interconnection queues
on this page
overview
even when generation capacity exists, data centers cannot connect without completing interconnection studies and building transmission infrastructure. queue wait times have extended from 2 years (2008) to 8+ years (2025).
the interconnection process requires multiple study phases, financial deposits, and coordination across utilities, regional transmission organizations, and grid operators. delays stem from study backlogs, limited transmission capacity, and the sheer volume of new requests driven by ai data center demand.
queue backlog by region
| rto/region | queue backlog | wait time | notes |
|---|---|---|---|
| pjm | 265gw | 8+ years | was under 2 years in 2008 |
| miso | 242gw | 3-4 years | 650-day delays common |
| ercot | 2,000+ requests | faster | 20% suspension rate |
| dominion (va) | 40+ gw | 7 years | data center alley congestion |
| oncor (tx) | 186gw | variable | 250% increase 2023-2024 |
regional context
pjm (mid-atlantic, midwest): the largest rto in north america, covering 65 million people across 13 states and dc. queue times have grown 4x since 2008 as data center demand concentrates in northern virginia and ohio.
miso (midwest, south): interconnection queue includes 242gw of requests, with average study completion taking 3-4 years. recent reforms attempt to streamline processes but face resource constraints.
ercot (texas): operates independently from federal oversight, enabling faster reforms. however, rapid load growth in dallas-fort worth and austin has created new bottlenecks.
dominion energy (virginia): serves data center alley, the world’s largest concentration of data centers. 40gw pipeline represents nearly 10x current data center load in the territory.
project success rates
the majority of interconnection requests never reach operation:
- only ~10% of queued projects reach operation within 3 years (enverus analysis)
- ercot: 20% suspension rate
- pjm, spp, nyiso, iso-ne: 46-79% suspension rates
causes of project failure
- financial barriers: deposit requirements and study costs exceed project budgets
- timeline uncertainty: 5-7 year waits make financing difficult
- transmission costs: network upgrade costs can exceed $100m for large projects
- site issues: inability to demonstrate site control or commercial readiness
- market changes: power purchase agreement failures or technology shifts
high suspension rates create a cycle: speculative projects clog queues, slowing legitimate projects, which then encourages more speculative filings to secure queue positions.
study costs and deposits
ferc order 2023 reforms (2023-2024)
federal energy regulatory commission reforms aimed to improve queue management:
- increased deposit requirements: higher upfront financial commitments
- site control demonstration required: proof of land rights before study
- commercial readiness deposits: evidence of project financing
- withdrawal penalties: financial disincentives for abandoning queue position
these reforms reduced speculative filings but increased barriers for smaller developers.
miso-specific fees
- d1 application fee: $7,000 (non-refundable)
- m2 milestone deposit: $8,000 per mw studied
- additional deposits: required for system impact studies
for a 100mw data center project in miso:
- application fee: $7,000
- m2 deposit: $800,000
- potential study costs: $200,000-500,000
- total upfront: $1m+ before any construction
cost of grid delays
transmission cost pass-through
grid upgrades required for data center connections are typically passed to ratepayers:
- pjm (2024): $4.3 billion passed to ratepayers for data center connections
- virginia bore: $1.98 billion of this cost (46%)
- pjm capacity auction: $2.2b to $14.7b single-year spike (2025/2026 delivery year)
the capacity auction spike reflected tightening reserve margins and anticipated data center load growth. residential customers in pjm territory saw bills increase $20-40/month.
project-level impact
- renewable project cancellations (h1 2025): $22b in abandoned projects
- sunk cost per abandoned mw: ~$200,000 in study fees and deposits
- developer exits: smaller renewable developers unable to compete with hyperscaler resources
queue delays disproportionately affect renewable projects, which have lower profit margins and less access to capital than data center-backed generation.
the behind-the-meter response
faced with 5-7 year queues, operators pursue on-site generation to bypass interconnection entirely.
advantages
- bypass queue entirely: no interconnection studies required
- 2-3 year timeline: vs 5-7+ years for grid connection
- no transmission constraints: generate at point of use
- independent of utility approval: simplified regulatory path
disadvantages
- higher upfront capital: $1,000-1,500/kw vs $600-800/kw for grid connection
- air quality permitting challenges: local opposition to gas turbines
- fuel supply contracts required: natural gas pipeline capacity constraints
- environmental opposition: emissions conflicts with corporate sustainability goals
result
54.5% of gigawatt-scale projects now use on-site or hybrid power strategies (enverus, 2025). this represents a fundamental shift from grid-reliant data centers to distributed generation models.
hybrid approaches
many projects combine strategies:
- grid connection for baseline load
- on-site generation for peak demand
- battery storage for reliability
- renewable ppas for sustainability credits
regional examples
northern virginia (dominion energy)
northern virginia hosts the world’s largest data center market:
- 40gw pipeline (dec 2024): up from 21gw (july 2024)
- 7-year wait: for new large customers (100+ mw)
- 500kv transmission loop: $1.2b project under construction
- load growth: 2.5gw/year vs 0.5gw/year historical average
dominion’s service territory represents 70% of global internet traffic routing through data centers.
texas (oncor/ercot)
texas combines regulatory flexibility with rapid load growth:
- 186gw total pipeline: across all generation types
- 137gw formal interconnection requests: in active study
- 30gw high-confidence projects: with demonstrated site control and financing
- $36b capital plan (2025-2029): transmission and distribution upgrades
oncor’s territory includes dallas-fort worth metroplex, where hyperscalers are developing multi-gigawatt campuses.
ohio (aep/firstenergy)
columbus and central ohio emerging as alternative to virginia:
- lower interconnection costs: less congested transmission system
- shorter timelines: 3-5 years vs 7-9 years in virginia
- fiber connectivity: established carrier-neutral facilities
- land availability: greenfield sites with utility coordination
reform efforts
ferc initiatives
- order 2023 (2023): cluster study approach, increased deposits, site control requirements
- order 2023-a (2024): clarifications on commercial readiness and withdrawal penalties
- proposed reforms (2025): data center-specific interconnection track under consideration
rto-specific improvements
pjm:
- transition to cluster study process
- expedited review for projects with on-site generation
- new capacity market rules to reflect load diversity
ercot:
- streamlined process for projects under 100mw
- direct utility interconnection option
- improved coordination with public utility commission of texas
miso:
- regional transmission planning improvements
- cost allocation reforms
- queue management automation
utility responses
utilities are creating dedicated teams for large load customers:
- dominion: data center interconnection specialists
- oncor: major customer delivery group
- aep: economic development fast-track program
future outlook
interconnection delays will persist through 2027-2028 as current queue backlogs work through reformed processes. key indicators to watch:
- ferc policy evolution: potential data center-specific regulations
- transmission buildout: major 500kv projects in virginia, texas, ohio
- on-site generation adoption: shift from exception to standard practice
- ai load concentration: whether hyperscalers diversify geographically
- reform effectiveness: whether ferc order 2023 reduces queue times by 2026
the fundamental tension remains: grid infrastructure operates on 5-10 year planning cycles while ai compute demand operates on 6-18 month deployment cycles. this mismatch will continue to drive infrastructure innovation and market disruption.
sources
- lawrence berkeley national lab “queued up: 2024 edition”
- ferc order 2023 and order 2023-a
- pjm interconnection queue reports (2024-2025)
- enverus 2025 interconnection queue outlook
- miso transmission expansion plan 2024
- ercot large load task force reports
- dominion energy investor presentations (2024)
- oncor 10-year transmission plan
- utility dive infrastructure coverage
- wood mackenzie north america power forecast