Comment on the Draft Strategic Energy Assessment 2026-2032

Filed with the Public Service Commission of Wisconsin on September 24, 2026, in docket 5-ES-113, Strategic Energy Assessment 2026-2032. The comment is on the record. The public comment period ran from June 25 to September 28, 2026.

My name is Garold Dalton. I live in Milwaukee, Wisconsin, and I am a residential customer of We Energies. I began my working life as a maintenance mechanic and then an electrician at a foundry. I went back to school for a degree in civil engineering and practiced for several years before starting an information technology services company, which I ran from 1996 to 2014. Since then I have done programming work in artificial intelligence and studied its mathematics. I retired in 2022.

I own WEC Energy Group stock, the parent company of We Energies. I also own stock in Microsoft and Meta, both building data centers in Wisconsin, and in Google and several semiconductor companies. Those technology holdings are a significant part of my portfolio.

The draft assessment evaluates Wisconsin’s electric supply against four goals, and the first is adequacy. My main comment is about the measurement of adequacy. The assessment has a floor but no target or maximum.

The floor is the 14.5 percent planning reserve margin the Commission set in docket 5-EI-141 in 2008, PSC REF# 102692. It is an installed capacity figure derived from a loss of load expectation of one day in ten years. The order says a utility may reasonably rely upon it in long term planning. The floor tests one thing, whether the state is short of capacity, and nothing else about adequacy.

Planning is uncertain when using only a minimum reserve margin. Historically, we have:

1. 2008: the actual reserve margin was 26.6 percent, on an installed capacity basis.
2. 2013: the 2012 assessment forecast 21.9 percent on the same basis, against the 14.5 percent floor, and said the state was “enjoying sufficient capacity.”
3. 2014 and 2016: the assessments said Wisconsin was “experiencing a surplus” and “currently has surplus capacity.”

For summer 2024 the accredited margin was 18.5 percent against a MISO requirement of 8.4 percent, on an unforced capacity basis.

What results occurred during this time?

1. From 2013 through 2024 the state retired more capacity than it added, and installed capacity fell.
2. Retail sales were mostly unchanged over the same years, and the fixed cost of the surplus was spread over flat sales.
3. Wisconsin’s residential rate has been second only to Michigan’s in the Midwest in every year since 2014 but one, on this draft’s own Table D-1.

Now, Table 1-1 of this draft forecasts a 15.71 percent increase in peak demand in 2028 alone, and Table 2-3 shows resources below MISO’s requirement in 2028 and 2029, on an unforced capacity basis. To meet this demand, build would already need to be underway.

The floor is not predictive because it cannot differentiate surplus from being on course. Above it, no build. Below it, build. But build requires lead time, four to eight years in my estimate. According to Lawrence Berkeley National Laboratory, the median project reaching commercial operation in 2025 took 61 months from its interconnection request.

The floor needs to be supplemented. I would like the assessment to measure against a target rate of demand growth, and against a band on excess capacity.

Low, positive growth in demand paces new generation onto the system at a rate the build cycle can follow. Growth slowly consumes excess capacity, so surplus returns to target without anyone deciding to stop building. And idle capacity under falling demand is a failure rather than adequacy, because idle capacity forces rates higher than the system’s cost justifies.

The record above is also why the target belongs on demand and not only on capacity. Renewables were built and other plants were retired early, and with demand flat the cost of both landed on the same customer base. Three things follow from a low, positive rate of demand growth that no band on capacity produces by itself.

1. Underlying demand growth renews infrastructure continuously, because a fleet is renewed by building and building follows load.
2. It grows the customer base, and a growing base spreads the fixed cost of the system over more sales, which is the one direction in which rates can fall without cutting cost.
3. Growth in the demand for energy is what a state adding employers and households looks like. It is both the sign of a more dynamic state and part of what produces one.

My request has four parts.

1. State a target rate of underlying demand growth.
2. State a band on excess capacity, with the existing loss of load floor as its bottom and a ceiling above which capacity is idle.
3. Report every edition against both, with generation and grid ahead of demand by a margin within the band, and with every reserve figure naming its basis, installed or unforced capacity.
4. State the rule that follows from each reading:
   • below the floor, build or buy;
   • above the ceiling, moderate the build;
   • growth below target, take committed load;
   • growth above target, pace new large load against capacity confirmed built or firmly scheduled, through the certificates for the generation and transmission that serve it, which the Commission already issues under Wis. Stat. section 196.491.

I view setting an underlying demand growth target as the most important of the four parts. I refer to growth in the load the system serves, normalized for weather and measured after efficiency programs, which is a quantity the assessment already forecasts. Table 1-1 reports peak demand with its percent change from the previous year. When the forecast rate and the target diverge, each edition should say by how much and which rule in the fourth part applies, so that the certificate and rate dockets follow the same instructions.

I am not proposing any numbers. The draft states that docket 5-EI-161 is reviewing whether the 14.5 percent guideline remains appropriate. That is where the numbers belong, and I ask that this structure be carried into that review.

A level stops working the day it is met. The 14.5 percent margin has not changed since 2008, and the renewable portfolio standard has been met every year since 2013 and requires nothing further. A level asks whether the state has arrived. A rate asks whether it is still moving.

Ten reporting requests follow. The first two are about seeing demand, the next three about the stock the band is read against, and the sixth and seventh correct the document. The last three stand apart. The target and band need the first seven, but each stands on its own.

First, the assessment needs to see the demand. The one public count of large load requests I have found came out of discovery in docket 6680-TE-115. The Citizens Utility Board asked Wisconsin Power and Light for every request it had received in twelve months to serve new data center load of 100 megawatts or more. The public answer, PSC REF# 588729, is twelve numbered rows, each reading “Transmission Indicative Study,” with the city and the megawatts blank. The Final Decision, PSC REF# 594678, records twelve requests over 100 megawatts in various locations. The draft concedes the same gap under Figure 1-1, where utilities “indicate that data center developers are considering additional prospective data centers across the state, suggesting additional increases to peak load beyond the forecasts.”

I ask that each utility report, once per assessment cycle, the count and aggregate megawatts of large load interconnection requests by stage: requested, under study, agreement executed, and energized, in ranges wide enough to protect confidentiality, with no names, terms, or locations. PSC 111.11 already asks each utility for the peak demand effect of every capacity purchase, sale, and load interruption, and nothing about load being added. I am asking for the other side of the same table.

Second, the assessment needs a solid number for customer-installed solar and an accounting of where it enters the arithmetic, and it has neither. Chapter 3 counts 464 megawatts of customer-owned generation as nameplate, and Table A-2 counts only the 510 to 560 megawatts registered with MISO. The rest enters the tables only by lowering the metered peak, by an amount the draft does not state.

Why this matters. Customer solar is not in the supply tables, and it is in the demand line only as a lower peak. Four things follow.

1. The reserve margin is computed against that lower peak. If the solar is not producing at the peak hour, the real peak is higher than the one the margin was set against. No one has accredited that solar, and no one is obligated to run it.
2. The demand target reads metered load. As customer solar spreads, metered load grows more slowly than what customers actually consume, so the target reads low.
3. Both effects are small now and grow with every installation.
4. On-site generation at large customers that is never registered as a capacity resource does the same thing, in larger units.

I ask that the assessment report customer-installed solar and other customer-sited generation both as installed and as its estimated contribution at each seasonal peak, and state how it enters the adequacy arithmetic: how much is counted as a behind-the-meter resource, how much is netted from the peak forecast, and how much is neither. Commission staff has collected the number, type, and capacity of non-utility distributed resources from every provider since 2016, so the inventory exists. The ask is to connect it to the tables.

Third, the assessment reports planned additions in nameplate capacity and measures adequacy in accredited capacity. Providers plan more than 7,600 megawatts of new renewable capacity between 2024 and 2030, nearly all solar, which MISO’s indicative results for planning year 2025-2026 accredit at 38 percent of installed capacity in summer and 21 percent in fall. The draft’s 2,250 megawatts of planned storage is even sharper: MISO gives storage three accreditation methods at once, 39, 50, or 62 percent in summer depending on the method, and says its storage modeling will change again. Commission staff’s fifth data request in this docket, PSC REF# 608484, has since asked the five largest providers to refile their supply schedules two ways, under current accreditation throughout and under MISO’s DLOL method from planning year 2028-2029, using MISO’s indicative seasonal values. That is the calculation this request needs. I ask that the final assessment publish both results by season, beside nameplate, state which one its adequacy tables use, and say which accreditation it assumes for storage.

Fourth, the assessment reports imports in megawatts and never in dollars. Wisconsin has imported energy every year since 1990, including 2008, the year of the highest margin in the assessments I have read, so energy imports are a price decision and not a shortage. Capacity is different. A provider short of its requirement buys the difference in MISO’s annual planning resource auction, which sets one price for each season of the coming year. For Wisconsin’s zone the summer price was $666.50 per megawatt-day for 2025-2026 and $424.30 for 2026-2027, and the price averaged across the four seasons was $217 and $126.19. The draft says its adequacy analyses reference the requirement “rather than financial incentives.” I ask that each edition carry one table, for Wisconsin’s providers: capacity bought in the auction by season, with the clearing price and the total paid; energy imported by month, with its cost and the annual total. Both are figures the providers already settle with MISO.

Fifth, the assessment should say how accredited capacity is raised quickly when capacity approaches the band’s floor, and it does not. Its response to the shortfalls it projects for 2028 and 2029 is that the Commission will continue to monitor. The fastest accredited capacity in the document is demand response: 701 megawatts of program capacity in 2025, of which 515 is registered with MISO and credited toward the requirement. The Commission has treated curtailable large load as capacity before. In docket 5-UR-110, PSC REF# 455451, it reopened the Cp-FN interruptible tariff to new customers, citing the need for capacity. The load creating the 2028 step is absent from that discussion. In docket 6630-TE-113 the Commission declined a firm and non-firm designation for very large customers as not shown necessary. I ask the assessment to report demand response in accredited terms by season, state what share of new large load is interruptible or registered with MISO, and assess curtailable large load as the fastest accredited capacity when the band is breached.

Sixth, two tables report different capacity requirements for the same season and year, and the requirement is the floor of the band. Table 1-2 gives the summer 2026 planning reserve margin requirement as 14,821 megawatts and Table 2-3 gives it as 15,157 megawatts, both on a forecast peak of 13,711 megawatts and the same MISO margin of 8.1 percent on an unforced capacity basis. The row labels differ, utilities in one and load serving entities in the other, so there may be a reason, and the assessment does not give one where the tables appear. I ask that it do so. The shortfalls this document projects for 2028 and 2029 rest on the higher figure, and Table 1-2 does not extend past 2026, so a reader cannot check those years against the other basis.

Seventh, the assessment has stopped publishing its own forecasting record. Every prior edition I have read, from April 2009 through November 2024, carried a table of what each earlier assessment had forecast for each planning year, Table 1-3 in the 2024 edition. This draft does not. That table is how a reader judges whether the instrument tracks anything, and an assessment projecting a shortfall in 2028 is asking to be believed about the future. I ask that the table be restored, that prior editions stay available on the Commission’s system, since several are no longer served there, and that the cells be kept to one quantity: the 2024 edition’s two newest columns carry MISO’s requirements where the older columns carry Wisconsin’s forecast margins, and the table does not say so.

Eighth, Chapter 4 counts the lines not yet built and not the capacity of the lines Wisconsin has. Grid enhancing technologies appear once in the draft, as a note on a FERC rulemaking, and reconductoring is not mentioned. ATC’s own current ten year assessment says of dynamic line ratings that “We are investing in research to maximize the use of existing transmission capacity,” and states that ATC is installing its first composite core conductor on a pair of lines in south central Wisconsin. The Energy Systems Integration Group, an industry technical body, reports that reconductoring raises line capacity by 25 to 100 percent and deploys in months against years for a new route. I ask that Chapter 4 report what capacity is available on existing lines through reconductoring and dynamic line rating, and what ATC’s program has produced to date.

Ninth, on what should fill the growth. I would put the goal in five parts.

1. Diversity of source.
2. Clean sources over emitting ones.
3. Both concentrated and distributed, because they do different jobs and the state needs both.
4. Emitting generation only for as long as it is needed.
5. Existing sites and corridors before new ones, including the interconnections at retiring plants and the early site permit work at Kewaunee.

The providers’ carbon goals in Table 1-3 speak to the second and fourth. I ask that the assessment speak to the other three.

Tenth, the draft does not ask the reliability question this load raises. It plans for one kind of loss of load, too little generation. It does not plan for load that removes itself. NERC’s review of the July 10, 2024 event records about 1,500 megawatts of data center load dropping at once on a 230 kV fault, and its 2025 white paper on large loads records a training campus swinging 50 megawatts in under a second. The draft mentions neither ride-through nor load fluctuation, and NERC’s first recommendation is to identify the standards that do not reach these loads. I ask the assessment to characterize forecast large load by type, training against inference, and to report the ride-through and flexibility the serving utilities have specified for it.

My basis is my own research. I read the filings in this docket and in the related rate and tariff proceedings over the past few months, and I follow the Commission’s dockets monthly.

I affirm that these comments are true and correct to the best of my knowledge and belief.

Garold Dalton