The gap between requested and deliverable power for U.S. data centers is staggering: Wood Mackenzie projects utilities will firm up only 28% of the 1,066 gigawatts currently in interconnection queues, exposing a planning crisis that threatens grid reliability, generation investment signals, and the pace of AI infrastructure deployment.
Queue Inflation Has Reached Crisis Proportions
The 1,066 GW figure represents more than the entire installed generating capacity of the United States, which sits at roughly 1,250 GW across all fuel types. That a single load category – data centers – could request nearly the equivalent of the nation’s full power plant fleet signals a fundamental breakdown in the interconnection process. Wood Mackenzie’s breakdown shows only 12% of projects have secured firm service commitments, with another 17% rated “likely” to proceed. The remaining 71% exists in a speculative limbo that distorts transmission planning, generation procurement, and regulatory resource adequacy assessments.
The mechanics driving this inflation are straightforward. Hyperscalers and colocation providers simultaneously file interconnection requests with multiple utilities across different regions, treating the process as a cost-free option on future power access. A single campus might appear in PJM, ERCOT, and MISO queues simultaneously, each filing counting toward the headline gigawatt total. Developers secure land options and preliminary permits in parallel, then narrow choices based on power pricing, fiber proximity, and regulatory certainty – but the withdrawn requests linger in queues for months or years before formal withdrawal.
This behavior mirrors the speculative queuing that plagued renewable generation interconnection in the 2010s, but with a critical difference: data center loads are firm, continuous, and often require 99.999% reliability. A 100 MW solar farm that never builds leaves a hole in renewable procurement targets; a 100 MW data center that never materializes after a utility has upgraded substations, built transmission, or contracted generation creates stranded assets and ratepayer exposure.
Utilities Are Weaponizing Financial Barriers to Filter Reality
The defensive measures utilities are deploying – hefty application fees, substantial collateral requirements, and top-tier credit rating mandates – represent a profound shift in the risk allocation between load and grid. Historically, utilities bore the obligation to serve new load within their territory, recovering costs through rate bases. Now, facing queues that could require hundreds of billions in transmission and generation investment, they are treating large-load interconnection as a commercial negotiation rather than a regulatory obligation.
Application fees that once ran in the tens of thousands now reach seven figures for gigawatt-scale requests. Collateral requirements – letters of credit or cash deposits tied to projected upgrade costs – can exceed $50 million for a single large campus. Credit rating thresholds effectively exclude private equity-backed developers and smaller colocation operators, concentrating viable projects among the handful of hyperscalers (Microsoft, Google, Amazon, Meta) and well-capitalized colocation giants (Digital Realty, Equinix, QTS).
This filtering mechanism will accelerate industry consolidation. Developers unable to meet financial thresholds will either partner with capitalized players or exit. The result: fewer, larger projects with stronger balance sheets – but also reduced competition for utility service territories and less geographic diversity in data center siting.
Grid Operators Are Rewriting the Rules in Real Time
PJM’s move to improve large-load forecasting and ERCOT’s temporary approval halt signal that regional transmission organizations (RTOs) recognize the current process is broken. PJM’s queue reform efforts, already underway for generation, are being adapted for load: cluster studies, milestone-based withdrawal penalties, and readiness requirements. ERCOT’s pause is more drastic – a regulatory timeout that effectively freezes new large-load interconnection while the Public Utility Commission of Texas evaluates whether the current framework can handle the volume.
These interventions will likely spread. MISO, SPP, and CAISO face similar queue pressures, though with different regional dynamics. MISO’s footprint includes growing data center corridors in Minnesota, Wisconsin, and Illinois; SPP sees requests in Oklahoma and Nebraska; CAISO contends with both data center load and the electrification mandates driving overall demand growth. Each will adopt variants of the PJM/ERCOT playbook: stricter entry requirements, faster withdrawal timelines, and cost allocation reforms that shift more upgrade risk to the interconnecting customer.
Cross-Cutting Analysis: The Generation Procurement Mismatch
The queue distortion creates a dangerous signal for generation developers and resource adequacy planners. If utilities plan transmission and generation for 1,066 GW but only ~300 GW materializes, the result is overbuilt infrastructure and inflated capacity prices – or, conversely, if planners discount the queue too aggressively and 500 GW arrives, the result is reliability shortfalls and emergency procurement at premium prices.
Consider the generation side: roughly 2,000 GW of generation and storage sits in U.S. interconnection queues, with median wait times exceeding five years. Data center load requests now compete with renewable and storage projects for the same constrained transmission upgrades. A utility facing a 500 MW data center request and a 500 MW solar-plus-storage request for the same substation upgrade must choose – or force cost-sharing arrangements that neither party anticipated. The data center’s firm load profile and credit strength often win, crowding out clean energy projects that need the same wires.
This dynamic also undermines resource adequacy modeling. North American Electric Reliability Corporation (NERC) assessments rely on queue data to project future supply-demand balance. When 70%+ of load requests are phantom, both loss-of-load expectation (LOLE) calculations and capacity market parameters become unreliable. PJM’s capacity auction, already under pressure from generation retirements and delayed new entry, now faces demand-side uncertainty of unprecedented scale.
My estimate: the “real” data center demand pipeline – projects with land control, capital commitment, and a defined utility path – is likely 250-350 GW through 2035, not the headline 1,066 GW. That aligns with WoodMac’s 28% commitment rate but implies a more concentrated buildout in fewer regions, primarily Northern Virginia, Texas, Ohio, Arizona, and the Pacific Northwest. The implications for gas turbine orders, nuclear restart evaluations, and long-duration storage deployments are significant: developers should plan for concentrated, creditworthy demand clusters rather than a geographically diffuse wave.
Who This Affects
- Utility transmission planners: Must redesign interconnection study processes to incorporate financial milestones, cluster evaluation, and dynamic withdrawal penalties – or face continued queue clogging that delays legitimate projects.
- Generation and storage developers: Should prioritize interconnection positions in regions with credible data center demand (Northern Virginia, Texas, Ohio) and expect increased competition for upgrade cost-sharing from load customers with stronger balance sheets.
- State public utility commissioners: Need to establish clear cost allocation frameworks for transmission upgrades driven by speculative large loads, preventing ratepayer exposure for infrastructure that serves projects that never build.
- Data center developers and REITs: Face higher upfront capital requirements (fees, collateral, credit support) and must demonstrate project maturity earlier – favoring large, well-capitalized players and accelerating consolidation.
- Grid operators (RTOs/ISOs): Must overhaul load forecasting methodologies to distinguish firm commitments from speculative requests, or risk resource adequacy models that either over- or under-state future peak demand by hundreds of gigawatts.
What to Watch Next
- ERCOT/PUCT rulemaking outcome (Q4 2024-Q1 2025): The Texas pause resolution will set precedent for how other regions handle large-load queue management – watch for mandatory readiness deposits, study cost caps, and withdrawal penalties.
- PJM large-load cluster study results (late 2024): First test of reformed process; the number of projects that post collateral and advance to Phase 2 will calibrate the “real” pipeline for the largest U.S. market.
- FERC Order 2023 compliance filings for load interconnection: Whether RTOs apply generation queue reforms (cluster studies, commercial readiness) symmetrically to large loads – or create a two-tier system.
- Hyperscaler earnings calls (quarterly): Capital expenditure guidance and site-specific deployment timelines from Microsoft, Google, Amazon, Meta – the only entities with visibility into which queued projects are real.
- Gas turbine and nuclear OEM order books (2025-2026): GE Vernova, Siemens Energy, Mitsubishi Power, and nuclear restart developers (Constellation, Holtec, NextEra) will signal where firm power purchase agreements are actually being signed.
Bottom line: The 1,066 GW headline is a planning artifact, not a demand forecast. The industry must recalibrate around the ~300 GW of creditworthy, sited projects – and build the financial and regulatory filters to keep the next wave of speculative requests from paralyzing the queue again.
Read the full report at Energy Central
Note: facts and figures attributed above to reflect that outlet's original reporting. Broader context, cross-sector connections, and forward-looking scenarios reflect independent analysis by our editorial team.
About this article: Drafted by Energy Ai with AI-assisted research and writing based on public reporting, then reviewed under our editorial process before publication.
Leave a Reply