Everus Construction Group

IThe business

Everus does the electrical and mechanical work inside large buildings, and builds and repairs power lines for utilities. It wires data centers, fire-sprinkles hotels, pipes hospitals and strings overhead lines across the American West. It is headquartered in Bismarck, North Dakota, and works through 16 operating companies, most under local names, with about 9,400 employees at the end of 2025 and about 4,000 customers on more than 44,000 projects that year.³¹¹ About 85% of its employees are represented by labor unions.³

Its history is short as a public company. The business started in 1997 inside what became MDU Resources, grew through more than 25 acquisitions, and was spun off to MDU’s shareholders on 31 October 2024.³ MDU’s shareholders received 100% of the shares, and the first close in regular trading, on 1 November 2024, was $52.49.¹²¹ So there are two audited years as a stand-alone company, 2024 and 2025, plus nine months of 2026. Anything earlier is a figure from MDU’s filings or from the information statement written for the spin-off. I label those MDU-era throughout. They describe the same business, but costs were allocated by a parent, interest was on loans from MDU, and the scope changed at times, so I do not treat them as exactly comparable.⁴¹³

The money works like this. A customer, often a general contractor, awards a job by competitive bid or by negotiation. Everus does the work with its own crews and subcontractors, bills as it goes, and books revenue as costs are incurred. In 2025, 52% of contract revenue came from fixed-price jobs, down from 59%, 43% from cost-reimbursable jobs, up from 34%, and 5% from unit-price jobs.³ On a fixed-price job, a cost overrun is Everus’s to eat. On a cost-reimbursable job it is paid for, but there is less to gain. Backlog is the signed work not yet done, and it is the main number to watch.

Two segments, one engine

Electrical and Mechanical, or E&M, is the engine. It was 77% of contract revenue in 2025 and 82% of revenue in the third quarter of 2026 C (calculated).³² Transmission and Distribution, or T&D, is the older and steadier utility business, with a better margin and slow growth.

Where the growth came from

Almost all of it is one customer type. Revenue from commercial customers, a group that holds data centers, hotels and ordinary commercial buildings, was $2.07B in 2025, against $1.20B in both 2023 and 2024. That is 55% of the company’s revenue, and it supplied 98% of the $897M rise in revenue in 2025 C (calculated).³ The company does not report data centers separately. In the third quarter, E&M revenue rose $368M, or 48%, which the release puts down mainly to data center and high-tech work, offset in part by renewables.² On the call, as written in a third-party transcript that I have not checked against audio, the CEO said data center work is “the largest part of our backlog” and that its customers “tend to be large hyperscalers.”⁵ So I know the direction and not the size. Anything I say about the exact data-center share is a guess, and I mark it so.

Other end markets did not grow. Institutional work, such as schools and hospitals, fell from $365M to $353M in 2025. Industrial work was flat at about $320M, after $473M in 2023. Utility revenue was $750M, flat on 2024.³ The company has also said that its “revenue mix has changed significantly in the last couple of years” and that a loss of data center business “could have a material adverse effect.”³¹¹ I take both statements at face value.

Backlog

Backlog is the other half of the picture. It was $3.23B at the end of 2025, $3.68B in March, $4.55B in June and $5.48B in September 2026. That is 86% above a year earlier, and 1.5 times the whole of 2025 revenue C (calculated).²¹⁵¹⁶³ About $742M of it came with the two acquisitions of 2026, so organic backlog is about $4.74B, up about 61% on a year earlier and 47% since December C (calculated).² The CFO told analysts that backlog burns at about 80% over twelve months, and that Epsilon’s modular work converts more slowly.⁵ At 80%, about $4.4B is due to turn into revenue by the third quarter of 2027, before any new work that is won and done inside the year.

Two cautions. First, backlog is not a contract to be paid. The company defines it to include letters of intent and work orders, and says contracts can be delayed, cut or cancelled.³ Second, the figures to 2022 are MDU’s and the later ones Everus’s, and I can’t be sure the two count alike. The jump to $2.13B in 2022 is in MDU’s own 10-K, so it is not an effect of the spin-off.¹⁹

What changed in 2026

Three things happened at once. The business grew faster: 25%, 34% and 40% in the three quarters, 31% organically in the nine months.¹⁶¹⁵² It bought two companies: SE&M Constructors of North Carolina, founded in 1923 and strong in pharmaceutical and industrial mechanical work, for $158M of base price in April, and Epsilon Industries, a maker of prefabricated mechanical and electrical modules, for $295M in cash on 1 September.²⁰²¹²² And the margin stepped up: gross margin was 14.2% in nine months against 12.3% a year earlier.² Net cash spent on acquisitions in the nine months was $380.5M, net of cash acquired; I can’t reconcile it exactly to the two headline prices until the third-quarter 10-Q is filed.²

The guide leaves a fourth quarter that is no better than the third. EBITDA of $127M to $137M compares with $140M in Q3, which included a $7.6M one-time gain from settling old billings on a past job, worth about 55 basis points of margin.²⁵ That is still 50% to 62% above the fourth quarter of 2025, so profit is not shrinking against last year C (calculated). But at the midpoint it stops rising from one quarter to the next, as it did last year, when Q4 EBITDA of $84.8M followed $89.0M in Q3 C (calculated).²³² Management said Q4 would continue a “similar margin trend.” Asked whether the third-quarter margin without the gain, about 9.6%, was the new base for 2027, the CFO said “I wouldn’t call execution benefits normalized margin.”⁵ Companies guide cautiously, and this one raised its forecast at each of the last three reports. Still, the gap between a record quarter and a flat forecast is where the debate sits.

That is the fact that the share price hit on 4 November. The stock closed at $112.73 on 3 November, up 9.5% on the day, before the release after the close. It closed at $79.71 the next day, down 29.3%.¹ I could not find a filing, a company statement or a dated analyst note that explains a fall of that size, and I do not know the cause. The only filing since the release is the 8-K itself, and the third-quarter 10-Q is not yet filed. A third-party write-up of the call put the drop down to investors looking past the beat to the quality of some margin gains and to acquisition integration; that is the writer’s reading, not the company’s, and it gave a midday figure of 25%, not the close. Analysts on the call asked about the one-time gain, about the margin and about talk of data-center pauses or moratoriums.⁵ The company said no customer has changed its behavior.⁵ That is what I can document. The rest is my reading, and it is in the valuation section.

IIThe moat

The honest starting point is that contracting is a hard business to protect. Everus wins most work by bidding, the customer can pick someone else next time, and the margin on each job is small: gross margin was 12.1% in 2025.³ A contractor has no patent and no network. What it has is closer to a reputation, in a business where reputation decides who is invited to bid.

Four things add up to an advantage that I would call real and local, not wide.

• Skilled labor. The binding limit on this industry is electricians, linemen and pipefitters. Everus has union relationships in many states and is at “record employment,” in the CEO’s words, with a workforce that the CFO says it has quadrupled over 10 to 12 years.⁵ A rival with a bigger order book than crew can’t take the job.
• Safety and bonding. Public and many private jobs need a surety bond, and customers screen on injury rates. Everus’s 2025 recordable incident rate was 1.06, and a share of its executives’ pay depends on it.³¹⁰ A weak safety record gets a contractor dropped from the bid list.
• Local brands, national customers. Sixteen operating companies keep local ties, and together they can follow a large owner from state to state. The 10-K says that ability to support national customers in most of its regions is part of why it has grown.³
• Prefabrication. Building pieces in a controlled shop uses fewer field hours. The company is spending on it, and Epsilon, which makes modules, is the largest step so far.²¹⁵

Evidence it is holding

The margin has held up better than a pure price-taker’s would. EBITDA margin stayed in a band of 7.2% to 8.5% from 2021 to 2025, through a year when revenue was flat, and gross margin did not drop below 10.2%.⁴³ Return on capital is high because the customers fund much of the work: contract liabilities, which are bills sent ahead of the work, were $522M at the end of September, against receivables of $1.07B.² On my arithmetic, operating profit after tax was about 29% of average capital in 2025 C (calculated).³ That is a business that doesn’t need much money to grow, as long as customers keep paying on time. Receivables took about 71 days of sales in the third quarter C (calculated).²

And in 2026 the margin rose. Gross margin was 14.2% in nine months against 12.3% a year earlier, with the company citing “increased workloads, solid project execution, project timing and a dispute resolution.”² If the cause is that crews are scarce and Everus is among the few with capacity, that would be pricing power, and the cost-reimbursable share rising from 34% to 43% would fit that story as well.³

Evidence it is narrowing

The first doubt is about who the customers are. One customer was 16.5% of 2025 revenue, about $618M, and the top 10 were about 43% against 33% in 2024. A single E&M customer was 21.1% of that segment, and another was 10.4%.³ A buyer that large sets the terms. The filing doesn’t name it, and I won’t guess.

The second is the one that matters most to the margin. A customer that is a hyperscaler has a team whose job is to buy construction well, and a handful of contractors who can do the work at scale. If crews become less scarce, price power flows the other way. The third is that the best rivals are far bigger, which I turn to next. The fourth is the one I can measure: what the company calls “changes in estimates” on work already under way added a net 6.2% of revenue in the second quarter of 2026, against 5.3% a year earlier and 3.0% for all of 2025.¹¹³ I come back to it under Financials.

So the protection, in plain terms, is that Everus is among the few firms with the crews, the safety record and the credit to do very large electrical jobs, and that is a durable position while demand outruns supply. It is not a position that survives a pause in demand with its margins intact.

IIIThe competitive advantage

The company lists its rivals as Comfort Systems, EMCOR, IES, MasTec, MYR, Primoris, Quanta and Sterling, plus private firms such as M.C. Dean and Rosendin.³ Three of them report this quarter’s numbers in time for a side-by-side. They are all bigger, and they are all growing quickly, which is the point: the whole group is riding the same wave.

Where Everus wins

On price and balance sheet. It has little debt: net leverage was 0.7 times EBITDA, against a stated target range of 1.5 to 2 times, so there is room to buy more.²⁵ And the price is lower. At $79.71 the equity is worth 8.3 times four quarters of its Q3 operating profit, against 12.8 for EMCOR, 15.3 for Comfort Systems and 21.3 for Quanta C (calculated).²⁹ If the margin and growth were equal in quality, the gap would be a gift. I don’t think they are equal, as the next part says. Its backlog grew 86% in a year, from a business whose quarterly revenue is under an eighth of Quanta’s.²⁸

Where it loses

On growth and on margin. Comfort Systems grew 54% in the quarter and Quanta 57%, against 40% for Everus, although Quanta’s includes acquisitions.⁶⁸² Comfort Systems made 16.6% in the quarter, almost double Everus’s 8.8%.⁶ Its mix includes more service and maintenance, which carries better margins, and it has been at this for longer. Everus’s T&D business makes 15% EBITDA, but it is 18% of revenue and its backlog is flat. On scale, too: Quanta’s quarterly revenue is nearly nine times Everus’s, and EMCOR’s four times.⁸⁷ Large owners that are building across the country give more work to the contractors with more crews and more credit.

And on history. Comfort Systems’ operating margin went from 6.7% in 2020 to 14.4% in 2025 C (calculated).²⁴ Everus’s step is two quarters old: EBITDA margin was 8.6% in the first quarter, 10.4% in the second and 10.1% in the third. The market has paid for Comfort Systems’ margin, and as of 4 November it pays much less for Everus’s.

The demand underneath

The case for all of these companies starts with the same fact: the biggest buyers of computers are spending as never before. In the first nine months of 2026, Alphabet spent $132.0B on property and equipment against $63.6B a year earlier, Amazon $175.7B against $92.3B, and Meta $83.1B against $48.3B. Together that is $390.8B against $204.2B, up 91% C (calculated). Microsoft spent $39.4B in the quarter to September against $19.4B.^25262728

That is the demand. The question for Everus is how long it lasts, and who is paid for it. Purchases of this size are decisions by a handful of boards. Their size is why the order book has grown, and it is also why a change of mind by two or three of them would show up in Everus’s backlog within months. The electricity has to come from somewhere too, which is the T&D business’s hope and also the tail risk the company itself mentions: projects “require continuous, adequate power sources, transmission capacity and permitting.”³ In the call, two analysts asked about election outcomes that could bring pauses or moratoriums. Management said it had seen no change of behavior.⁵

IVManagement

The team is the one that ran the business inside MDU. Jeff Thiede has been chief executive of the construction group since 2013. The operating chief has about 25 years in the MDU and Everus companies. The chief financial officer is new to the business, from August 2024, and so is the legal chief, who came from MDU’s own legal department.¹⁰ Dale Rosenthal, a former chief financial officer of Clark Construction Group, chairs a board of eight in which all but Thiede are independent. Betty Wynn on the board was chief financial officer of MYR Group, a rival.¹⁰

At the May meeting every director was elected. The weakest vote was for Edward Ryan, with 35.6M shares for and 4.6M against, or about 89% in favor C (calculated). The pay vote passed with about 96% in favor C (calculated).²⁹

How they are paid, and what that pays them to do

Thiede’s 2025 pay was $4.86M: $850,000 of salary, $1.87M of cash bonus, $2.01M of stock at grant value and $129,000 of other items. That is 52 times the median employee’s $93,574.¹⁰ Over 80% of his target pay is at risk.

The bonus is the part to read. It rests 80% on adjusted EBITDA and 20% on safety. For 2025 the plan EBITDA was $224.1M, set in February 2025. Reported EBITDA for 2024 had been $232.2M, so the plan sat below what the company had just earned C (calculated). The result was $319.8M, or 142.7% of plan, and the bonus paid the maximum, 200% of target.¹⁰³ Pay goes up if EBITDA goes up, and the definition lets the committee exclude costs of acquisitions and the effect of acquired businesses.¹⁰ So the plan rewards growth, including growth that is bought. There is no measure of return on capital and none of cash. The long-term awards are 60% performance shares, half on adjusted EBITDA over three years and half on total return against a peer group, and 40% time-based shares.¹⁰ The TSR half is a real test, and after the 4 November fall it will bite. The EBITDA half has the same flaw as the bonus.

I do not say this leads to poor decisions. A contractor with a 16.5% customer and fixed-price jobs has plenty of reasons to be careful about risk. I say that the plan wouldn’t flag a business where profit grows while cash lags, which is the pattern in the five-year numbers below.

How much of their own money is in

Directors and officers as a group, 15 people, owned 167,722 shares on 20 March 2026, or 0.33% of the company, worth about $13.4M at the close C (calculated).¹⁰ Thiede’s 66,426 shares are worth about $5.3M C (calculated), against an ownership rule of five times salary, or $4.25M.¹⁰ That meets the rule, but the stake is small next to a company worth $4.1B, and it is not what a founder holds. The largest holders are BlackRock with 11.9% and Vanguard with 11.0%. ¹⁰

What they did with capital over a full cycle

There is no full cycle to judge. Two years as a public company, both in a boom, is a short record. What the numbers show is a business that, until 2026, spent little: net capital spending of about $35M to $57M a year, no deals, no dividend, and debt paid down from $300M to $285M.³ In 2026 that changed. In nine months it spent $380.5M on deals, its debt rose from $285M to $571.5M, and net leverage went from 0.4 times to 0.7 times EBITDA.² The target range is 1.5 to 2 times, so management intends to use the balance sheet more.⁵

One fact stands out. Management paid about 9 to 12 times EBITDA for Epsilon in September, when the shares traded near $130. At $79.71, Everus’s enterprise value is 9.0 times the midpoint of its 2026 EBITDA guide C (calculated). The deal looks better than the stock’s fall implies, or the stock looks cheaper than the deal, depending on which you trust. I can’t judge Epsilon from outside; it was bought a month ago and its first month of purchase accounting was “lumpy,” in the CFO’s word.⁵

VFinancials

Ten years of this business show steady growth until 2024, a flat year, then a jump. The first five years come from MDU’s filings, the next three from the spin-off’s information statement, and the last two from Everus’s own accounts. I mark each. They are the same business, not the same accounting.

Revenue has compounded at 15% a year since 2016 and earnings at 22% C (calculated). Earnings margin rose from 3.2% to 5.4%. But the growth was not smooth. Revenue was flat from 2023 to 2024, at $2.85B, and the backlog went down in 2023. Growth of 31.5% in 2025 and 34% in the first nine months of 2026 is far above anything in the first decade, and 2026 is guided to about $5.05B.³²

What an owner could really take out

Reported earnings and cash are different things for a contractor that is growing. Over the five years from 2021 to 2025, net income was $0.72B. Free cash flow, the company’s own measure, was $0.40B, or 56% of net income C (calculated).⁴³ Take off the stock pay of $6.3M in 2025 and the figure is lower still.³ The gap is mostly money tied up as the company grows: in 2025 receivables and contract assets rose $269M, offset by $186M of higher payables and customer billings, a net drag of about $83M on cash from operations, plus $57M of capital spending net of equipment sales C (calculated).³

In 2025 the conversion was 50%: $100M of free cash flow on $202M of net income. In the nine months of 2026 it is 94%, $221M on $234M. Management said “our 2026 free cash flow had some timing benefits,” and that operating cash rose on “favorable working capital timing.”⁵ I don’t treat that as the new normal. My model takes off 9% of added revenue for working capital, which is what 2025 cost C (calculated), and capital spending net of depreciation, so the owner cash in my base case for 2027 is $235M against about $350M of net income I expect for that year E (estimate). A company growing 20% a year cannot turn all of its profit into cash. If growth slows, more does.

Quality of earnings

Two points need care. The first is the one I raised in the section on the business’s protection. Everus revises its estimate of the cost of each job as work proceeds, and when a job comes in under budget, the gain is booked at once for the part already done. The company reports the net effect on revenue. It was 2.9% of revenue in 2024 and 3.0% in 2025, or about $112M in 2025, against operating income of $265M. In the first half of 2026 it was 5.1%, and in the second quarter 6.2%, or about $76M against operating income of $112M C (calculated).³¹¹

The company breaks that net figure into good and bad. In the second quarter, favorable changes were 7.5% of revenue and unfavorable ones 1.3%. It names the favorable causes as labor efficiency, lower material costs, risk reduced, and change orders. The unfavorable causes were labor inefficiency and delays and overruns.¹¹ The filing doesn’t give the profit effect. In my reading, because the costs are already spent, most of a revenue catch-up reaches profit, so the share of operating income is probably large E (estimate). Three observations. This is how contractors normally book, and it is not a sign of manipulation. It does mean part of the 2026 margin step is gain on work won in earlier years at earlier prices. And it is where bad news would land: on a fixed-price job that has gone wrong, the same arithmetic runs backward.

The second point is the one-time gain. In Q3 a settlement of past billings added $7.6M to gross profit, or about 55 basis points of margin.²⁵ Q3 also carried higher incentive pay of $13.4M and $6.7M of acquisition costs, which partly offset it.² The gain is small next to the changes in estimates. I raise it because analysts asked about it on the call.⁵

The balance sheet

The balance sheet is sound for a contractor, with a larger load than a year ago. At 30 September, Everus had $244.8M of unrestricted cash against $571.5M of gross debt, which is net debt of $327M, or 0.7 times trailing EBITDA. It had $247.8M available on its revolver.² The covenants allow net leverage of up to 3.0 times and require interest cover of at least 3.0 times; interest expense was $15.4M in nine months, against operating income of $312M C (calculated).¹¹² Goodwill and acquired intangibles are now $580M, or 21% of assets, up from $143M of goodwill in December, so tangible equity is about $284M C (calculated).²

Smaller items. Operating leases add $137M of liabilities that I do not count as debt, because the lease cost is already in operating income.² The company contributes to multiemployer union pension plans and says it can’t estimate any withdrawal liability.³ It needs surety bonds for a material share of its work, which depend on its balance sheet and its record.³ Not in the numbers above: a contingent payment to the sellers of SE&M, a fair-value liability that rose $1.9M in Q3.²

The shares have gone from $52.49 on the first day of regular trading to a record close of $169.16 on 6 May 2026, and back to $79.71. That is a rise of 222% and a fall of 53% C (calculated).¹ The low close in that time was $33.60 on 4 April 2025. I read the 2026 level of $118 to $169 as the market paying for the margin step. On 4 November much of that was reversed in a day.

VIValuation

First, how this goes badly wrong

Before the numbers, I asked what would most likely hurt a holder of this stock, and worked back. Three things could, and none needs a shock.

• The big buyers pause. Three companies alone spent $391B on property and equipment in nine months, almost twice the year before, and some share of it reaches Everus C (calculated).²⁵²⁶²⁷ They can stop without notice, and the company has told us a loss of data-center work “could have a material adverse effect.”³ A pause of two or three quarters would show first in new orders, then in backlog, then in revenue a year later. Everus has taken on labor and bought two companies for it. The one customer at 16.5% of revenue is part of the same risk.
• The margin was borrowed from the cycle. Gross margin stayed between 10% and 12% for five years and went to 14.2% in 2026. If part of the step is scarce crews, good weather, favorable re-estimates and a one-time gain, it can go back. A fixed-price job bid at the top of the market and finished in a slack one would carry it back faster, and 52% of the work is fixed-price.³²
• Good cash is spent badly. The balance sheet is clean, the plan pays on EBITDA, and the company has just paid about 9 to 12 times EBITDA for Epsilon.²¹¹⁰ Management’s stated range for leverage is 1.5 to 2 times, which on trailing EBITDA of $442M leaves room for $340M to $560M more net debt, and more as EBITDA grows C (calculated).⁵²³ If the next purchases come at the next peak, returns suffer.

The first two are one question seen twice: how long does the build-out last, and how much of the margin is a price for scarcity? They set my bear case. The third I keep outside the model, because I can’t size it. Capital spent on deals is neither in my cash flows nor in my balance sheet.

What the price implies

At $79.71 the market values Everus at $4.08B on 51.2M diluted shares: 51.1M outstanding on 30 September plus 0.1M for options and share units C (calculated).²¹ Add net debt of $327M for an enterprise value of $4.41B. That is 9.0 times the midpoint of 2026 EBITDA guidance, and 12.9 times the $6.17 I expect for 2026 earnings per share E (estimate).² On 3 November, before the release was priced in, it closed at $112.73, which was 18.3 times the same figure C (calculated).

What does the price pay for? Hold the margin on my base path and the price needs revenue to grow 11.0% a year for ten years, to $14.4B in 2036 C (calculated). Or take my base growth, and the price works only if the margin never falls below 8.7% for ten years, against a base that fades to 7.2% C (calculated). So the price is a bet that this margin is the new normal. It doesn’t need a bubble. It does ask for a business that is never again as hard to run as it was in 2022.

What I can’t judge

Two inputs decide this valuation, and I can’t know either. One is how much of the business is data centers and how long that spending runs. The company gives no share, only that the commercial group is 55% of revenue and that data centers are “the largest part” of backlog.³⁵ I’d guess a third to a half of revenue; it is a guess E (estimate). The other is the normal margin. Before 2026 EBITDA margin had never been above 8.5% in the years I can see, and 2026 is guided at 9.7%. There is no earlier public year in which to see how it behaves when work slows. So here is a range and not one number. Read my $64 with the grid next to it.

The discount rate

I use 11.0%. The ten-year Treasury yielded 4.90% on 4 November, and Damodaran’s implied equity risk premium was 4.41% on 1 November.³⁵³⁶ My judgement for a beta is 1.3, for a cyclical small company with one large customer, which gives 4.90% plus 4.41% times 1.3, or 10.6%. I round up to 11.0% for size and concentration E (estimate). A beta from the data is not much help: against the S&P 500 on 25 months of returns it comes out at 2.6, with a monthly swing of 18% in the share price, and I don’t trust a number built on a spin-off year, a crash and a spike C (calculated).¹³⁷ It is above the 10.0% on my Netflix page and 10.5% on Broadcom, and below the 12.0% on Vertiv, whose orders face the same data-center cycle at a much higher price.

Three cases

The model is simple on purpose. I set revenue growth and an operating margin for each year from 2027 to 2036, measured before the amortization of acquired intangibles, which is a non-cash charge. I tax at 26%, the nine-month rate. Then I take off capital spending of 1.7% of revenue less depreciation of 0.8%, working capital at 9% of added revenue, which is what 2025 cost, and stock pay of 0.2% of revenue. What is left is owner cash. I start from the middle of the 2026 guide, $5.05B and an 8.95% margin, and I do not include any purchases beyond the two already made. Net debt is the 30 September figure, with no credit for fourth-quarter cash, which may reverse some of the year’s timing gains.

Near-year numbers: my 2027 revenue of $6.06B is 20% above 2026, and about 4 points of it is the full year of the two deals E (estimate). I did not find a published estimate for 2027 I could verify, and the company says it will give 2027 guidance in February.⁵ Where I differ from the optimists is after 2028.

Where would $7.7B of revenue in 2030 come from? Start with 2027, on one condition: that backlog is still near $5.5B at the end of 2026, which means about $1.4B of new orders in the fourth quarter to replace the revenue worked off. At 80%, $4.4B of that turns into revenue in 2027. My base 2027 revenue of $6.1B then needs about $1.7B of work that is booked and done inside the year. In 2026 the equivalent figure is about $2.5B: the $5.05B guide less the $2.59B of backlog due within twelve months at the start of the year, though that includes revenue from the two acquisitions C (calculated).⁵³² So if backlog holds, 2027 asks for less new work than 2026 delivered. The doubt starts at 2029. By 2030 the base needs revenue to be 1.5 times 2026 four years on, which requires the backlog to keep refilling after the current wave of orders is spent. I can’t see that far, and this is where my base is a judgement and not a count.

The price sits 24% above my base, and that is the plain result. Why might the market be right? Because it sees what I see, plus things I can’t. The margin has risen faster than any five years of history suggest; the CFO would not call the third-quarter margin normal; a customer with 16.5% of revenue can leave. On the other side, the stock has fallen 29% in a day on a beat and a raise, and at 12.9 times my 2026 earnings it already prices a lot of fading. If my base is too low, the likeliest reason is that crews stay scarce and the margin holds near 9%, which is the bull path. The $99 bull is a fair picture of that world.

Other uses of the same dollar

The ten-year Treasury pays 4.90%. Against that, my base here returns 9.6% a year and my bear 5.9% E (estimate).³⁵ The closest comparison on this site is Vertiv, which sells the power and cooling for the same data centers. There my base returns 3.8% a year against 12.0% required, at $232.50. GE Vernova returns 5.9% against 10.0%. So on my numbers Everus is the one of the three closest to its own hurdle. That isn’t a reason to prefer it: Vertiv and GE Vernova make equipment that is scarce, and Everus supplies labor. The pages use their own rates and methods, so I treat gaps of a point or so as noise. Everus’s gap to its hurdle is 1.4 points, against 8.2 for Vertiv and 4.1 for GE Vernova.

What past lows looked like

The stock has fallen more than 30% from a high twice: 56% from the $76.76 close of 24 January 2025 to the April low, and 53% so far from the May record.¹ At the April 2025 low of $33.60 the shares were 12.0 times 2024 earnings of $2.81 C (calculated). Today’s $79.71 is 12.9 times what I expect for 2026 E (estimate), so the multiple is similar and the earnings behind it are much higher. That is not a forecast. It shows that this stock was priced at about 12 times trailing earnings within the last two years.

The margin of safety

• Price$79.71Close, Q4 2026 [1]
• Base value$64My estimate E (estimate)
• Margin of safety below$36set from my bear case C (calculated)

I set the margin of safety from the bear case: at $36, my $27 bear value is three quarters of the price paid, and the price is 44% under my base value. Why the bear? Because the danger here is the cycle ending, and a discount to base wouldn’t protect against that. At $36 my bear returns 9.2% and my base 16.0% a year E (estimate). The shares last closed below $36 on 21 April 2025, a year in which Everus earned $202M against about $316M I expect for 2026.¹³ At $79.71 the price is 121% above that level, so there is no margin of safety on my numbers, and there is less of a gap to the base than there was a day ago.

Two different questions

How good is the business, and how good is the price? They are separate questions. As a business Everus is good on most tests I can apply: very fast growth, a return on capital near 29%, low debt, customers who fund the work, and a record backlog. Its flaws are the ones it admits: a thin gross margin, a customer that is 16.5% of revenue, and only two public years. As for the price, it was poor at $112.73, when it was 76% above my base and returned 7.8% a year on it. At $79.71 the gap is smaller and still on the wrong side. So this looks like a good business at a price a little above my base, on the day the price fell 29%. That is a different thing from a good business at a bad price. It is also not a good business at a good price, and my range says why: the answer depends on how long a cycle lasts, which I cannot know.

VIIThe case for the price

What most people miss

Three facts cut in Everus’s favor and get less attention than the backlog. The mix of contracts moved toward safer work: fixed-price jobs were 52% of 2025 contract revenue, down from 59%, while cost-reimbursable work rose from 34% to 43%.³ Second, customers fund a lot of the business. Contract liabilities, which are bills sent ahead of the work, rose from $305M in December to $522M in September.² Third, the backlog is mostly organic: excluding the $742M that came with the two deals, it is about $4.74B, up about 61% on a year earlier C (calculated).²

The bull case

In plain words: the biggest buyers of computing have committed to build for years, there are few contractors who can wire a campus on time, and Everus has the crews, a clean balance sheet and a growing share of that work. Management makes this case itself. The CEO said that the backlog “underscores the strength of our industry position” and that opportunities are strong “including the data center, high tech and hospitality submarkets.” On the call he said “We haven’t seen any change of behavior from our customers.”²⁵ The company has raised its 2026 forecast three times, and expects EBITDA about 50% above the midpoint of its February range C (calculated).²³² I found no letter, filing or statement by a named outside investor, fund or analyst making a case for Everus that I could verify and date. The largest holders, BlackRock with 11.9% and Vanguard with 11.0%, have published no case for the company that I found.¹⁰ The analyst questions on the call, from Cantor Fitzgerald, Stifel, D.A. Davidson and Oppenheimer, were about margin, Epsilon and election risk. They are not statements of a view.⁵

Against reality

The bull case needs three things at once: the buyers keep spending through 2028, the margin stays near 9%, and cash catches up with profit. If they do, my bull value of $99 is possible, and the price of $79.71 sits between my base and bull. Where is the evidence today? The spending is real, at 91% growth for three buyers in nine months.²⁵²⁶²⁷ The margin has risen, but it rests in part on re-estimates that were 6.2% of second-quarter revenue and on a one-time gain of $7.6M.¹¹² Cash conversion over five years was 56% of net income.⁴³ The forecast for the fourth quarter says EBITDA will be no higher than the third. My base value of $64 assumes the first holds to 2028 and the second fades to 7.2%.

It helps to multiply it out. My base needs four things. Big-buyer spending staying high through 2028: say 70%. The margin staying above 8% in 2029: 60%. No large fixed-price loss: 85%. The two deals earning their price: 75%. Those are my guesses, and they multiply to 27% E (estimate). They overlap, so in practice it is somewhat higher, perhaps a third. The price needs more than that: a margin that never fades below 8.7% on top of the same growth. That is the bull path for the margin, with my base growth, and I put it nearer one in five E (estimate).

What a 10x would take. Ten times $79.71 is $797 a share, or about $40.8B of market value. At 20 times earnings that is $2.0B of net income, 6.5 times the $0.32B I expect for 2026 E (estimate). At Everus’s present net margin of about 6.3% it needs $33B of revenue in 2036, 6.5 times 2026 and 21% growth a year for ten years. At Comfort Systems’ 2025 operating margin of 14.4%, or about 10.7% after tax, it needs $19B, or 14% a year C (calculated).²⁴ The nearest precedent is Comfort Systems itself: its shares went from $46.32 on 4 November 2020 to $1,091.31, or 23.6 times, on revenue that went from $2.86B to $9.10B and an operating margin from 6.7% to 14.4%. Operating income rose 6.9 times, so the multiple the market paid rose about 3.4 times C (calculated).³⁸²⁴⁹ That took a data-center boom, a doubling of margin and a market willing to pay more each year. Everus starts at a margin of 8.8% and at a value of $4.1B, against Comfort Systems’ $1.7B. A tenfold rise by 2036 is possible only on a path like Comfort Systems’; I put it under 5% E (estimate).

Where I come down

What moved me is the order book: 61% organic growth in a year, and a 2027 that, if backlog holds, needs less new work than 2026 delivered. What didn’t is the margin. I can’t tell how much is scarcity and how much is old jobs finishing well, and the company has not told us either. What would raise my values: re-estimates falling to 3% to 4% while gross margin stays above 13%, and cash conversion near two thirds in 2027. A customer-concentration figure that drops, or a second quarter of orders above $2B that is not from acquisitions, would help too.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• Power. Data centers need power and transmission, and the company says availability creates “uncertainties surrounding the longevity” of this work.³ If connections are slow, projects slip, and delays hit a fixed-price contractor first. Analysts asked on the call about election-linked pauses or moratoriums.⁵
• One customer. 16.5% of revenue and 16.3% of receivables in 2025.³ I can’t see the contract. A dispute or a re-bid would show up at once.
• Labor and cost. About 85% of staff are union members, which helps with supply and also means wage and pension terms are fixed by agreement.³ The company participates in multiemployer pension plans and can’t estimate a withdrawal liability.³ The CFO acknowledged “there is a labor challenge” over the long term.⁵
• Acquisitions. Two deals in six months, one in modular building, which converts backlog more slowly. The CFO said Epsilon will convert backlog “at a slower pace” and be “a little bit lumpy.”⁵ Goodwill and intangibles are 21% of assets.²
• Estimates and accounting. Revenue is booked on cost-to-cost estimates, and the company discloses large favorable adjustments.¹¹ The auditor changed in January from Deloitte to KPMG, with no disagreement stated.³⁴ Bad news on one large job could change the picture quickly.
• Insurance and surety. Premiums have risen since the separation and may keep rising; bonding capacity depends on the balance sheet.¹¹³
• Short history. Two audited public years, both in a boom. Earlier figures are carve-outs, and my model uses them only as background.⁴
• Tax in the spin-off. Actions such as an acquisition of Everus or a large share issue could make the distribution taxable to MDU, and Everus could owe under the tax matters agreement. This limits some capital raising.³