PJM just handed hyperscalers and data center developers a new kind of term sheet. In the old world, getting “in the queue” was the long pole. In the new world, getting energized comes with a condition: either you bring new generation (or equivalent) to the table, or you accept that your megawatts can be curtailed earlier than the public when the system is tight.
This is why “Connect and Manage” matters. It turns AI load growth from a pure real estate and substation problem into an explicit reliability product. The grid is essentially saying: you can connect faster, but only if you agree to be interruptible under defined conditions. The throttle is no longer a rumor. It is the contract.
[wnr_snapshot title=”PJM ‘Connect and Manage’ – at a glance” items=”Rule shift|Connect faster, accept curtailment|New large loads face earlier curtailment or BYO-power pathways;Threshold|50 MW+ at one POI|PJM large-load definition for fast-path work;Capacity market stress|$333.44/MW-day cap|2027/28 BRA cleared at the cap;Reliability gap|6,623 MW short|Cleared supply below requirement;Data center cost share|~$6.5B (40%)|Share of last auction cost (IMM via reporting);Load growth signal|~32 GW by 2030|PJM estimate of peak growth mostly from data centers”]
The new contract: megawatts now come with a curtailment clause
PJM’s board package (born out of the 2025 Critical Issue Fast Path process) is explicit about the direction. PJM wants two parallel on-ramps for large loads: (1) bring your own new generation (or otherwise prove you are not worsening the adequacy math), or (2) connect under a “connect and manage” framework that is subject to earlier curtailment when the system needs it. PJM also indicated it intends to design curtailment allocations based on how much a given shortfall contributes to reserve-margin pressure, and to develop load-shed prioritization and allocation through 2026.
That last part is the key: this is not “we will study you.” This is “you are agreeing to be a controllable load.”
Why PJM is doing this now: capacity prices hit the cap and the math still broke
PJM’s December capacity auction for the 2027/2028 delivery year cleared at the $333.44/MW-day cap across the footprint and still landed 6,623 MW short of PJM’s reliability requirement. That is the market screaming “tight,” with the operator still saying “not enough.”
The Independent Market Monitor’s data center accounting is what turns this into a political story. In the last auction, data center load accounted for about $6.5B of the ~$16.4B cleared cost (roughly 40%). Across the last three capacity auctions, costs related to data center forecasts above existing data center loads totaled ~$21.3B, or 45% of the ~$47.2B cleared capacity cost (Monitoring Analytics figures as summarized in reporting). That is a giant spotlight on the difference between “forecasted” load and “built” load, and it is why PJM is simultaneously tightening forecasting processes while hardening interconnection rules.
[wnr_math title=”Quick math: how reserve margin and capacity prices turn into real dollars” items=”Capacity price cap (2027/28)|$333.44 per MW-day|PJM BRA clearing price at cap;Annualized capacity cost|$121,705.60 per MW-year|333.44 x 365 (derived);Capacity as $/MWh equivalent|~$13.89 per MWh|$121,705.6 / 8,760 (derived);100 MW campus capacity line|~$12.17M per year|100 x $121,705.6 (derived);Reliability shortfall scale|6,623 MW|Enough to equal ~66 x 100-MW campuses (order-of-mag);PJM data center planning default|~70% utilization|Used in PJM planning assumptions (reported);30 GW at 70% utilization|~184 TWh per year|30,000 MW x 0.70 x 8,760 (derived)”]
What “Connect and Manage” does to capex and siting
Once curtailment becomes a standard condition for new load, the winners are the developers who can either self-supply or credibly flex. That means the project’s electrical design starts to look less like a passive consumer and more like a hybrid resource: on-site generation, battery ride-through, workload shifting, or contractual curtailment obligations tied to grid conditions.
It also makes co-location more rational. If front-of-meter interconnection timelines are stretching, the economic appeal of building next to an existing generator rises, even if regulators get twitchy about cost shifting. Google has publicly described transmission interconnection delays as its number-one challenge, with at least one utility citing “12 years” just to study interconnection timing (per Reuters). That kind of timeline forces alternative architectures.
[wnr_table title=”Three ways to ‘plug in’ under the new reality” headers=”Path|What you get|What you give up|Who wins” rows=”Traditional firm service|Highest firmness and standard rights|Time (studies/upgrades) and long lead equipment timelines|Workloads that cannot curtail and can wait;Connect and Manage (curtailable)|Faster connection pathway under defined conditions|Earlier curtailment risk; operational constraints become contract terms|Flexible loads; campuses with storage/backups and orchestration;Bring-your-own new generation / co-location|More control of energization and adequacy math|Capex, permitting, fuel/tech risk, regulatory scrutiny|Hyperscalers with balance sheet + developers near plants” notes=”PJM’s board package emphasizes BYO-new-generation or earlier curtailment frameworks for large load additions; co-location rules are also moving at FERC.” ]
FERC is building the parallel toolbox for co-location
Separate from PJM’s board package, FERC has been forcing clarity on “co-located generation + load” arrangements in PJM. Legal summaries of FERC’s order describe new transmission service concepts including an interim, non-firm network-style product (Interim NITS) that allows service before upgrades are complete but requires curtailment in advance of system emergency conditions, plus “contract demand” style products that limit how much the load can withdraw from the grid.
Translation: even the regulator is converging on the same design pattern. Speed to power is increasingly traded for firmness.
[wnr_table title=”Co-location services (FERC-directed concepts) and what they imply” headers=”Service concept|Core idea|Why it matters for AI load” rows=”Interim NITS (temporary, non-firm)|Get connected before full upgrades, accept curtailment risk|Turns time-to-power into a curtailment trade;Firm Contract Demand|Buy a defined firm MW block, no withdrawals above it|Formalizes ‘grid as backup’ instead of grid as full supply;Non-Firm Contract Demand|Flexible, non-firm withdrawals when available|Useful for maintenance/outage coverage if on-site generation carries base load” notes=”Summarized from legal analyses of FERC’s PJM co-located load order; PJM compliance details and rates are still being developed.” ]
Historical context: PJM’s load curve just flipped into a new regime
PJM’s own forecasts show why the operator is moving fast. In its 2026 long-term load outlook, PJM reported sharply higher growth rates versus its prior forecasting regime, with annualized 10-year growth rates in the multiple-percent range (summer and winter peaks both meaningfully higher than the sub-1% era). PJM has also stated that data centers dominate its near-term load growth expectations, with an estimate of roughly 32 GW of peak load growth from 2024 to 2030, almost entirely attributable to data centers.
This is the real “AI electrification” story: a historically slow-moving grid operator is being asked to absorb utility-scale load additions on tech timelines, while also keeping a one-in-10 reliability standard and containing consumer backlash from capacity price spikes.
Bottom line
PJM is turning new AI load into a choice between capex and curtailment. If you want speed, you accept interruptibility. If you want firmness, you wait (or you build and finance new supply). Either way, the era of “just get in the queue” is ending. The new currency is controllability.
Sources (primary)
• Reuters (Jan 16, 2026): PJM plan requiring BYO power or early curtailment; AI/data center demand surge — https://www.reuters.com/legal/litigation/pjm-unveils-plan-tackle-ai-driven-power-demand-surge-2026-01-16/
• PJM Inside Lines (Jan 16, 2026): Board outlines actions; Connect and Manage framing; quotes — https://insidelines.pjm.com/pjm-board-outlines-plans-to-integrate-large-loads-reliably/
• PJM Board decisional letter (Jan 16, 2026, PDF): CIFP-Large Load Additions package; 50 MW+ definition; curtailment allocation intent; assumptions — https://www.pjm.com/-/media/DotCom/about-pjm/who-we-are/public-disclosures/2026/20260116-pjm-board-letter-re-results-of-the-cifp-process-large-load-additions.pdf
• PJM (Dec 17, 2025): 2027/2028 BRA results; $333.44/MW-day; 6,623 MW short; $16.4B cleared supply value — https://insidelines.pjm.com/pjm-auction-procures-134479-mw-of-generation-resources/
• PJM (Jan 14, 2026): 2026 long-term load forecast highlights; regime shift in growth rates — https://insidelines.pjm.com/2026-long-term-load-forecast-report-released/
• PJM (2026 Load Forecast report, PDF): peak and energy forecast tables; zonal adjustments — https://www.pjm.com/-/media/DotCom/library/reports-notices/load-forecast/2026-load-report.pdf
• Utility Dive (Jan 7, 2026): Data centers share of capacity costs; Monitoring Analytics report summary — https://www.utilitydive.com/news/data-centers-pjm-capacity-auction/808951/
• Monitoring Analytics (Nov 10, 2025, PDF): IMM executive summary on large load additions; BYO-new-generation framing — https://www.monitoringanalytics.com/reports/presentations/2025/IMM_CIFP_Executive_Summary_20251110.pdf
• Reuters (Jan 14, 2026): Google on transmission delays and co-location as a strategy — https://www.reuters.com/business/google-says-us-transmission-system-is-biggest-challenge-connecting-data-centers-2026-01-14/
• K&L Gates (Jan 15, 2026): Summary of FERC order directing new transmission service options for co-located load (Interim NITS / Contract Demand) — https://www.klgates.com/FERC-Orders-PJM-to-Reform-Tariff-for-Co-Located-Generation-and-Load-1-15-2026
• Perkins Coie (Jan 13, 2026): Summary of FERC co-located load order and service concepts — https://perkinscoie.com/insights/update/ferc-orders-pjm-revise-tariff-accommodate-co-located-load-arrangement
• DOE (Jan 16, 2026): Emergency procurement auction push and cost allocation posture — https://www.energy.gov/articles/trump-administration-calls-emergency-power-auction-build-big-power-plants-again
Leave a Reply