Grid Access, Large-Load Service & Speed-to-Power

In the chip-bound era, the gating question was how many accelerators you could buy. In the power-bound era it is how many megawatts you can energize, and when. A bureaucratic process decides the answer, and most engineers learn how it works only after it has cost them eighteen months. The interconnection queue is the scarcest asset in the project. A live cluster needs steel, power and cooling, and any of the three can control the schedule. Which one controls is project-specific. Line up the utility's study, upgrade and service dates against civil works, fit-out, commissioning and equipment delivery to find it. This chapter covers how the grid-service process works, where it breaks, how the 2024–2026 reforms reshaped it on four continents, and the fork every speed-to-power strategy turns on: firm-but-slow vs flexible-but-fast.

The forks carry direct costs. Site into the wrong regime and your cost-allocation rule flips from "the network socializes upgrades" to "you pay 100% of everything you touch." Insist on firm power and you take the full queue wait. Accept an executable curtailable service offer and you can weigh its earlier power against curtailment exposure, bridge-generation cost and merchant-tail risk. Order the building before the transformer and the slab sits dark for two years past mechanical completion. This chapter owns the grid-impact process. The macro load-growth story behind the crisis is in Chapter 16.1, and turning grid flexibility back into revenue is in Chapter 15.8.

Do not start from one universal interconnection funnel. Begin by working out which processes actually apply to your project. Every project starts with large-load service, retail or transmission, under the tariffs of the host utility and the ISO/RTO and under the state's rules. A co-located or behind-the-meter arrangement brings its own rules on top of that. Generator interconnection is a further, separate process, and it applies only when your on-site generation exports power or otherwise connects to the grid as a generating facility. Each path has different deposits, studies, cost allocation, milestones, withdrawal rights and contract endpoints.

Large-load service path. Use the host utility or ISO/RTO tariff and the state process: application, load-flow, short-circuit and dynamic studies as required, facilities and network-upgrade scope, then the retail-service, transmission-service or large-load agreement. Co-located / behind-the-meter path. Add the applicable withdrawal, standby, netting, curtailment and operating rules. Generation/export path. If the campus generation exports or otherwise qualifies as a generating facility, run the separate generator-interconnection procedure and agreement. Record the governing tariff or order, the study endpoint and the contract name, and do not substitute LGIA terminology for load service.

The upgrade scope and cost that bind you come out of the studies the governing tariff or utility process requires. Where that process studies interacting requests, another project's withdrawal can trigger a re-study. Where it does not, do not import generator-queue mechanics. Record the actual study dependency, cost-allocation rule, milestone and contract endpoint for your site.

FERC Order 2023 reforms generating-facility interconnection. Its cluster studies, readiness requirements, withdrawal penalties and transmission-provider deadlines do not create a national large-load queue or a default data-center study process. They matter directly only to a generating facility the tariff covers, so for a load project they are supply-side context.

Large-load screening is instead being revised through utility tariffs, state rules and market-specific proceedings. Some utilities and RTOs use readiness deposits, site control, duplicate-request screening or flexible-service products, but authority and contract terms differ by jurisdiction. Use ERCOT's published large-load population only for ERCOT, and LBNL's queue only for generation and storage. The United States has no comparable national large-load queue dataset, so do not transfer generator attrition rates to data-center load requests.

Read the table as a siting screen. The regime is a decision variable you select, and the choice carries through to your upgrade costs, your firmness and your schedule.

ERCOT is a single energy-only grid, and SB6 does not make it a fast lane. Before you connect, the serving utility must be able to curtail you in a firm load shed, and NOGRR282 adds ride-through for large computational loads. Neither speeds service. Classification, the audit, studies and ERCOT's approval to energize still gate it. PJM is the opposite lesson: deep, congested and slow, with generator application-to-commercial-operation stretching past eight years and a generation shortfall that makes every new large load politically fraught. FERC's December 2025 PJM order, with its four transmission-service options and a behind-the-meter cost-causation reform, tries to make co-location workable. It is not a speed promise.

SPP is the quiet outlier. Its HILL/HILLGA study pathways and its CHILLS conditional non-firm service are approved; PALS, its design for price-adaptive load, is still only a proposal. Finishing a study there does not give you an energization date. Vertically integrated utilities are the wild card: one counterparty can move fast or capture you entirely, and line-extension policy often puts 100% of the distribution upgrade on your bill.

Outside the US the process differs in kind, not just degree, and the constraint is often a hard moratorium rather than a queue you can pay to jump. The question stops being "firm or flexible" and becomes "is new large-load connection legally available at all, and on what conditions."

EU / UK. Great Britain's connections queue, run by the National Energy System Operator (NESO, the former National Grid ESO) under Ofgem, ballooned to hundreds of GW. A 2025 reform culls stalled "zombie" projects and re-orders the queue. Its Gate 2 requires both readiness and strategic alignment before a project gets a confirmed connection date, point and queue position, and transmission-connected demand is an eligible alignment category. FERC Order 2023's readiness reforms apply only to generating facilities, so the two processes are not interchangeable.

Ireland is the cautionary extreme. In November 2021 EirGrid and the CRU set restrictive connection criteria for data centers. On December 12, 2025, the CRU swapped the blanket restriction for a conditional route to apply. The system operators can still refuse a connection once they have assessed local network constraints, and the policy does not mean capacity is available in Greater Dublin. Any new data center with a maximum import capacity of 10 MVA or more must bring dispatchable generation or storage, on site or nearby. That plant must be separately connected and metered, match the import capacity once de-rated, and take part in the wholesale market. Projects of 1 MVA or more must, within six years of energization, cover at least 80% of their annual demand from additional renewable generation located in the Republic of Ireland. Renewables that have held or now hold REFIT, RESS or ORESS contracts do not count. In effect, Ireland redefined interconnection as "bring your own firm power and grid services."

The Nordics (Norway, Sweden, Finland) remain the relief valve, with firm hydro, cold-climate free cooling and comparatively available transmission. Grid operators in Norway and Sweden have started putting connection requests in priority order. Finland's draft of September 18, 2026 would sort demand and storage connection requests into four priority groups by connection capacity and purpose. It would also weigh how ready each project is and what dispatchable or new local generation it contributes. The target start is January 1, 2027, and earlier requests would be caught too unless they already hold a concluded or conditional connection agreement. Fingrid's October 9 response asks for further clarification of the proposed law. The Middle East (Saudi HUMAIN, UAE) and India offer state-coordinated, fast-tracked access where sovereign-AI ambition aligns connection policy with the project.

APAC is shaped by moratoria. Japan faces transmission constraints around Tokyo and Osaka. Singapore ran a hard moratorium from 2019, and its successor regime rations capacity by competitive bid. The December 2025 DC-CFA2 call drew more than 20 proposals, and in August 2026 it provisionally allocated 200 MW, 50 MW each to four operators on Jurong Island, on efficiency and green-energy commitments rather than first-come. Malaysia (Johor) absorbed the spillover and is now itself power- and water-constrained.

A 2026-specific risk sits on top of the regime map: the rules may change mid-queue. The Secretary of Energy moved on October 23, 2025, invoking Section 403 of the DOE Organization Act, an authority seldom used. The directive told FERC to consider an ANOPR, an advance notice of proposed rulemaking, on connecting large loads, generally those above 20 MW. FERC created docket RM26-4 and took comments through late 2025. On 18 June 2026, rather than issue a nationwide rule, it opened Section 206 show-cause proceedings against all six RTOs/ISOs. Each must justify or reform its large-load interconnection, cost-allocation and transmission-service tariff, and the RM26-4 docket stays open for possible later action. Those regional proceedings are separate from the DOE proposal. The clock then slipped: all six RTOs won roughly 90-day abeyances in mid-August 2026, pushing tariff filings toward late 2026. The contemplated reforms include standardized large-load study procedures, a target study timeline for flexible loads and joint co-located load-plus-generation filings.

PJM's Board did not wait. On 2026-07-27 it directed a Large Load Registry, tracking each large site's location and supply status, and an Interim Resource Adequacy Service (IRAS), which would have curtailed new large loads ahead of Pre-Emergency Load Management unless they brought their own generation by 2027-06-01 or otherwise secured supply. FERC split the filing. Its October 9, 2026 order (197 FERC ¶ 61,019, ER26-3515-000) accepted PJM's Large Load definition and the registry, effective October 12. A Large Load is 50 MW or more of peak load at one site, with affiliated facilities within one mile counted together. The same order rejected IRAS, the requirement that new load be matched with new capacity, and the Bring Your Own New Capacity rules, and PJM owes a compliance filing within 30 days. IRAS is not an available service, so do not build service or curtailment terms around it.

PJM's Reliability Backstop Procurement addresses supply, and it runs on its own track. FERC accepted it for filing on September 29, 2026. The order suspends it for five months, until February 28, 2027, and makes it subject to refund and further proceedings. Cost allocation, transmission-owner exit rules and load-serving-entity collateral remain under consideration. The procurement was meant to launch in September 2026, so do not key first power or customer security terms to that date. Track the tariff and procurement dates separately from your service agreement.

Texas showed that connection rules can also freeze mid-process. The governor's August 3, 2026 directive paused ERCOT's Batch Zero large-load study pending a project-by-project audit; what happened next is below.

Congress is a third policy vector. Introduced on September 30 with bipartisan sponsors, the American Affordability and Jobs Act of 2026 had not passed as of October 12. It would apply to data-center or high-density computing load of 20 MW or more, at one site or spread across sites under common ownership or control, counting any phased full build-out that can reasonably be expected. Under the bill, a utility would demand financial assurances or contributions large enough to cover the cost before it builds or upgrades the generation, transmission or distribution facilities that load needs. Amounts already provided for the same facility would be credited. Track the bill alongside RM26-4 when you negotiate construction security.

The collision is jurisdictional. Large-load interconnection has traditionally been state and utility territory. The DOE and FERC moves assert a federal interest in standardizing it, and several states have pushed back to protect their authority over retail load, siting and cost allocation. For a developer the consequence is regulatory tail risk: an interconnection agreement negotiated under today's state or RTO rules could be reshaped mid-construction, changing your cost allocation, flexibility obligations or study timeline. Build optionality into your terms, which means flexibility you can offer rather than have imposed. Track RM26-4 and the Section 206 filings as live schedule items. Do not bet the project on a cost-allocation rule that subsequent federal orders or accepted tariff revisions might overturn. The macro driver is in Chapter 16.1.

ERCOT's Batch Zero status is a dated sequence. The pause came on August 3. On August 20 the PUCT approved procedural exceptions, and ERCOT's August 21 notice, M-A080326-02, opened a route to conditional classification. On September 3, ERCOT issued conditional classifications covering 362 base or studied projects. Notice M-A080326-04 says ERCOT notified interconnecting DSPs and TSPs of provisional Batch Zero inclusion, subject to verification, dynamic-model cure or good-cause conditions. A load that fails to satisfy a condition can be removed from the batch. Under the PUCT exceptions, base loads with a conditional classification can enter the August and November quarterly stability assessments. The Batch Zero study timetable itself is still being revised.

Classification is not service. A conditional classification does not establish a completed study, a signed service offer or permission to energize. Under ERCOT's September update, any new data center or crypto facility of 75 MW or more stays unenergized while verification, the audit and the community-impact review run. A state permit gate now runs alongside it: on September 21, 2026 the governor directed TCEQ to halt permits sought by data centers until ERCOT completes its audit (Chapter 3.9).

ERCOT's State and Community Impact RFI of September 14, 2026 reaches two groups. One is unenergized Batch Zero data centers at transmission voltage (60 kV or above) that hold a conditional base or studied classification. The other is unenergized, substantiated data centers of 25 MW to under 75 MW, at any voltage, that were reported through the August 19 Medium Load RFI. The developer answers questions on ownership, grid dependence, water, cooling, community impacts and public assistance, attaches supporting documents and signs a notarized attestation. Its TSP or DSP files the package through RIOO, due by 17:00 Central Prevailing Time on October 12, 2026. ERCOT aims to issue two separate reports, one on eligibility verification and one on community impact, by December 10, 2026. Put your project's actual classification and remaining conditions in the integrated master schedule (IMS), and do not treat the notice date as a first-power promise.

In practice the firm-but-slow / flexible-but-fast fork resolves into an energization sequence. Rather than waiting years for firm service to the full campus peak, the operator energizes in tranches and covers the gap with flexibility and bridge generation. The pattern is now standard at the GW-scale frontier:

• Phase 1: bridge power. While interconnection and network upgrades work through study, energize the first 50–100 MW tranche on behind-the-meter generation (gas reciprocating engines or aeroderivative turbines). That often stands up the first IT load on a project-specific schedule, before firm grid service arrives. The bridge-power thesis and generation choices are in Chapter 3.5, and electrical integration is in Chapter 4.8.
• Phase 2: flexible grid tie. Connect early on curtailable or non-firm terms, such as an ERCOT-approved conditional arrangement, SPP CHILLS or a utility interruptible tariff, without waiting for the firm upgrade. You accept curtailment during system stress and cover the curtailed hours from the bridge plant.
• Phase 3: firm grid. When the network upgrade and the long-lead transformer finally arrive, move to firm service and re-purpose or retire the bridge generation, or keep it as backup or grid-services capacity (Chapter 15.8).

Energized capacity leads commissioned IT load. A campus may have 100 MW energized while only 40 MW of IT is commissioned, because energization is staged ahead of the fit-out. The whole sequence exists to defeat one number, the queue wait, by turning a hard schedule gate into a manageable operating constraint. Flexibility is worth offering when an executable offer of earlier service is worth more than the curtailment, bridge-plant and workload costs it brings. The operator who designs for curtailment captures the speed. The one who insists on firm and is later forced to curtail anyway pays twice.

Winning the queue does not energize the building. The host utility's large-load studies identify the facilities and network upgrades your load needs (new lines, reconductoring, substation expansions), and those upgrades can be multi-year builds. The US built roughly 1,700 circuit-miles of high-voltage (345 kV-plus) transmission a year in 2010–2014; by the early 2020s the annual figure was down to a few hundred. Working from revised FERC data, ACEG/Grid Strategies' July 2025 addendum counts about 450 miles for 2023 and 888 for 2024. From DOE's 2024 National Transmission Planning Study, the same analysts work out a need for roughly 5,000 miles of high-capacity regional transmission a year. DOE modeled capacity scenarios, and the mileage is their conversion of them.

Higher-voltage backbones are one answer. ERCOT's 765 kV program is the headline example. On April 24, 2025, the PUCT approved three 765 kV import paths into the Permian Basin. ERCOT's comparison of January 27, 2025 put the import paths at $9.06 billion, or $13.77 billion with their associated local upgrades. The full statewide STEP was estimated at $32.99 billion, and it mixes voltages: 2,468 miles of new 765 kV line, new and upgraded 345 and 138 kV lines, and 69-to-138 kV conversions. Separately, ERCOT's December 2025 constraints report counts 1,329 miles of 765 kV line approved for construction in the Permian region. ERCOT's Board endorsed two more projects on December 9, 2025: the Drill Hole–Sand Lake–Solstice project, and the Eastern Backbone, estimated at $9.384 billion and about 1,108.8 miles of new right-of-way. According to the same report, the Board also endorsed the rest of STEP. Board endorsement is not PUCT routing approval, and the Eastern Backbone's eastern segments still need it before construction. Construction milestones and energization dates are separate again. A single 765 kV circuit moves roughly the power of three double-circuit 345 kV lines in less right-of-way.

Existing corridors are the other option, so study grid-enhancing technologies (dynamic line rating, advanced reconductoring) too. A February 2024 summary on advanced conductors from DOE and Idaho National Laboratory says they could potentially double a line's capacity for roughly one-third the cost of a new line. Those figures are the report's estimates, and no project is guaranteed to achieve them. Before you count on extra service capacity, get from the utility the limiting element and contingency analysis, plus the upgrade's scope, cost and completion date.

Even with the line built, the project waits on iron. The long-lead transformer can be as hard a constraint as the queue itself. Wood Mackenzie reported US substation-transformer lead times above 160 weeks in May 2026, and generator step-up units are a separate category. Incumbents (GE, Siemens, Hitachi/Mitsubishi) quote multi-year backlogs. Order the building first, discover the transformer is three years out, and the slab sits dark long past mechanical completion. Order the gate, not the building, first: place transformer and switchgear orders against the interconnection in parallel with, or ahead of, civil works, and treat the long-lead register as a board-level schedule artifact. The full equipment supply chain and lead-time register is in Chapter 2.3.

The forks above compound into a single speed-to-power posture, and mismatching them is expensive. Three pairings recur.

• Regime × firmness. In ERCOT, conditional service plus bridge generation can make a staged supply plan, provided ERCOT approves the project's studies, operating limits and energization. PJM-firm is coherent but slow, and PJM-flexible is constrained by the co-location order. In Ireland and Singapore, each jurisdiction's connection and sustainability conditions decide which service options you can have. Picking a regime implies a firmness option set, so once the site is fixed you do not get to choose freely.
• Flexibility offered vs imposed. The operator who designs for curtailment and ride-through from day one captures the speed and can monetize the flexibility (Chapter 15.8). The one who insists on firm and is later subject to SB6-style mandates pays for the firm upgrade and eats the curtailment. Offer it; do not wait to have it taken.
• Gate ordering. Wherever the dependencies allow, run the service process, equipment delivery, network upgrades and facility completion in parallel, and plan first IT service around whichever milestone controls.

The downstream economics of all three (nodal pricing, congestion exposure, network-upgrade cost allocation and the "who pays" question) are quantified in Chapter 3.3. The supply strategy that monetizes the firm/flexible split (PPAs, behind-the-meter islands, co-location) is in Chapter 3.4.