Grid Interconnection Queue for Data Centers, Explained

Load Interconnection vs. Generation Interconnection

These get conflated constantly, including in press coverage. Generation interconnection is the process a power plant — solar, wind, gas, batteries — goes through to connect to the grid and sell electricity. Load interconnection is the process a large consumer, like a data center, goes through to connect and draw electricity. They run through different study processes, sometimes different teams at the same utility, and are governed by different rules depending on the region and grid operator.

A parcel's data center prospects depend on the load queue and available deliverable capacity at the nearest substation — not on how many generation projects are waiting to connect elsewhere on the system.

The rules that govern this process also vary a lot more by region than most site selectors expect going in. Formal RTOs like PJM or MISO — the MISO South sub-region covers Entergy Mississippi’s territory, for instance — run standardized, published queue processes. PJM in particular has spent 2026 restructuring its large-load process with new expedited tracks — see our PJM interconnection queue guide for specifics. Southwest Power Pool has gone further than most in the central US, pairing a 90-day study track with a curtailable bridge-service product — see our SPP interconnection queue guide for how that works. California's CAISO, by contrast, is still building its large-load framework from a queue designed mainly for generation — see our CAISO interconnection guide for where that stands in 2026. Plenty of active markets sit outside that structure entirely: Utah’s Rocky Mountain Power and New Mexico’s PNM territory are both in the Western Interconnection without full RTO membership, which means the study process runs through the utility's own procedures rather than a regional market operator's published tariff — worth confirming directly rather than assuming a standardized timeline applies. Some states are now legislating queue and cost-allocation rules directly: Oklahoma’s 2026 ratepayer protection law requires any large load over 75 MW to sign a long-term agreement covering its own interconnection costs, while Kentucky’s PJM-territory utilities have seen queue interest surge fast enough — nearly 30 potential projects at one utility alone in early 2026 — that state lawmakers have introduced (though not all passed) bills addressing who pays for the resulting grid upgrades. Michigan shows how fast this can move in the other direction too: an uncapped 2024 tax exemption drew so much large-load interconnection interest — DTE alone is evaluating roughly 3 GW of potential new capacity — that lawmakers introduced bills to walk the incentive back within about eighteen months. New York illustrates a different pattern entirely — the state's zone-based grid pricing means queue and congestion dynamics vary sharply within the state itself, with downstate zones running far tighter than upstate territory near NYPA hydropower. New Jersey shows how fast a PJM state's queue can fill even without a new grid operator or rule change — PSEG alone reported its large-load pipeline surging to roughly 9.4 GW by mid-2026, over 90% of it data centers, a pace of growth that outstrips what most utility planning cycles were built to absorb. Idaho shows a different variant of the same problem — Idaho Power's large-load planning has to account for an existing major industrial customer (Micron's ongoing Boise-area expansion) alongside new data center requests, and rising rates tied to that combined growth prompted a 2026 legislative proposal that would have blocked new 20+ MW customers from being approved without proof they wouldn't raise costs for other ratepayers. Montana shows yet another variant — rather than an open queue, NorthWestern Energy has been negotiating individual development agreements with named developers (Sabey Data Centers, Atlas Power) while separately filing a Large New Load Tariff application with state regulators to formalize how large loads pay for the capacity they use, before most of the announced load actually comes online. And North Dakota illustrates how differently this can play out under a rural electric cooperative rather than an investor-owned utility: Applied Digital is funding roughly $75 million of a $110 million substation-and-transmission project itself for its Fargo-area campus, with the cooperative owning the finished infrastructure — a path around a slow queue that only works because the developer is willing to pay for the capacity up front. South Dakota shows the more conventional version of the same principle — the Sioux Falls-area Gemini project was specifically sited next to Xcel Energy's Split Rock substation and the Angus Anson power station, existing infrastructure that's reportedly enough to support the project's full planned 500 MW draw without a comparable developer-funded build. New Hampshire illustrates a structurally different variant still — it sits inside ISO-NE, a multi-state regional grid operator covering all of New England rather than a single-state utility footprint, and frustration over regional power costs has gotten serious enough that the state funded a study on withdrawing from ISO-NE entirely, a reminder that queue dynamics can be shaped by regional grid governance, not just a single utility's own planning cycle.

Why the Load Queue Has Gotten More Attention

Utilities plan capacity additions on multi-year cycles tied to forecasted demand and capital budgets. Data center and AI infrastructure demand has grown quickly enough, in a short enough window, that many utility territories are now seeing more large-load requests arrive than their existing planning assumptions anticipated.

When several large loads request service in the same area around the same time, utilities commonly evaluate them together in a study cluster to understand the cumulative impact on shared infrastructure — a substation, a transmission segment — rather than approving each request in isolation. That grouped review can extend timelines relative to what a single, isolated project might experience on its own.

What Determines Queue Position and Speed

How This Connects to Site Value

Two parcels with identical acreage and zoning can be worth very different amounts once interconnection is factored in. See why land near substations matters for how proximity translates into value, how much power a data center needs by facility size, and the full data center site requirements developers screen against before interconnection even comes up. Power and interconnection decide whether a site is viable at all; once that's cleared, fiber and connectivity requirements shape how quickly and cheaply it can actually go live — and the utility power study itself is usually the longest-running item in a site's full due diligence checklist. Queue pressure varies by utility too — Washington's Grant County PUD is now near its maximum load and has slowed new large-load construction as a result, while Colorado's Xcel Energy has proposed a large-load tariff that shifts the full cost of new capacity onto big users rather than rationing it through a slower queue. Where the queue itself is the bottleneck, more developers are turning to on-site and behind-the-meter power to bypass it entirely — and a smaller number are underwriting nuclear and small modular reactor capacity as a longer-dated baseload alternative, though that path is realistic for far fewer sites. Clearing the queue is also no longer the last hurdle — see our guide to transformer and equipment lead times for why the physical equipment needed to actually energize a site can now take as long to arrive as the interconnection approval itself. And getting through the queue isn't the end of the power story either — see our guide to curtailment and demand response programs for how ERCOT, PJM, and other grid operators are now writing rules that can interrupt a facility's draw even after it's fully interconnected and online. A site that's already documented its power feasibility ahead of a formal interconnection request is exactly the kind of head start state shovel-ready site certification programs are designed to capture.

Frequently Asked Questions