Utility planning was built for a predictable world.
The grid has never been more complex, or the decisions harder to defend.
Utilities are managing the largest capital cycle in their history, on a grid that has stopped behaving predictably. Load forecasts carry more uncertainty than ever. Interconnection queues are stalling the capacity that resource adequacy plans depend on. All of it pushes upward on rates at the moment regulators and consumers most expect energy to stay affordable. When the answers come from separate consultants and point tools, they don’t reconcile. An IRP filing and a rate case say different things, and no one can explain the gap.
Utility executives are being asked to defend, optimize, and grow the grid simultaneously, on timelines and under uncertainties their planning processes weren’t built for.
Protecting the largest rate bases in utility history against regulatory scrutiny, ratepayer pressure, and cost-recovery uncertainty. The rate case has become a filter for which loads enter the queue, not just a cost-recovery exercise.
Marginal resources, congestion patterns, and price-formation logic have all changed. The control-room mental model (dispatch the next-cheapest unit until load is met) was built for a market that no longer exists.
Building generation and transmission against interconnection queues, load forecasts, and fast-track processes that often disagree with each other. In some markets by 2030, the risk may be oversupply as much as undersupply.
Each pressure lands differently across utility types. A vertically-integrated IOU feels all three acutely. A wires-only utility feels Defend and Optimize more than Grow. A distribution-only utility feels Defend most, where rate-base recovery and reliability dominate. A G&T co-op feels Grow as member-load reality, not queue dynamics. Municipal and public-power utilities land somewhere on the same map, weighted by whether they own generation, wires, or both. The taxonomy is shared. The intensities aren’t.
Each pressure also has a planning gap underneath it. Different as they look, all three eventually route through the same work: planning the grid, designing it, connecting it, and optimizing how it runs. And underneath every one of them is the engineering that turns a decision into something you can actually build. The next three pages take each one in turn — what Enverus‘s research desk is seeing, and what the gap looks like underneath.
The conventional read is that utilities respond to load growth by raising rates and filing rate cases. The growth in data center interconnection requests creates pressure on existing ratepayers, and rate cases recover the cost of new infrastructure from the loads causing it. That story is true, but it’s incomplete.
A different pattern emerges when you carefully read recent rate filings. Utilities aren’t only redesigning rates to recover cost. They’re redesigning rates to screen which loads enter the queue in the first place.
AEP Ohio proposed a new data center tariff in 2024 and the Public Utilities Commission of Ohio finalized it in 2025. Enverus Intelligence® Research (EIR) found that within months of the rate taking effect, AEP Ohio filings showed a roughly 50% drop in data center connection requests. The rate recovered cost and filtered speculative load out of the queue before it could clog the planning process.
That pattern isn’t isolated. EIR analysts examined 94 large-load tariffs across 36 utilities and found new structural components consistent across filings, including capacity ramp rates that force developers to commit to load profiles, collateral and credit requirements that separate funded projects from optionality plays, and long-term contract minimums that make speculation expensive. These are queue-filtering mechanics built into the tariff itself, not cost-recovery mechanics.
The same tariffs reshape developer economics from both ends. EIR modeled a 100 MW data center paying nearly $10 million more in first-year costs under AEP Ohio’s data center rate than under general service, while a flexible 100 MW load in Southern California Edison territory could save more than $20 million a year by moving to real-time pricing. The design rewards loads that can prove they’re real and flexible, and prices out the ones that can’t.
large-load tariffs analyzed across the Lower 48
drop in AEP Ohio data center requests after rate finalization
annual savings available to a flexible 100 MW load under RTP (SCE territory)
added first-year cost for a 100 MW facility under AEP Ohio's new data center rate
The traditional rate-case mental model is to recover the cost of new infrastructure from those causing it, while the emerging mental model is to use rate design itself to gate who gets to cause new infrastructure. That shifts where the planning work happens, and which team owns it. Tariff design is no longer downstream of integrated resource planning. In some filings, it’s becoming a primary planning tool. And because reliability has become the most powerful capex-recovery argument available to utilities in front of a PUC, the tariffs that successfully filter speculative load protect ratepayers and the reliability defense that anchors the next rate case.
The data center tariff is the most visible case, but the shift is bigger than data centers. Rate design is becoming a planning instrument inside the rate case itself, and for any utility, the rate case is where you decide which growth you’ll defend and which you’ll price out.
The question shifts. It’s no longer how do we recover the cost? It’s which loads are real, and which are filtering themselves out?
For most of the industry’s history, price formation followed fuel cost. Combined-cycle gas set the floor, while peakers set the ceiling. The merit order was a stack of heat rates, and the control room’s job was to dispatch the next-cheapest unit until load was met.
That description still fits parts of the day, but no longer the parts that matter most.
EIR analysts examined ERCOT 2025 settlement data and found a market that has reorganized itself. Combined-cycle gas remains the baseline marginal resource, setting the clearing price 63% of the time at an average of $35/MWh. But batteries, a small share of the energy mix, are setting the price 23% of the time, capturing a $21/MWh flexibility premium when they do. System price formation has decoupled from local nodal realities. The system marginal price is often low; congested nodes hit their physical limits and clear at very different prices on the same interval.
The structural shift isn’t about which fuel is cheapest, but who is bidding most sophisticatedly. Ten firms set prices for more than half of all intervals in ERCOT and a specialized cohort of storage operators sets the price during the highest-volatility windows. These are portfolio operators bidding opportunity cost, not utilities dispatching units.
of ERCOT intervals where batteries set the clearing price
average flexibility premium captured by storage when it sets the price
of ERCOT intervals where combined-cycle gas still sets the price ($35/MWh avg)
that set prices for more than half of all ERCOT intervals
ERCOT is the most pronounced case today. But the underlying logic (opportunity-cost bidding by sophisticated portfolio operators displacing fuel-cost dispatch on the intervals that matter most) is not unique to ERCOT. It follows battery economics and market structure, both of which are developing across ISOs. ERCOT is the clearest window into where that development leads.
For market participants, knowing what each unit costs matters less than knowing what each operator believes their unit could earn somewhere else. Bid behavior is the game, not heat rates alone.
For utilities planning generation and transmission, the read is even sharper. Locational congestion has become a separate market with its own clearing logic, as opposed to a footnote to the system price. A unit dispatched at $35/MWh on system price and a unit clearing at multiples of that on local congestion in the same interval are not anomalies — they are the structure.
And for the many utilities that buy more power than they generate, this isn’t abstract. The same congestion and price formation show up directly in purchased-power costs, the line-item regulators scrutinize most closely.
Planning against this market is a different discipline than planning against the old one. The starting point is modeling where congestion forms and why, not just which fuel is on the margin. From there, understanding how the most sophisticated operators bid matters as much as knowing what their units cost. The system price and the locational price have to be held in view at the same time, because they are no longer the same number. The utilities that adapt will site generation, value flexibility, and defend transmission against the market that is actually clearing. The ones that don’t will keep planning against a merit order that stopped setting the price.
The question shifts. It’s no longer what does this unit cost to run? It’s what is the market actually clearing, where, and why?
Every utility executive has heard the queue paralysis story. Across PJM, MISO, SPP, and ERCOT, interconnection queues are stretching years. Load forecasts climbing. Fast-track processes emerging to relieve the bottleneck. The implicit assumption inside most planning conversations is that even with fast-tracking, supply will struggle to meet demand by 2030.
EIR’s modeling tells a more complicated story, and so does EIR’s analysis of the load forecasts driving it.
By 2030, EIR projects PJM, MISO, and SPP will be better supplied with power than they are today. PJM’s Reliability Resource Initiative cleared 11.8 GW across 51 projects, with natural gas making up 69% of cleared capacity. MISO’s Expedited Resource Addition Study has cleared roughly 12 GW across two cycles. SPP’s combined ERAS and HILL processes could push the region’s 2030 reserve margin to 19–24%. These aren’t proposals. They’re projects already moving through accelerated queues with commercial operation dates in 2030–2031. The study outputs that move those projects through the queue need to be auditable and repeatable, or the cost allocation decisions they underpin won’t survive intervenor scrutiny.
On the demand side, EIR’s Load Forecast Benchmarking analysis found a similar problem. ERCOT’s and PJM’s own published load projections imply data center investment in just those two regions would need to exceed total planned U.S. hyperscaler capital spending. The numbers don’t add up. Their combined projections would require roughly $2.67 trillion in new data center investment through 2030, 35% more than the $1.96 trillion consensus estimate for all U.S. data center growth capital in the same period. ERCOT’s forecast compounds the problem further, assuming near-full execution of almost every announced hydrogen project. EIR’s analysis puts the likelihood of that at close to zero.
The risk flagged is the opposite of undersupply. If load forecasts overshoot, as EIR’s analysis suggests they do in PJM and ERCOT, and fast-track processes deliver as planned, some regions face material oversupply by 2030. That matters for three reasons:
cleared by PJM's RRI fast-track process across 51 projects
projected 2030 SPP reserve margin under fast-track scenarios (EIR's modeling)
implied data center capex required by ERCOT and PJM load forecasts through 2030 (35% above total U.S. consensus)
Most utilities are being asked to assume the queue is the bottleneck and build to meet projected load. That frame may produce the wrong answer in some regions, not because the data is bad, but because it only admits one scenario at a time. And this isn’t only an ISO-market problem. Any utility building its capital plan around a single, bullish load forecast carries the same exposure if the demand doesn’t arrive. Better planning isn’t faster planning. It’s planning that holds two scenarios at once, supply-tight and supply-loose, load-forecast-correct and load-forecast-inflated, and stress-tests which one is actually unfolding.
The question shifts. It’s no longer can we build enough fast enough? It’s what does ‘enough’ actually look like, and what happens if we get the answer wrong in either direction?
The Defend pressure asks utilities to redesign rate structures faster than rate cases historically allow, and to use those structures to do work they were never built to do. Queue filtering.
The Optimize pressure asks utilities to read a market whose logic has changed. The merit order has shifted from a stack of heat rates to a behavior pattern across a small number of sophisticated portfolio operators.
The Grow pressure asks planners to choose between two contradictory scenarios, supply-tight or supply-loose, load-forecast-correct or inflated, and commit to one. That decision sits on capital plans with twenty-year tails, where a wrong assumption about coal retirement timing or generation build sequencing compounds across every subsequent IRP cycle.
Each pressure has a different surface, but underneath they share the same problem.
The data, models, and workflows utilities use to make planning decisions were built for a slower, more predictable era. They were built by function: generation here, transmission there, regulatory affairs over there, market analytics somewhere else entirely. They were built to support decisions that happened on annual or multi-year cycles. The three pressures of the current decade don’t move on those cycles, and they don’t respect the boundaries between those functions.
A new tariff filing is a regulatory document with implications for the interconnection queue, which has implications for transmission planning, which has implications for generation siting, which has implications back into the next rate case. A single decision touches four teams whose data lives in four systems whose analytical frameworks weren’t designed to talk to each other.
Faster versions of existing tools won’t solve this. What’s needed is analyst-grade, cross-functional intelligence built for decisions that span the three fronts at once.
This is the kind of problem the next decade will reward solving.
The previous pages described what EIR’s research desk has been seeing in tariff filings, ISO settlement data, fast-track queue clearances, and the gap between published load forecasts and what independent modeling suggests is realistic. The research and the analytical tools come from one place, organized around the same three fronts. Here’s what that looks like in practice.
July: Full summer heat arrives. Both analog years skew above-average load rather than flat average, reinforcing a firmer demand outlook. The 2023 analog stands out as the warmer of the two, with one to two Chicago days above 95°F alongside below-average wind — supporting higher net load risk. D.C. failed to reach 95°F in either analog year, which tempers the case for sustained, broad-based Mid-Atlantic heat-driven upside. Congestion risk is front-loaded into the first half of the month, adding a further layer of price asymmetry to an already bullish temperature setup.
On the Optimize front, the finding was that price formation has decoupled from fuel cost. A small number of sophisticated portfolio operators bidding opportunity cost now set the price during the intervals that matter most, and local congestion has become a separate market with its own clearing logic.
That analysis draws on marginal resource data, congestion patterns, nodal price formation, and operator behavior tracked across ISOs. The same view is available to your trading desks, planners, and operations teams. The market reorganized itself, and the intelligence to read it has reorganized with it. The picture of who is actually setting prices, and why, is already there.
The Grow finding was that planning under a single-scenario frame produces the wrong answer in some regions, and that both the supply side (fast-track queue acceleration) and the demand side (inflated load forecasts) contribute to that risk.
Two complementary data layers sit underneath this work.
The first draws on capacity potential by substation owner, hyperscaler land-bank tracking, and risk-adjusted scenario modeling. It’s the same dataset your transmission planning team can run their own scenarios against.
The second is Enverus’s interconnection study automation, validated by MISO, SPP, and PJM and having processed more than 500 GW of interconnection studies. FERC Commissioner David Rosner, in letters to all six ISOs (ferc.gov, March 2025), highlighted the MISO implementation specifically, noting it reproduced a large interconnection cluster study that took nearly two years to complete in just 10 days, arriving at largely similar results. Your projects’ commercial operation dates depend on outputs that come from this work. In SPP and MISO, the queues processing interconnection requests are increasingly running on it.
The work each utility does on each front is its own, but the intelligence underneath it doesn’t have to be built from scratch.
The decisions described across these pages are not theoretical. They are happening, in real filings and real planning cycles, inside organizations like yours.
The most useful thing this document can do isn’t to argue. It’s to give you three questions worth sending to the people on your team who are closest to the work, including VPs of Planning, T&D, Regulatory, and Trading. See what comes back.
What share of our recent interconnection requests would survive a tariff structure designed to filter speculative load?
If your team has not modeled this, the answer is informative on its own. If they have, the comparison to peer utility filings, across 94 tariff structures in this analysis, is the next conversation.
Does our operational intelligence tell us who is actually setting prices in our interconnected market, or are we still reading the system marginal price as the whole story?
The system price tells you one story. The nodal picture, the operator behavior, and the congestion pattern tell another. The gap between them is where the real market structure lives.
Which substations in our footprint have the highest hyperscaler land-bank exposure? And do our transmission planning decisions reflect that view, or are we still planning to load growth assumptions that independent benchmarking suggests are materially overstated?
The hyperscalers already know where they’re building. The question is whether your planning team has the same map, and whether the load forecast underlying your capital plan has been stress-tested against a scenario where ERCOT and PJM projections prove as inflated as the benchmarking analysis suggests.
The planning playbook for this decade isn’t written yet.
Defend, Optimize and Grow are three pressures that don’t sit inside any one team’s job, and won’t resolve on the cycles utility planning was built around. The utilities making the most defensible decisions in the next ten years won’t be the ones with the most data. They’ll be the ones whose data, analyst work, and workflows are organized around the actual shape of the decisions they have to make, with a single intelligence layer beneath planning, interconnection, operations, and regulatory work, so the decisions that cross all four rest on one auditable picture. Analyst-grade context. Cross-functional intelligence. Stress-tested against the disagreements that matter.
EIR is the intelligence layer built for that work — and the work is built for this decade.
The utilities that write their playbooks first will define what the next decade’s job description looks like for themselves, for their regulators, and for their customers.
Talk to a utility expert about your load forecast, your capital program, or your interconnection backlog.
About Enverus Intelligence® | Research, Inc. (EIR)
Enverus Intelligence® | Research, Inc. (EIR) is a subsidiary of Enverus that publishes energy-sector research focused on the oil, natural gas, power and renewable industries. EIR publishes reports including asset and company valuations, resource assessments, technical evaluations, and macro-economic forecasts and helps make intelligent connections for energy industry participants, service companies, and capital providers worldwide. See additional disclosures here.
Let’s get started!
Let’s get started!
We’ll follow up right away to show you a quick product tour.
Ready to Subscribe?
Ready to Get Started?
