Cornerstone Guide

Utility Procurement Intelligence Guide

How to build a procurement intelligence function for utility distribution. Data sources, market signals, competitive monitoring, and decision frameworks for equipment sourcing.

Last updated September 9, 2026 Published March 31, 2026 39 min read DistroForge Research

Why Procurement Intelligence Matters

Most electrical distributors make procurement decisions with incomplete information. They know their own pipeline, their primary supplier relationships, and whatever market color they pick up from trade shows and industry contacts. That worked when the market moved slowly. It does not work in 2026.

Procurement intelligence is the systematic collection, analysis, and application of market data to sourcing decisions. It turns pricing trends, lead time shifts, competitive moves, and regulatory changes into inputs a procurement team can act on.

That reframe is no longer ours alone. Writing in Utility Dive in May 2026, ULE Group President Danielle Pirrone argued that equipment procurement has crossed from a purchasing function into a grid reliability function. When transformer, switchgear, and breaker schedules slip, the reliability work they support slips with them, and the budget overruns follow. A procurement team that treats lead time and pricing data as reliability inputs, not just cost inputs, is reading the same signal a utility planner reads, and reading it earlier.

Public Data Sources

The foundation of procurement intelligence is public data. Government databases, regulatory filings, and industry reports contain signals that most procurement teams never see because nobody is looking.

EIA (Energy Information Administration) publishes monthly data on utility capital expenditures, generation capacity additions, and electricity demand by region. These datasets reveal where equipment demand is heading before it shows up in manufacturer lead times.

FERC filings contain rate case details, capital spending plans, and equipment procurement data from investor-owned utilities. When a utility files a rate case requesting $500M in grid modernization spending, that demand signal is public information months before the RFPs go out. FERC dockets also surface policy shifts with direct procurement consequences. In April 2026, nine utilities filed docket EL26-58-000 seeking to suspend competitive bidding for transmission projects in MISO and SPP, a ruling that could consolidate or fragment equipment procurement across 18 states. Updated April 2026.

The federal regulatory calendar is itself a procurement signal. FERC has committed to act by the end of June 2026 on its large-load interconnection rulemaking (Docket RM26-4), the framework that will govern how data centers and other loads above roughly 20 MW connect to the interstate grid. The action slipped from an original April 30 target, and legal analysts at Troutman, Holland and Knight, and Snell and Wilmer expect a proposed rule rather than a final one, which points to standardized rules taking effect in late 2026 or early 2027. The procurement read is timing. Once the rule lands, expect a wave of new interconnection applications and the transformer, switchgear, and substation orders that follow them, and expect the option-to-build provision, if it survives, to open a direct purchasing channel for hyperscalers buying breakers, relays, and cable outside the utility. Teams that map their late-2026 and early-2027 buys to that milestone avoid paying for standard studies the rule may supersede. Our FERC large load interconnection coverage carries the equipment crosswalk. Updated 2026-06-16.

State PUC dockets provide similar visibility into municipal and cooperative utility spending plans. The signal quality varies by state, but the major markets (Texas, California, New York, Florida, Ohio) have well-documented regulatory proceedings.

SAM.gov lists federal procurement opportunities including utility equipment purchases by federal facilities, military installations, and DOE-funded projects.

National laboratory technical reports are the least-watched source on this list and often the most granular. DOE-funded national labs publish primary-research reports that trade coverage summarizes months later, without the underlying tables. A May 2026 example is NLR/TP-5700-96742, a large power transformer supply chain gap analysis funded through DOE’s Hydropower and Hydrokinetic Office. It re-baselines lead times with a citable federal series, 12 to 18 months in 2010 stretching to a 30-month average in the first half of 2025, and up to 60 months for extra-high-voltage units, and names the entire US manufacturing base above 345 kV: four plants, Hyosung HICO, Virginia Transformer, HD Hyundai Power, and Prolec-GE. Its most useful finding inverts the usual procurement conversation. Bushings, not the transformer tank, are the longest lead item at up to 130 weeks, and a spec that locks in a single bushing or tap-changer make turns a component the market can usually substitute into a self-inflicted multi-year wait. Our bushing and component-spec breakdown has the full lead-time table. Watch DOE program-office publication pages the way you already watch FERC dockets. A gap-analysis report is a capacity and import-reliance signal months before it becomes a headline.

Building a Monitoring System

The challenge is not data availability. It is data volume. Manually checking these sources is unsustainable. Effective procurement intelligence requires automated monitoring with human judgment applied to the filtered output.

A practical approach: set up keyword alerts on FERC filings for equipment categories you sell. Monitor EIA monthly reports for demand trends in your territory. Track PUC dockets in your key states for capital spending authorizations. Review SAM.gov weekly for direct procurement opportunities.

What Is the Date Backed By?

Every section of this guide so far treats a lead time as a number to be read. There is a second variable sitting behind that number, and most procurement teams do not track it at all: what the date is contractually backed by.

Speaking at the Energy Projects Conference and Expo in Houston in June 2026, Jeff Gulach of National Grid Ventures, who spent 36 years on the EPC side before moving into origination, put the change plainly. “The supply chain is still broken,” he said, and “original equipment manufacturers and distributors have retreated from delivery guarantees once backed by liquidated damages” (POWER, August 3, 2026). Note who is named in that sentence. The distribution channel is in it, not the factories alone.

That retreat is what makes the contractual strength of a date an independent variable from the date itself. A promised delivery date backed by liquidated damages is a priced, enforceable commitment: the supplier has put a number on its own slip. The same date offered without that backing is a forecast. Two quotes can carry identical stated lead times and entirely different risk, and nothing on the face of either quote tells you which is which.

The mechanism by which a date quietly stops being real is worth understanding, because it is a contract term rather than a factory problem. Gulach again: “I hear constantly, ‘My switches were supposed to be here eight and a half months ago, and they’re not here.’ You took your eyeball probably off of the production slot, and your [purchase order] allowed the distributor or the OEM to sell your production slot.” The equipment was never being built for you in the sense the buyer assumed. The purchase order permitted the slot to be reassigned to a buyer who was watching more closely, and it was. In the same remarks Gulach cited a four-year lead time on his most recent 345-kV breaker purchase, which is the scale of commitment now riding on paperwork most buyers never read past the price line.

He also named the staffing asymmetry that lets it happen: “I still see 300-person engineering teams and 1,200 craft heading to a major project, and then I’ll see a three-person procurement team still boarding all the stuff.” That ratio is worse, not better, at a municipal utility or a cooperative, where the procurement function is often one person with other duties.

So add three questions to any lead-time conversation, and ask them of the quote rather than of the market:

Is this date contractual or indicative? Ask for the answer in writing. A supplier that will commit in writing is telling you something a supplier that will only commit verbally is also telling you.

What is the remedy if it slips? Liquidated damages, a price adjustment, a substitute unit, an escalation path, or nothing. “Nothing” is a common and honest answer in this market, and it is still an answer you need before you build a schedule on the date.

Does the purchase order let anyone reassign my production slot? This is the clause Gulach’s story turns on, and it is the one most likely to be absent from a buyer’s own checklist. If the slot can be resold, the date is conditional on your continued attention, and somebody has to own that attention by name.

None of this requires price data or a supplier relationship you do not already have. It requires reading the terms attached to a date you were already being quoted, which is why it belongs in an intelligence function rather than in a negotiation.

Who carries the consequence of a weak date has shifted alongside the date itself, and the two changes compound. At the same conference, Zachry Group senior vice president Mike Kotara described owners going out to buy “extended scope packages” under a limited notice to proceed that “allows you to buy equipment at risk, at the owner’s risk,” covering equipment the EPC contractor used to absorb. Centuri Holdings states the same position from the contractor side in its Form 10-Q for the quarter ended June 28, 2026: “Generally, our contracts provide that the customer is responsible for supplying the materials for their projects.” So the buyer is now committing capital earlier, against less design, on a date carrying weaker backing than it did three years ago. Our read on owner furnished equipment procurement risk works through what a limited notice to proceed actually commits, and which attributes of an early order a late design change can invalidate. Updated 2026-08-26.

Price Indexes and the Pass-Through Read

Government dockets tell a procurement team what utilities plan to spend. Commercial price indexes tell it what the equipment underneath those plans already costs, often a quarter or two before the same pressure reaches a distributor’s own quotes. The renewable power purchase agreement market is the clearest public example, because PPA prices absorb equipment, tariff, labor, and permitting costs and pass them straight through to the offtaker.

LevelTen Energy’s Q1 2026 PPA Price Index, published April 14, 2026 from 291 offers across 207 projects in six North American markets, set records on both sides. Solar PPA prices rose 4.6 percent for the quarter and 13 percent year over year to $64.49 per MWh. The cheapest quartile of wind prices, the P25, rose about 24 percent year over year, which means even the most competitive projects now pay far more than a year ago. The increase was steepest in CAISO. LevelTen and the developers it interviewed named the drivers plainly: tariffs on imported modules, inverters, transformers, and steel, plus permitting delays, labor shortages, rising insurance, and data center buyers willing to pay a premium for firm delivery dates.

Two things make this a procurement signal rather than a renewables headline. First, North American PPA prices kept climbing while European prices fell, which isolates the cause to United States cost pressure rather than global module or turbine pricing. Second, LevelTen flagged that tax credit expiration under OBBBA had not yet become a primary driver, which implies a second leg of price pressure still ahead. For a distributor quoting collector substation transformers, GSU and step-up units, medium voltage cable, or inverters into utility-scale renewable projects, the read is to assume continued upward pressure and to treat the CAISO premium as real when pricing California work.

The demand side of the same picture shows up in the services layer. CBRE reported Q1 2026 revenue of $10.5 billion, up 19 percent, with its new Critical Infrastructure Services line growing 71 percent year over year to $580 million and management guiding to more than 60 percent growth for the year. CEO Bob Sulentic called the firm’s pivot into data center and critical power services at least as profound as its outsourcing pivot of the 1990s, and faster. The point for procurement intelligence is cross-confirmation. Rising PPA prices and a services firm reorganizing around critical power are two independent readings of the same demand, one from the price side and one from the operations side. When two unrelated public signals agree, the demand is durable enough to plan inventory and supplier commitments against.

State Governor 3-Benchmark Frameworks and ATT Procurement Gates

A new class of state-level pressure signal emerged in late April 2026 that procurement intelligence teams should monitor as the leading indicator for IOU capex deferral. On April 29, 2026, Pennsylvania Gov. Josh Shapiro sent letters to 24 electric, gas, and water utilities articulating three benchmarks that utilities must meet to earn his support for future rate cases: cheaper debt financing through DOE Loan Programs Office loans (with debt as “a clear majority” of the ratemaking capital structure), transparent cost-benefit analysis disclosing the share of rate increases flowing to shareholder dividends versus customer benefits with proof that existing grid resources have been maximized before new capex, and competitive return-on-equity bidding to replace utility-set ROE. PECO had requested 10.95%, which competitive bidding would likely cut by several percentage points (WHYY, April 2026; Vista Today, May 2026). Stock prices of Exelon, FirstEnergy, and PPL dropped the next day. PECO subsequently withdrew its rate hike under earlier Shapiro pressure.

The procurement signal is not the political theater. It is the explicit “maximize existing grid resources before requesting new capex” benchmark, which is the operational definition of non-wires alternatives and grid-enhancing technologies procurement. PA utilities now face a documented public standard for capex justification that materially favors distribution automation, advanced inverters, dynamic line ratings, reconductoring with high-performance conductors, and front-of-meter storage targeted at constrained feeders, while disfavoring greenfield substation builds and traditional transmission line projects. Distributors selling to PECO, PPL, and FirstEnergy PA should expect lengthening sales cycles for traditional gear and shortened cycles for ATT-study-collateral-eligible equipment.

The framework is replicable across states, and the parallel North Carolina case is the second instance of the same pattern. Jigar Shah’s argument, with backing analysis from Brattle Group, is that NC Governor Stein has authority to require data centers to co-locate battery storage at the state’s existing 7,200 MW of solar capacity as a condition of interconnection. Modeled outcome: 5% reduction in Duke Energy customer bills and a 10% improvement in grid utilization translating to a 3.4% rate decline by 2030, with national savings estimated at $110–$170 billion over 10 years for vertically integrated utility ratepayers (Energy Empire Podcast, May 2026; Utility Dive, April 2026). Duke is currently seeking a 13.5–13.9% rate increase despite $5B annual profits, with 80% of projected demand growth attributable to data centers, exactly the political setup that makes the Stein/Shah co-location argument live. The Minnesota Google agreement, projected to deliver $1.7–$1.9B in customer savings through co-located storage at hyperscaler load, is the operational template.

A separate but reinforcing PA signal is HB 2233, which the Pennsylvania House passed unanimously on May 5, 2026. The bill requires utilities to study advanced transmission technologies (high-performance conductors, dynamic line ratings, advanced power flow controllers, topology optimization software) before approving traditional transmission upgrades, and gives the PUC authority to mandate ATTs in approved projects (Utility Dive, May 2026). For procurement teams, this converts ATT-study collateral from a “nice to have” into a procurement gate. Distributors who can package DLR sensors (LineVision, Heimdall Power, Ampacimon), advanced conductor (CTC Global ACCC, 3M ACCC), advanced power flow controllers (Smart Wires), and FOM-storage interconnect kits as a bundled bid response gain advantage. Intelligence teams should add three new monitoring categories to their PUC docket scans: (1) governor letters to utilities containing 3-benchmark or similar capex-justification language, (2) ATT-study mandate bills in PA-style format moving in NC, NJ, VA, OH, GA, and IL, and (3) DOE LPO loan applications by IOUs, which signal both BABA-compliant equipment preference and a willingness to accept federal financing terms (Updated 2026-05-06).

A fourth category emerged in mid-July: commissions turning affordability scrutiny onto the cost-recovery mechanism itself, not just new capex requests. Indiana’s URC opened a joint review of authorized-versus-earned ROE (AEP’s Indiana Michigan Power subsidiary earns 12.6 percent against a 9.85 percent authorization) and of TDSIC, the tracker that recovers 80 percent of approved transmission, distribution, and storage capex as incurred. A tighter tracker regime does not remove distribution capex demand, which is structural, but it can slow the pace of approvals and shift order timing, and TDSIC is a widely studied template other states are likely to copy. Category (3) above already has a live illustration: DOE’s Energy Dominance Financing office closed a $3.26 billion loan to AEP Texas covering roughly 100 transmission projects and 2,800 miles of reconductoring tied to 41 GW of new load. Our DOE transmission loans piece has the terms. Category (3) also deserves a BABA-specific watch item: domestic-content thresholds for manufactured products step from 40 percent in 2024 to 55 percent in 2027, and for large power transformers the math does not favor the obvious choice. Cleveland-Cliffs is the only US producer of grain-oriented electrical steel, an input that runs 20 to 25 percent of transformer production cost, but its grades are priced above Nippon Steel’s imported high-efficiency steel while offering no efficiency advantage. A buyer chasing the domestic-content bonus on a federally funded project can be trading a credit for a cost and efficiency penalty on the single largest material input. The federal waiver library already holds more than 2,000 approved or proposed waivers, so budget for a waiver application as a normal step, not a fallback.

Federal financing has a second channel that cooperatives reach far more often than DOE’s, and it carries a supplier constraint DOE’s does not. The Rural Utilities Service lends rather than grants, and on August 3, 2026 it opened the Powering Affordable Reliable Technology program: roughly $410 million in Inflation Reduction Act budget authority under Section 22001, individual awards of $1 million to $100 million, forgiven up to 40 percent subject to conditions that live in the notice of funding opportunity rather than in the Federal Register notice (Docket No. RUS-26-ELECTRIC-0232). Two terms matter more than the headline. Project loans finance up to 75 percent of capitalized cost, and the remaining 25 percent must be cash or equity that “may not be derived from debt instruments,” which rules out the supplemental borrowing a thin-margin distribution cooperative would normally reach for. An award also makes the recipient an RUS borrower, so 7 CFR 1728.70(a) applies: for the material categories the RUS List of Materials covers, a borrower may purchase only listed materials or materials holding a current RUS technical acceptance that meet the Buy American requirement. Federal money and a federal qualified-products list arrive together, which is a supplier-qualification question rather than a financing one, and it belongs in the specification rather than in the award. Letters of interest are evaluated on a rolling basis in the order received between September 8 and October 9, 2026, so the opening date governs rather than the closing one. Our read on the PART program works through eligibility, the storage condition, and the equipment consequences. Updated 2026-08-22.

Reading PJM Market Design Signals

State governor frameworks tell a procurement team what regulators will allow utilities to spend. PJM market design signals tell the team something different: who will actually buy the equipment, through which channel, and on what timeline. For distributors operating in the PJM footprint (PA, NJ, MD, DE, OH, WV, KY, VA, NC, IL, IN, MI, DC), PJM market design procurement intelligence has become the single most useful read on where 2027 and 2028 demand lands. Four developments from spring 2026 define the current picture.

Cycle 1 quantifies the demand wave. On April 29, 2026 PJM disclosed the results of its first reformed interconnection cycle: 811 projects totaling 220 GW, the largest single-cycle queue in U.S. RTO history. The mix is the signal. Gas-fired generation leads at 106 GW (48 percent), battery storage follows at 67 GW (30 percent), and nuclear appears at a record-high 18 GW. The reformed first-ready-first-served process targets a one-to-two-year review versus the prior four-to-seven-year wait, and it lands on top of capacity prices that have run to $329 per MW-day, roughly ten times the level of two years prior. Every gas block and storage project in that queue pulls step-up transformers, collector-substation gear, HV breakers, and protective relays. With large power transformer lead times still running two to four years, the binding constraint is not whether the demand is real. It is whether procurement decisions get locked early enough to clear the supply window. Our Cycle 1 equipment analysis breaks the mix down by equipment class.

Utilities are refusing the counterparty role. On its Q1 2026 earnings call, FirstEnergy CEO Brian Tierney called PJM’s planned reliability backstop auction flawed and stated plainly that the company “will not sign contracts where our companies take commodity risk on generation and energy.” That is a major investor-owned utility publicly declining to be PJM’s collection agent for data-center capacity costs. The procurement read is that generation-side equipment demand is shifting away from the traditional utility channel and toward independent power producers and their EPCs. Distribution-side spend (substation upgrades, transformer replacement, hardening) stays with the utility because it carries the lowest political risk. The PJM inflection analysis covers the backstop redesign pressure in detail.

FERC is now policing vertical integration. In late April 2026, PJM’s independent market monitor, Monitoring Analytics, urged FERC to reject Hull Street Energy’s acquisition of two gas peakers totaling 1,267 MW from Rockland Capital, arguing that Hull Street’s parallel data-center development business creates an incentive to withhold capacity from the wholesale market. The dollar figure is small against PJM’s roughly 180 GW of installed capacity, but the ruling sets precedent. If FERC permits the structure, expect a wave of generator and data-center co-location deals that route equipment procurement through private substations and customer-owned switchgear rather than utility interconnection. If FERC blocks it, the conventional utility procurement channel holds. Procurement teams should track the docket outcome as a fork in their channel forecast.

The customer-pays template is spreading. While PJM utilities, governors, and the market monitor argued over who pays, OG&E and Google produced a clean answer. On April 30, 2026 OG&E announced it will serve three new Google data centers in Oklahoma under a structure where Google covers 100 percent of grid-connection costs, commits to pay contracted costs regardless of usage, and brings two solar facilities for capacity, all under a new large-load tariff that shields existing ratepayers. The model is replicating across jurisdictions. In May 2026 the Oregon Public Utility Commission approved a Portland General Electric framework built on the same customer-pays principle, and PJM moved up its backstop auction while urging states to write ratepayer-shield rules of their own. For municipal and cooperative procurement officers fielding hyperscaler inquiries, the practical effect is that the customer, not the utility and not its EPC, increasingly owns the procurement decision for the substation, transformers, and dedicated feeders. That distinction matters more than it reads, because the obvious assumption runs the other way: bring an EPC in early and the contractor absorbs the supply-chain problem. Two independent primaries say the opposite. Mike Kotara, Senior Vice President and Power Market Executive at Zachry Group, described what owners are actually doing at the same Houston conference cited above: “Now, what they’re having to do is go out and buy extended scope packages…You’re doing that under maybe a limited notice to proceed that allows you to buy equipment at risk, at the owner’s risk. Obviously, the EPC can’t take on that risk, so you have two things that are working against each other” (POWER, August 3, 2026). Centuri Holdings, a major union electrical contractor, states the same division in its Form 10-Q for the quarter ended June 28, 2026: “Generally, our contracts provide that the customer is responsible for supplying the materials for their projects.” A conference remark and a filed financial statement, from unrelated parties, agree. Contractor consolidation changes who supplies the crews. It does not move equipment purchasing to the contractor. The owner is buying, earlier than it used to, under a limited notice to proceed issued before design is complete, and at its own risk. That is a larger and more expensive information gap than a buyer handing over a turnkey scope ever carried, and it is the gap this guide exists to close. Updated 2026-08-18.

Read together, these four signals point to one conclusion: in PJM, the question of who buys the equipment is being renegotiated in real time, and the answer differs by deal structure. A procurement intelligence function that maps each major interconnection, backstop filing, co-location case, and large-load tariff to its likely equipment channel will see demand shifts months before they reach a lead-time quote. That mapping, with the channel-by-channel equipment crosswalk and the lead-time compression math by project class, is the work product our procurement intelligence reports deliver.

The newest turn in that renegotiation is ownership. In March 2026 Alphabet closed a $4.75 billion purchase of developer Intersect Power and kept its development arm to build generation for Google’s own data centers, the first time a hyperscaler bought a power developer outright. Microsoft’s behind-the-meter gas plant in West Texas, Google’s backing of a 1.5 GW small modular reactor project in Ohio, and its $15 billion Missouri commitment extend the same move. When a hyperscaler owns or finances the generation, it becomes a direct, schedule-driven buyer of step-up transformers, high-voltage switchgear, and battery balance-of-system, competing with utilities for the same constrained capacity. Our hyperscaler-owned generation analysis traces what that ownership shift does to the equipment channel. Updated 2026-06-25.

Capacity-Market Price Signals and the Flexible-Interconnection Bargain

Market-design signals tell a procurement team who buys the equipment and through which channel. The capacity auction tells it how hard the clock is ticking. On June 30, 2026 PJM opened its 2028/2029 Base Residual Auction, with bidding closing July 7 and results posted July 14. The auction ran under a price collar of roughly $325 per MW-day at the cap and about $175 at the floor. PJM’s own filing states that without the collar, the cap for this auction would sit near $550 per MW-day. It cleared at the $325 cap (Updated 2026-07-30), a 2.5 percent decrease from the prior auction and, per PJM’s own release, the third consecutive auction to clear at the cap. Note that each of those three auctions pinned a different FERC-approved ceiling rather than a single figure: $329.17 per MW-day for 2026/2027, $333.44 for 2027/2028, and $325 for 2028/2029. Set that against a 2024/2025 auction that cleared close to $29 per MW-day. A capacity price that has now held at its ceiling for three straight delivery years is the clearest public confirmation that the PJM supply-demand imbalance is structural rather than temporary. PJM concedes the caps do not fix the shortage. They cap the pain for ratepayers, and the market monitor credits the prior two auctions’ collar with holding roughly $13.1 billion off customer bills.

The reliability math behind the price is the demand signal a procurement team should read. For 2027/2028, PJM fell 6,623 MW UCAP short of its requirement, an actual reserve margin near 14.8 percent against a target close to 20 percent, and the first time the entire footprint including self-supplying FRR areas has fallen short. Nearly 5,100 MW of the 5,250 MW peak-load increase driving that gap came from data centers. That is the demand the next wave of equipment orders chases, and it is concentrated in exactly the classes already running long: step-up transformers, medium voltage switchgear, breakers, and protective relays.

PJM is now trying to convert the price into steel. Alongside the auction, the PJM board advanced a “sprint for new capacity” to close the 2028/2029 gap and a “Connect and Manage” model for certain large loads, both slated for a FERC filing the following month. Under Connect and Manage, PJM gathers information on new large loads and shares it with transmission owners and distribution companies so they, with their state regulators, can set service-priority decisions when the system is stressed. A companion fast-track interconnection path would take up to ten requests per year for capacity resources of at least 250 MW, require backing from a state’s primary siting authority, and target a three-year completion with roughly ten months from filing to a Generation Interconnection Agreement, sunsetting at the end of 2027. That three-year build mandate collides directly with the two to three year lead times still quoted on large power transformers, and the collision is where a procurement intelligence function earns its keep. Projects that lock equipment orders early clear the supply window. The ones that wait for fast-track certainty do not.

Flexibility is the lever that decides how much gets built and how fast. On June 18, 2026 FERC ordered system operators to provide transmission service for flexible large loads, part of a broader push that put six RTOs on a 60-day clock to rewrite their large-load tariffs, and Pennsylvania released a first-of-its-kind large-load model tariff on May 18. Live pilots already show the range. Silicon Valley Power requires 100 percent control of the load-side breaker as its condition for faster interconnection to an NVIDIA data center. Portland General Electric demonstrated a 20 percent reduction in a simulated emergency. Salt River Project ran a peer-reviewed 25 percent workload ramp across a three-hour peak. EPRI’s DCFlex program spans nine demonstration sites with a five-class flexibility framework, and independent work at Boston University puts training and inference flexibility at 18 to 55 percent. The procurement read is that flexibility does not remove equipment demand. It relocates and conditions it. Silicon Valley Power’s breaker-control requirement implies utility-grade sectionalizing, protection, and remote-control gear at the data-center interface, reclosers, motor-operated switches, protective relays, and communications, not just the service transformer. Curtailment-enforcement equipment is becoming a specified line item in large-load interconnection, sitting alongside the traditional transformer and switchgear package.

Texas moved from federal show-cause pressure to an enforceable state rule days later. On July 10, 2026 the Texas PUC unanimously approved ERCOT’s NOGRR282, the first mandatory load-side ride-through requirement in the country, covering any site or co-located group at 75 MW or more where computational load makes up half or more of demand. The rule sets voltage and frequency ride-through envelopes at the point of interconnection, caps load current at 125 percent of normal during a disturbance, and gives ERCOT disconnection authority for non-compliance, with a softer standard for projects that qualify before January 1, 2028 and a stricter one after. Unlike the tariff and moratorium fights below, this is not a cost-allocation signal. It is an equipment-specification signal: protective relays with reconfigurable under-voltage and under-frequency elements, dynamic fault recorders, and UPS chains rated to hold through a 0.5 per-unit sag are becoming named line items in large-load interconnection packages rather than assumed capability. CAISO is drafting a parallel requirement and is explicitly benchmarking ERCOT, so treat NOGRR282-style ride-through envelopes as a national specification trend headed for RFQs outside Texas within the next one to two years, not a one-state rule.

For municipal utilities and cooperatives in the PJM footprint, the combined signal is a hard cost forecast and a hard operational exposure. Capacity near $325 per MW-day for 2028/2029, against $29 four years earlier, is a number to budget now. Connect and Manage hands service-priority and curtailability decisions to distribution companies and states, which means the muni or co-op inherits the operational rules and the capacity-cost pass-through without the legal and engineering staff an investor-owned utility brings to a case-by-case curtailment negotiation. Standardized tariffs, the Pennsylvania model and FERC’s June 18 order among the first, are the emerging protection. Mapping each auction result, Connect-and-Manage filing, and flexible-interconnection pilot to its equipment channel and its lead-time window is the work our procurement intelligence reports deliver.

Large-Load Tariffs Codify: Rate-Class Design as an Equipment Capex Signal

FERC’s June 18 order put six RTOs on a 60-day clock at the federal level. The states are moving faster. On July 7, 2026 the Oregon PUC approved Portland General Electric’s Schedule 96 rates in docket UM 2377, the first rates in effect under a legislatively mandated separate rate class for large loads. Oregon’s 2025 POWER Act (HB 3546) draws a hard line at 20 MW. Above it, the customer pays 100 percent of distribution-network upgrade costs, carries a 90 percent minimum demand charge on contracted capacity, and signs a contract of at least ten years, stretching to 30 years for loads of 220 MW or more. Projects above 100 MW pay an added 1 cent per kWh surcharge, and a peak growth modifier can extend cost-allocation obligations past the contract term. The approved rates run about 29 percent higher for large-load customers while residential bills fall 1.3 percent. QTS, Flexential, and Aligned in Washington County and Amazon, Meta, and Google in rural Oregon take service under the new class. PacifiCorp, Oregon’s other POWER Act utility, remains in docket UE 463 without approved rates. The customer-pays framework noted in the PJM section above has now produced effective rates, not just an approved structure.

Duke Energy Carolinas filed the counter-model in late June inside rate case E-7 Sub 1329, with a parallel Duke Energy Progress case at E-2 Sub 1380. The proposal sets a 75 percent minimum-take on contract demand for loads of 50 MW and up at 80 percent load factor, or any request of 100 MW or more, effective 2027, with ten to 15 year contracts, a 25 percent early-exit penalty, and cash or letter-of-credit collateral, but no separate data-center rate class. Against Dominion Energy Virginia’s benchmark of an 85 percent minimum on transmission and distribution demand, 60 percent on generation, 14-year contracts, and $1.5 million per MW in collateral, Duke’s terms are light. North Carolina’s Public Staff wants roughly $200 million of Duke’s $247 million grid-upgrade request assigned directly to large-load customers, the state attorney general is pushing to separate data-center costs from the general rate base, and the NCUC order is expected late in 2026.

Three templates now compete. Oregon codifies a separate class with a statutory threshold. Duke keeps large loads inside existing classes and manages risk through contract terms. Pennsylvania’s model framework, built on a but-for cost-causation standard, sits between them: it gives a utility a clean legal basis to size and order the equipment attributable to a named large load without waiting for a general rate case. The spread is national. As of June 2026, 24 states have approved at least one large-load tariff, four more have approvals pending, and legislators in more than 18 states have introduced rate-class bills. Xcel proposed a data-center rate scale in Colorado in April.

A fourth axis, more fundamental than any tariff design, opened in July. New York enacted a one-year moratorium on data-center construction above 50 MW, effective July 14, 2026, becoming the second state after Maine’s April 2026 ban to stop hyperscale interconnection outright rather than price it. That is a different tool than anything in the templates above: Oregon, Duke, Dominion, and Pennsylvania all let the load connect and argue over who pays, while Maine and New York simply pause the interconnection pipeline that would have driven the transformer, switchgear, and protection-relay spend. Our moratorium demand-displacement analysis covers where that paused demand goes. It does not disappear under a freeze, it relocates to the tariff states, so a moratorium is a near-term demand-suppression signal for the frozen state and a tailwind for its neighbors. Virginia’s own tariff proceeding has now resolved, and it produced a fifth template. Dominion’s Rider T1 case, Case No. PUR-2026-00056, was heard July 14 to 16 and decided in a final order on July 31, 2026. The SCC approved an amended 12 Coincident Peak allocation with a minimum demand adjustment, cutting the projected Rider T1 increase for a typical residential customer from $2.90 to $0.94 per month, and separately ordered Dominion to file amendments to its line extension policy requiring a mandatory contribution in aid of construction for “direct connect” transmission facilities, meaning the substations and lines connecting a large load to the bulk system. The instrument is the important part. Virginia reassigned substation cost through an ordinary line extension tariff inside a rate case rather than through a statute or a dedicated rate class, which makes it the fastest and most copyable of the templates on this page. The proposed amended policy is due in a new docket within 90 days of the order, on October 29, 2026, and a separate status update on cost assignment between Dominion and neighboring distribution companies including ODEC is due at 120 days, on November 28. Our read on the Virginia CIAC order covers the definition fight and the cooperative exposure. Updated 2026-09-10. Add a moratorium-versus-tariff tag to state tracking alongside the existing three templates, and watch for a second state to copy New York’s freeze rather than the tariff route.

The procurement read is that tariff design is now a forward indicator for equipment orders. A POWER-Act-style class converts “will this load materialize and pay for the substation” from an open question into contracted revenue before the utility issues an RFQ, so expect faster and more confident transformer, switchgear, and feeder orders tied to specific interconnections in codified-class states. Where allocation is still contested, as in the Carolinas until the NCUC rules, equipment orders tied to large loads carry more risk of delay or descoping. For municipal utilities and cooperatives without 20 MW customers of their own, the second-order effect dominates: tariff certainty releases IOU capital, and those IOUs compete for the same transformer, switchgear, and HV cable capacity every smaller distributor needs. Rate-design progress at PGE and Duke is, indirectly, a lead-time signal for everyone else. Intelligence teams should add large-load rate-class filings to their PUC docket scans and tag each key jurisdiction by template: codified class, contract-term-only, or but-for framework. The June cost-allocation vote at FERC and the real drivers behind rising electricity prices trace how the same fight reads on the ratepayer side. Updated 2026-07-11.

A federal layer joined the state templates on July 21, when the House Energy and Commerce Committee ordered the Ratepayer Protection Act, H.R. 9340, reported by a vote of 52 to 0, narrowed from any large load at 100 MW or more to data centers only. The bill amends PURPA Section 111(d), a “consider and determine” standard that forces every state PUC to open a large-load cost-allocation docket without mandating an outcome, so it adds procedural pressure behind the state templates above rather than replacing them. Two provisions matter more than the headline for buyers below the transmission system (updated 2026-08-06): the bill enumerates “generation, transmission, or distribution” upgrades by name rather than saying “grid upgrades,” which is what puts substation and feeder scope on the developer, and it requires financial assurances or contributions from the large-load customer before any such upgrade proceeds. It also does not touch wholesale network-upgrade allocation, the gap five states flagged to FERC in docket EL26-67 on July 17, where PJM’s Independent Market Monitor now attributes 46 percent of the last four capacity auctions to data centers. Track the Ratepayer Protection Act’s House floor vote and any Senate companion alongside the state dockets above.

A fifth instrument arrived in two pieces, and it is the first one that writes an equipment obligation rather than a payment obligation. PJM’s board directed both on July 27, 2026, and PJM then filed them separately: the Reliability Backstop Procurement on July 31 in docket ER26-3380-000, and the Interim Resource Adequacy Service, with a Large Load Registry, on August 13 in docket ER26-3515-000, with a requested effective date of October 12 and comments due at 5 p.m. Eastern on September 3. Read them as two filings, because PJM tells FERC to. Its transmittal states that “the two filings are independent of each other” and that the backstop “can and should proceed as requested in this filing even if the Commission wishes to further examine portions of an IRAS filing that may be pending before the Commission at the time.” Trade coverage has repeatedly run the two together under the backstop’s docket number, which means anyone tracking curtailment obligations by docket is watching the wrong one. Under IRAS, new large loads that connect without dedicated supply must reduce load or switch to on-site backup when the system nears emergency conditions, with individual utilities writing the curtailment rules subject to their own state and local regulators, and PJM standing up a registry that captures each facility’s location, ramp schedules, and capacity details. A registry that collects ramp schedules implies verification, and verification at that resolution means revenue-grade metering at the point of delivery, communicating relays and recloser controls with a SCADA path to act on a dispatch signal, and transfer equipment able to move the block. That is recurring distribution-side spend attached to every large load in the footprint, and it does not end when the substation is energized. Duke’s Customer Protection Plus carries the same temporary-curtailment provision as a contract term, so a vertically integrated utility’s tariff and an RTO’s market rule reached the requirement independently in the same month. The registry also settles who counts, and the test is wider than the data-center framing implies. PJM’s board framework defines a Large Load as end-use customer load with a cumulative peak of at least 50 MW at a single site behind one or more delivery points within a one-mile radius, and it applies that test to growth as well as to new construction. If a 30 MW customer ramps to 60 MW after June 1, 2027, the 30 MW of new load counts as new Large Load megawatts, and the example is the framework’s own. Two dates follow that the trade coverage has not carried. Large loads already in service before June 1, 2027 must register by March 1, 2027, and every site must be registered before its in-service date and before PJM finalizes the load forecast for the auction the load first appears in. So a municipal utility or cooperative with a single industrial customer on a growth curve inherits a registration obligation, a ramp-schedule reporting obligation, and the metering and telemetry scope behind both, without ever signing a data center. Add a sixth column to any large-load tracking sheet, next to codified class, contract-term-only, but-for framework, and moratorium: whether the jurisdiction attaches a curtailment and telemetry obligation to service. Our read on the PJM curtailment rule works through the registry fields and the equipment each one implies. Updated 2026-08-19.

The backstop half of that board directive is a calendar rather than an obligation, and it repays reading because almost none of it is fixed. PJM asked FERC to act by September 29, 2026, intends to open a three-week offer window on September 30 if it does, and expects clearing to conclude six weeks after that window closes, which puts commitments in early December against a term running to the 2042/2043 delivery year. That October 21 close now circulating in coverage is arithmetic, not filing text. The proposed tariff language commits only to a window that “shall open for three weeks in 2026” and says PJM “shall post specific dates on the PJM website” before it starts, so every downstream date moves one for one with FERC’s docket clock. A second widely repeated figure deserves the same scrutiny. The 2028/2029 auction cleared 6,831.3 MW short at a collared cap of $325 per MW-day, and the $555 per MW-day figure attached to the backstop is not a raised ceiling on any offer. It is a maximum willingness to pay applied to the MW-weighted average of levelized procurement cost across all selected offers over the full term, which PJM describes as a budget approach “not restricting Sell Offers to a hard price cap.” Individual offers can clear above it. The supply side deserves the discount this guide already applies to load pipelines: the request for information run by CRA drew over 100 GW of prospective supply against a 6,831.3 MW target, about fifteen to one, and PJM cites that ratio as evidence the procurement will be competitive. Interest is an option book on the supply side exactly as a queue is on the demand side. Anything that clears has to be deliverable for a 2028/2029 delivery year, so working back through the generator step-up unit, medium-voltage switchgear, protection, and the generation-tie substation puts ordering decisions in the first quarter of 2027 at the latest, on equipment whose quoted lead times routinely run past it. Expect winners to be projects already holding equipment slots or interconnection positions, and expect a high clearing price against thin volume, which is the pattern April’s procurement already set. Updated 2026-08-19.

A sixth instrument is the one most of these templates assume away: the cooperative and municipal case, where the counterparty reviewing the contract is a state commission rather than an RTO. Kentucky is the working example. Big Rivers Electric, a G&T cooperative that owns 936 MW of generation, was named on July 29 as the wholesale provider for a $100 billion campus at the DOE Paducah Site supporting up to 1.8 GW of utility capacity, with a distribution co-op as the retail counterparty and the power service agreement subject to Kentucky PSC approval. The commission is already reviewing a separate Big Rivers data center contract, Case No. 2026-00115, whose terms are public: a 15-year agreement with take-or-pay obligations for the first six years, MISO wholesale pricing, and a mandatory curtailment provision that drops the load before residential and business customers. Two lessons for a tracking sheet. First, in commission-reviewed jurisdictions the contract structure becomes readable by every other buyer, which is the opposite of the arbitration-and-litigation path other states are on, so add the source of review as its own column. Second, the commission’s authority runs only to whether rates and service are fair, just and reasonable, meaning approval says nothing about equipment deliverability against the in-service date. Our read on the Paducah deal works through what the generation mix and the curtailment term imply for utility-side scope. Updated 2026-08-04.

Every column above assumes the load being tracked is real, and the Q2 2026 earnings window showed that assumption doing more work than it can carry. Six utilities published large-load figures in the same weeks and none of them counted the same way. PG&E disclosed a 12.7 GW pipeline against roughly 1.8 GW it expects to serve by 2030, about 7:1. CenterPoint submitted more than 17 GW to ERCOT Batch Zero expecting around 14 GW to qualify, about 1.2:1. DTE reported 2.4 GW executed against another 5 to 6 GW of development pipeline. Xcel split its disclosure into three labeled tiers, roughly 2 GW contracted now and roughly 4 GW contracted by end of 2027. Exelon identified 11 GW of high-probability load with only 4 GW behind signed Transmission Security Agreements and about $1 billion in collateral. AEP published the largest and cleanest-looking number, 69 GW contracted through 2030, which its own earnings call describes as supported by a combination of fully executed ESAs and LOAs, with 45 GW of it sitting in Texas under LOAs while the take-or-pay electric service agreements behind the company’s headline customer-benefit claim sit at its vertically integrated utilities elsewhere. The tracking-sheet lesson is a seventh column, and it belongs left of the other six: record which conversion tier a GW figure describes and what the project had to post to be counted, because a deposit-gated queue converts near unity while a free queue converts somewhere between one in five and one in ten. Our guide to discounting data center load pipeline vs contracted figures works through all six disclosures and the four questions that make any of them usable. Updated 2026-08-15.

The seventh column tells you how much to discount a number. An eighth tells you whether anyone outside the utility can check it, and Georgia is the first state to supply one. As part of a joint stipulation with Public Service Commission staff filed on August 26, 2026, Georgia Power agreed to publish a summary of its service agreement with OpenAI within ten days, covering 3,200 MW phased across 2028 to 2032 with up to 1,000 MW of committed flexible demand response. Every other large-load proceeding on this page produces a tariff, meaning a rule about who pays. This one produces a document about what one named counterparty signed. Two cautions belong in the column, because both are already being reported loosely. The forward-looking commitment is a semiannual portfolio report rather than a summary per contract, so Georgia has not created a standing publication regime for future agreements. And the developer-pays outcome in that contract is not new: a Georgia PSC rule adopted in April 2025 already required data center contracts above 100 MW to be filed with the Commission 30 days before execution, so the approval confirms an existing rule rather than creating one. Track the source of disclosure alongside the conversion tier, because a figure a regulator can force into the open is a different quality of input from one a press release supplies. Our read on the first data center utility contract disclosure covers the three terms worth reading the summary for. Updated 2026-08-31.

A ninth column asks who set the tariff and who reviews it, and TVA is the case that makes it necessary. On August 20, 2026 the TVA Board approved a distinct data center rate class effective October 1, applying a Capacity Commitment Charge to new or expanding data center load above 5 MW, framed as recovering incremental capacity cost rather than spreading it across the general rate base. TVA projects an all-in average billing impact near 10 percent for those customers, phased over three consecutive fiscal years, and the same action tightened contract-demand rules to match actual usage, added a stability contract option for existing manufacturers above 5 MW, and updated board policy for new power requirements above 100 MW. What separates it from every template above is the counterparty structure. TVA had 215 customers last year: 153 local power companies distributing to more than 10 million people, and 62 large industrial and federal customers served directly. TVA officials put the share of data center load served through those local utilities at 99 percent, so the charge lands on a municipal or cooperative distributor’s wholesale bill and reaches the data center only through a retail term that distributor wrote itself. There was no state commission review and no public comment period, and the process ran as direct conversations between TVA and the power companies with the Tennessee Valley Public Power Association carrying the collective position. Two consequences for a tracking sheet. First, record whether the entity that will buy the interconnection package was a party to the tariff or a third party to it, because that predicts whether the terms are ever legible to the buyer holding the equipment schedule. Second, record what the tariff has actually published: as of September 3, 2026 TVA had released neither the rate schedule nor the cost of service study, and the contract minimum and early-exit terms, the provisions that convert a tariff into a dated equipment order, remain unstated. Our read on the TVA data center rate works through the 5 MW threshold, the reclassification risk in the word “expanding,” and the questions worth putting to a wholesale account team before October 1. Updated 2026-09-03.

From Intelligence to Action

Raw data becomes intelligence when it changes a decision. Lead time data that confirms your existing approach is information. Lead time data that reveals a manufacturer falling behind schedule, prompting you to accelerate an order or switch suppliers, is intelligence.

The procurement intelligence function should produce two outputs: a regular market briefing (weekly or biweekly) that keeps the team informed, and ad-hoc alerts when time-sensitive signals emerge that require immediate action.

One filing type deserves a standing place on that watch list, because it is routinely the earliest readable date on a large program and it is not the one most teams track. On a multi-region transmission build, project approval is not the gate. Cost recovery is. MISO’s board approved its $21.8 billion Tranche 2.1 portfolio in December 2024 and could not order the supporting facilities built inside PJM, because its authority as transmission provider stops at its own system. That work sat until FERC approved a cost recovery and funding framework in docket ER26-1538 on August 14, 2026, which also named the utilities that will build it and, because the facilities classify as supplemental projects in PJM, exempted the work from competitive solicitation. Our read on what the FERC MISO Tranche 2.1 cost recovery PJM order buys works through the three equipment categories it releases. The general rule: when you are tracking a planned program, set the alert on the cost-recovery docket rather than the approval, and read the named builder as your counterparty. Updated 2026-09-04.

Building this function in-house means assembling and re-checking the same public sources every week. DistroForge Insider turns them into a standing procurement intelligence feed: lead times, supplier capacity, and regulatory signals in one place.

This guide is updated as new research is published. Last reviewed September 4, 2026.

In This Guide Series