedicated generation for data centers
curtailable workloads and resilient inference
energy availability as an AI infrastructure bottleneck
distributed or local inference versus hyperscale compute
compute sovereignty and grid dependence

Measured heat

no measured readings yet — the hourly heat pass fills this in

How the heat travelled

no chain yet — the hourly chain pass fills this in

Evidence (4) — ⭐ canonical anchor

source	object	author	score	comments
🟠 reddit	America's largest grid wants to cut power to new data centers first during shortages — 50MW-plus data centers must bring their own electricity generation to avoid shutoffs OpenAI	KeanuRave100	219	31
🟧 echo.other ⭐	PJM’s primary proposal defines Large Loads as 50 MW or more, requires new loads without Bring Your Own New Capacity to reduce demand or swit	PJM Interconnection	—	—
🟠 reddit	UBS models $4.1T in AI infrastructure spending by 2028 - it assumes the power just shows up artificial	Servola-Journal	2	3
🟧 hn	FERC approves SPP 'topology optimization' plan for cutting grid congestion	toomuchtodo	1	0

Interpretation history

2026-08-31T20:39:32Z
Repeated staleness checks have produced only engagement churn, with no PJM approval, finalized terms, or named project impact. Expire this dormant episode and reopen only on formal policy action or a consequential industry response.
2026-08-29T20:24:42Z
The latest reobservation is another negligible engagement-only change and adds nothing on PJM approval, final terms, or affected projects. The proposal remains credible but dormant; further review should be keyed to formal policy action or a named industry response rather than routine discussion churn.
2026-08-27T19:49:06Z
Only minor, repetitive engagement arrived; PJM approval status, final implementation terms, and project-level effects remain unknown. The proposal is still credible but dormant, so further review should wait for a formal policy action or named industry response.
2026-08-25T19:43:20Z
No substantive evidence arrived before the staleness review; the tiny engagement change does not clarify whether PJM adopted the regime or whether any project is affected. The credible first-party proposal remains dormant pending a formal policy-status change or named industry response.
2026-08-23T19:34:57Z
The refreshed comments only repeat the broader power-constraint framing and add no evidence of PJM approval, finalized terms, or impact on a named data-center project. The PJM-specific case remains dormant pending a formal policy-status change or consequential industry response.
2026-08-23T01:29:24Z
The SPP congestion-management approval shows another grid operator acting on transmission constraints, but it does not clarify PJM’s proposal status, terms, or effects on any data-center project. The PJM-specific hypothesis remains dormant pending a formal policy change or named project response.
2026-08-23T01:22:57Z
evidence attached: hn.story.49405423 — A concrete grid-congestion intervention materially contextualizes the infrastructure constraints and mitigation options facing large AI data centers.
2026-08-22T16:31:54Z
The added discussion strengthens the broader view that grid access can constrain AI buildout, making PJM’s proposal more worth tracking, but it does not independently confirm adoption, implementation terms, or project-level effects. The case advances to watching while remaining dependent on a formal PJM status change.
2026-08-22T16:23:15Z
evidence attached: reddit.post.1vvfxyq — Substantive evidence that grid interconnection queues, rather than GPUs alone, are becoming a binding constraint on AI data-center expansion.
2026-08-21T04:31:16Z
Repeated engagement updates add no evidence on approval, implementation terms, or project impact. The first-party proposal remains credible but dormant pending a formal PJM policy-status change or consequential industry response.
2026-08-19T03:32:01Z
The refreshed comments remain generic reactions and add no evidence that PJM approved the proposal, fixed its implementation terms, or affected a specific data-center project. The case’s meaning is unchanged and still depends on a policy-status change or consequential industry response.
2026-08-18T21:39:24Z
The refreshed discussion remains generic reaction to the known first-party proposal and adds no evidence of approval, implementation, or impact on a specific data-center project. The case still awaits a policy-status change or independent consequential response.
2026-08-18T17:32:02Z
Refreshed comments remain generic reactions to the known first-party proposal and add no evidence of approval, implementation, or impact on a specific data-center project. The case still awaits an independent consequential development.
2026-08-18T15:43:44Z
Refreshed discussion adds only public reactions and a restatement of the proposal; it does not establish approval, implementation details, or effects on specific data-center projects. The case remains a credible first-party proposal awaiting an independent consequential development.
2026-08-18T13:45:21Z
No new substantive evidence arrived; the modest Reddit engagement increase only repeats the already-grounded first-party proposal and does not clarify adoption status or implementation details.
2026-08-18T13:30:35Z
grounded: converges/medium — PJM’s proposed curtailment regime makes interruption tolerance an infrastructure requirement for some large AI deployments, strengthening Scott’s existing case
2026-08-18T13:27:41Z
origin walked (codex/luna, conf 0.98): anchor reddit.post.1vro1o1 -> echo.other.1ca08f1efd by PJM Interconnection
2026-08-18T13:25:59Z
case created — The reported PJM policy is a concrete infrastructure constraint whose adoption and effects on data-center development can be tracked.

Decision trace

09-01 06:39expireRepeated staleness checks have produced only engagement churn, with no PJM approval, finalized terms, or named project impact. Expire this dormant episode and reopen only on formal policy action or a
09-01 06:39alert_silentThe new delta is engagement-only and does not change the proposal’s status or consequences; no briefing attention is warranted.
09-01 06:39alert_routeThe new delta is engagement-only and does not change the proposal’s status or consequences; no briefing attention is warranted.
09-01 06:39drop_targetsquiet through full ladder or over cap 8
08-30 06:24repriceThe latest reobservation is another negligible engagement-only change and adds nothing on PJM approval, final terms, or affected projects. The proposal remains credible but dormant; further review sho
08-30 06:24alert_silentNo consequential new delta occurred; a one-point engagement change does not alter the policy’s status or implications, so this can wait for routine monitoring.
08-30 06:24alert_routeNo consequential new delta occurred; a one-point engagement change does not alter the policy’s status or implications, so this can wait for routine monitoring.
08-28 05:49repriceOnly minor, repetitive engagement arrived; PJM approval status, final implementation terms, and project-level effects remain unknown. The proposal is still credible but dormant, so further review shou
08-28 05:49alert_silentThe new delta is engagement-only and adds no consequential evidence; routine monitoring is sufficient until PJM changes the proposal’s status or an affected project responds.
08-28 05:49alert_routeThe new delta is engagement-only and adds no consequential evidence; routine monitoring is sufficient until PJM changes the proposal’s status or an affected project responds.
08-26 05:43repriceNo substantive evidence arrived before the staleness review; the tiny engagement change does not clarify whether PJM adopted the regime or whether any project is affected. The credible first-party pro
08-26 05:43alert_silentThere is no new consequential delta beyond negligible engagement, so this can wait for routine review focused on PJM approval, implementation terms, or project-level impact.
08-26 05:43alert_routeThere is no new consequential delta beyond negligible engagement, so this can wait for routine review focused on PJM approval, implementation terms, or project-level impact.
08-24 05:34repriceThe refreshed comments only repeat the broader power-constraint framing and add no evidence of PJM approval, finalized terms, or impact on a named data-center project. The PJM-specific case remains do
08-24 05:34alert_silentMinor engagement and generic discussion do not change the tracked policy’s status or consequences, so this can wait for routine review.
08-24 05:34alert_routeMinor engagement and generic discussion do not change the tracked policy’s status or consequences, so this can wait for routine review.
08-24 05:21sensor_dirtycomment_update
08-23 11:29repriceThe SPP congestion-management approval shows another grid operator acting on transmission constraints, but it does not clarify PJM’s proposal status, terms, or effects on any data-center project. The
08-23 11:29alert_silentThis is an adjacent development involving a different grid operator and mechanism, not a consequential change to the tracked PJM regime; it can wait for routine review.
08-23 11:29alert_routeThis is an adjacent development involving a different grid operator and mechanism, not a consequential change to the tracked PJM regime; it can wait for routine review.
08-23 11:23alert_silentFERC’s approval of SPP topology optimization concerns a different grid operator and congestion-management mechanism. It neither advances nor materially changes PJM’s proposed 50 MW large-load curtailm
08-23 11:23alert_routeFERC’s approval of SPP topology optimization concerns a different grid operator and congestion-management mechanism. It neither advances nor materially changes PJM’s proposed 50 MW large-load curtailm
08-23 11:22attachA concrete grid-congestion intervention materially contextualizes the infrastructure constraints and mitigation options facing large AI data centers.
08-23 11:22propose_attachA concrete grid-congestion intervention materially contextualizes the infrastructure constraints and mitigation options facing large AI data centers.
08-23 02:31repriceThe added discussion strengthens the broader view that grid access can constrain AI buildout, making PJM’s proposal more worth tracking, but it does not independently confirm adoption, implementation
08-23 02:31alert_silentThe new evidence is contextual commentary rather than a consequential PJM policy or industry response, so it can wait for the next briefing; alert only on approval, a binding filing, finalized impleme
08-23 02:31alert_routeThe new evidence is contextual commentary rather than a consequential PJM policy or industry response, so it can wait for the next briefing; alert only on approval, a binding filing, finalized impleme
08-23 02:24alert_silentThe attached Reddit post adds commentary but no consequential change beyond PJM’s already captured primary proposal. The proposal is a substantive infrastructure signal, but it is not shown here as a
08-23 02:24surface_candidateThe attached Reddit post adds commentary but no consequential change beyond PJM’s already captured primary proposal. The proposal is a substantive infrastructure signal, but it is not shown here as a
08-23 02:24alert_routeThe attached Reddit post adds commentary but no consequential change beyond PJM’s already captured primary proposal. The proposal is a substantive infrastructure signal, but it is not shown here as a
08-23 02:23attachSubstantive evidence that grid interconnection queues, rather than GPUs alone, are becoming a binding constraint on AI data-center expansion.
08-23 02:23propose_attachSubstantive evidence that grid interconnection queues, rather than GPUs alone, are becoming a binding constraint on AI data-center expansion.
08-21 14:31repriceRepeated engagement updates add no evidence on approval, implementation terms, or project impact. The first-party proposal remains credible but dormant pending a formal PJM policy-status change or con
08-21 14:31alert_silentNo new consequential delta occurred; renewed attention is only repetitive amplification and can wait for a formal decision, filing, or affected-project response.
08-21 14:31alert_routeNo new consequential delta occurred; renewed attention is only repetitive amplification and can wait for a formal decision, filing, or affected-project response.
08-21 03:21sensor_dirtyengagement_update
08-21 00:21sensor_dirtyengagement_update
08-20 21:21sensor_dirtyengagement_update
08-20 20:21sensor_dirtyengagement_update
08-20 18:21sensor_dirtyengagement_update