Thirteen utilities and infrastructure contractors reported on the same trading day, and one conclusion emerges unambiguously from their results: data-center electricity demand has stopped being a forecast.
And then the market did something that requires explaining: it paid generously for the companies that build the grid, and cut the utilities themselves - on the same day, on exactly the same news.
The Day's Scoreboard
| Company | Role in the chain | Move on the day |
|---|---|---|
| EMCOR | Electrical and mechanical contractor | +19.32% |
| Quanta Services | Transmission grid builder | +17.26% |
| GE Vernova | Power generation equipment | +9.07% |
| Monolithic Power | Power management chips | +5.40% |
| Edison International | Regulated utility | +0.13% |
| Xcel Energy | Regulated utility | -0.66% |
| Ameren | Regulated utility | -1.11% |
| Eversource | Regulated utility | -1.23% |
| AEP | Regulated utility | -1.25% |
| Trane Technologies | Data-center cooling and HVAC | -1.46% |
| Alliant Energy | Regulated utility | -1.54% |
| IDACORP | Regulated utility | -1.73% |
| Southern Company | Regulated utility | -1.78% |
| Exelon | Regulated utility | -3.08% |
The line separating the top of that table from the bottom is not the quality of the results. It is who gets paid now and who pays now.
The Side That Settles the Argument: How Much Power Is Actually Needed
For two years, AI electricity demand has been mostly a story. Today's reports turned it into signed numbers.
AEP signed six gigawatts of load agreements in a single quarter, primarily in Texas, bringing total contracted load through 2030 to 69 gigawatts. The company notes the customers include hyperscalers, data centers and industrials.
Alliant Energy forecasts roughly 60% load growth by 2031, with three data center developments underway, and states that "expected demand from large customers remains on track to materialize this year."
And Xcel Energy signed an agreement to power a new Google data center in Minnesota - on a term that deserves attention: Google will pay all costs for the new service for the duration of the contract.
Why that Xcel term matters more than the deal itself
The big problem utilities face with data centers is not demand - it is who pays for the infrastructure built to serve it.
When a utility builds a transmission line and a power plant for one enormous customer, the risk is that the customer leaves in a decade and the bill rolls onto everyone else - and that is exactly where regulators and politicians stop projects.
The Xcel agreement is structured so that Google bears the cost in full, in accordance with Minnesota's regulatory and legislative requirements for large loads. That turns the deal from a rate risk into an opportunity - and it is the model that will determine whether the data-center wave gets built on schedule or stalls in regulation.
And still: the agreement is subject to approval. Until it is approved, this is an intention, not a cash flow.
And the Side Collecting the Payment: Records Everywhere
While the utilities describe the demand, the contractors are already billing for it.
| Metric | Quanta | EMCOR | Trane |
|---|---|---|---|
| Quarterly revenue | $9.56 billion | $5.25 billion | $6 billion |
| Earnings per share | $4.24 (adjusted) | $9.06 (record, +34.8%) | $4.31 (adjusted, +11%) |
| Backlog / remaining obligations | $53.4 billion (record) | $17.14 billion (record, +43.9%) | $12.1 billion (record) |
| Full-year guidance | Raised across all metrics | Raised to $32.00-33.25 | Raised (revenue and EPS) |
Two figures deserve a pause:
The first - EMCOR's guidance raise. The company lifted its 2026 EPS range from $28.25-29.75 to $32.00-33.25. That is roughly a 13% raise mid-year - not a tweak, but an admission that the previous forecast did not correctly describe the pace of work.
And the second - Trane's bookings. Organic bookings rose 37%, and Americas Commercial HVAC rose 50%. Book-to-bill stood at 123%, meaning the company entered the next quarter with more work than it finished.
Why backlog is the important number in this sector
In infrastructure, a quarter's revenue describes work signed a year or two ago. The backlog describes what has not been executed yet.
Which is why a record backlog is the closest thing to direct evidence of future demand - it is not management's forecast and not an analyst's estimate, but signed contracts not yet recognized as revenue.
Quanta's backlog stands at $53.4 billion - more than five times its quarterly revenue. And EMCOR's remaining performance obligations grew 43.9% - faster than its own revenue.
When backlog grows faster than revenue, it means the company is signing work faster than it can execute it. As we wrote this morning about Quanta, that is precisely the condition in which the bottleneck moves from demand to the capacity to deliver.
So Why Did the Utilities Fall?
And here one has to resist an explanation that is too easy. The utilities did not report badly.
Southern Company grew EPS from $0.80 to $1.03 - a 29% increase. AEP raised its full-year operating earnings guidance to $6.25-6.55 from $6.15-6.45. IDACORP raised the bottom of its range. Alliant reports it is tracking in the upper half of its range. And Exelon and Edison affirmed their guidance and their multi-year growth rates.
Almost all of them reported well. And almost all of them fell.
Three reasons, and only one of them relates to the results
The first - utilities trade as bond substitutes. Their return is relatively fixed and predictable, so their price moves inversely to interest rates. In a week when the Federal Reserve held rates in a hawkish split vote, the whole group gets pressured - regardless of any single company's performance.
The second - and here lies the substantive difference: who bears the investment. A utility committing to serve 69 gigawatts must first spend enormous capital on generation, storage and grid, and earn a regulated return on it over decades. The contractor, by contrast, gets paid for that same investment within quarters.
The same demand wave is an expense at one and revenue at the other.
And the third - regulatory risk. The investment requires regulatory approval to enter the rate base, and the return on it is set in a commission rather than in the market. That is a system that protects against loss and caps the upside at the same time - which is why it does not price a demand wave the way a growth stock prices one.
And the Anomaly That Needs Its Own Explanation: Trane
Trane reported organic bookings up 37%, Americas Commercial HVAC up 50%, a record $12.1 billion backlog, and raised both revenue and EPS guidance for the full year.
The stock opened higher - and closed down 1.46%.
It gave up the entire gap it opened with over the course of the day.
And why that is interesting: Trane is not a utility. It sells cooling - and cooling is a genuine bottleneck in data centers, because AI servers generate heat that ordinary systems were never designed to remove. If there is one company the data-center thesis should lift, it is this one.
The most plausible explanation is expectations: after two years in which the stock was priced on that thesis, +37% bookings was what the market had already assumed. Adjusted earnings rose 11% - respectable growth, but not on the order of the bookings. The gap between the pace of bookings and the pace of profit is what got sold today.
The Bull Thesis
Whoever reads the day positively will see the end of the argument about demand: 69 gigawatts contracted at AEP, roughly 60% load growth at Alliant, and record backlogs at all three major contractors. This is no longer a story - these are contracts.
Beyond that: EMCOR's 13% mid-year guidance raise is the kind that tends to repeat when the original forecast missed a trend, rather than a one-off event. And anyone buying utilities on the dip is getting a higher dividend yield on a growing asset base.
The Bear Thesis
Whoever reads it negatively will point first to the gap between bookings and profit. At Trane, bookings rose 37% and adjusted earnings 11%. Backlog is a promise, not cash flow - and between the two stand labor, equipment and costs that can erode the margin.
Second, the utilities are entering a heavy capital investment cycle, funded with debt and equity issuance. In a high-rate environment that raises the cost of financing and dilutes shareholders before the return starts flowing.
And third, regulatory risk is not theoretical. Xcel's agreement with Google is still subject to approval, and any structure that passes cost to consumers will face public scrutiny.
The debate in one line
The bulls see demand that has moved definitively from forecast to signed contracts - 69 gigawatts at AEP, a $53.4 billion backlog at Quanta and a 13% guidance raise at EMCOR. The bears see bookings growing far faster than profit, an investment cycle diluting utility shareholders, and regulation that has not yet approved some of the agreements. Both sides are reading the same reports.
My Angle
A personal opinion of Ilan Abramov - not advice, not a recommendation
What catches me about this day is that the market made a correct distinction, and not necessarily for the correct reason.
The distinction between who pays for the demand wave and who bills for it is real and substantive. A utility signing up for 69 gigawatts is first and foremost committing to enormous capital expenditure that comes back to it in rates over decades. A contractor signing the same project gets paid within quarters. That is a genuine difference in cash-flow profile, and it justifies different multiples.
But today's move does not look to me like a distinction - it looks like a rotation. Companies that raised guidance fell alongside companies that merely affirmed it, in a narrow band of one to three percent. When an entire group moves together with no differentiation between a better result and a worse one, that is usually interest-rate pricing rather than business pricing.
And the figure I had marked for myself to check going into this day was the backlog, not the earnings. Over recent quarters I had seen backlog at the contractors growing faster than revenue, and that is exactly the pattern that sharpened today - EMCOR's remaining performance obligations grew 43.9%, faster than its own revenue. When that happens, the key question stops being "is there demand" and starts being "is there anyone to execute it."
And the point of caution I hold: bookings up 37% against adjusted earnings up 11% at Trane. A gap like that can come from mix, from timing, or from investing ahead in capacity - and each of those means something entirely different. I do not know which one is right, and that is exactly what makes it a figure worth tracking rather than drawing a conclusion from now.
And what I will watch next quarter: whether the record backlogs start translating into margin. If the contractors keep accumulating work without the operating margin expanding, then the bottleneck has moved from demand to execution capacity - and that changes who profits from this wave.
Summary
Thirteen utilities and infrastructure contractors reported on the same day, and demand ceased to be a question. AEP signed another six gigawatts in a single quarter, reaching 69 gigawatts of contracted demand through 2030; Alliant forecasts roughly 60% load growth by 2031; and Xcel signed an agreement for a Google data center in which the customer bears the cost in full.
And at the contractors - records: Quanta with $9.56 billion of revenue and a record $53.4 billion backlog; EMCOR with record EPS of $9.06 and a full-year guidance raise of roughly 13%; and Trane with organic bookings up 37% and a record $12.1 billion backlog.
And the market split the sector in two: EMCOR jumped 19.32% and Quanta 17.26%, while every regulated utility fell - including the ones that raised guidance.
The distinction the market drew is between who pays for the demand wave and who bills for it. The question left open is whether those backlogs translate into margin, or only into work.
Sources: the official second-quarter 2026 results releases of Quanta Services, EMCOR Group, Trane Technologies, Xcel Energy, American Electric Power, Exelon, Southern Company, IDACORP, Edison International and Monolithic Power Systems, as filed with the U.S. Securities and Exchange Commission on July 30, 2026, including revenue, earnings per share, backlogs and remaining performance obligations, load agreements and full-year guidance; Alliant Energy figures from its July 30, 2026 results release as reported in the financial press. Stock moves are closing changes for the July 30, 2026 trading session. Data accurate as of the time of writing. The charts are shown in real time via TradingView. Nothing herein constitutes a forecast, recommendation or advice - see the full disclaimer at the bottom of the page.