EMCOR Group
This is the archived full research. Read the story
Electricians and pipefitters at national scale: EMCOR wires and cools other people's buildings, now above all data centres, and wins each job by bid.
The scores
Growth3/4 ↑
Revenue grew 16.6% then 19.7%, but it narrows to one driver and a fifth of the half's growth was bought.
Business quality3/4 →
Bonded, staffed capacity at $200m project size is a real advantage with share gains behind it, but its own filing denies barriers.
Financial position4/4 →
$924m of cash, no drawn borrowings, no maturity wall, and $750m of acquisitions funded from the balance sheet without issuing anything.
Execution record4/4 →
Eight consecutive beats, guidance raised twice in 2026, and the disposal gain's margin effect disclosed by the company unprompted.
Per share capture4/4 →
A quarter of the shares retired since 2018, every deal in cash; existing holders' slices grew rather than shrank.
Potential2/4 →
A large prize, but the upside is somebody else's capital budget: no named customer, no contract term, no minimum volume anywhere.
Risk profile3/4 ↓
One serious dependency - a third of revenue on data centre demand - plus a single jurisdiction, disclosed and survivable, no binary event.
Valuation3/4 →
About 18% under a base-case fair value of about $739 at an 8% required return: fair, inside the 20% margin of safety.
Evidence quality3/4 →
Everything central is filed and current, but the acquisition economics and the demand outlook rest on management's account of both.
The Verdict
We rate EMCOR Attractive. It sells skilled crews, not products: licensed electricians and pipefitters it moves to whichever market is growing, now data centres, and the business is strong, with no debt and cash matching profit. We value it at about $739 a share against $608, a little short of the cushion we usually ask for. We accept the thinner cushion because the rest is unusually solid: a record backlog of signed work, and a price that assumes growth barely above the economy's even though our value already lets the record margin fall back. The risk is that margin: back at its old average, the value drops below the price.
The long caseData centres are the one part of American building still growing while the rest shrinks, and EMCOR moves its electricians and pipefitters to wherever that work is.
How we weighed it
The price against our value
Cuts both ways
Price at 82% of our value
The price is under our value by just less than the margin of safety, and that value already assumes the record margin falls back.
The business
In its favour
Strong
It moves crews to whichever market is growing, grew through a collapse in factory work, holds net cash and turns its profit into cash.
Growth against what the price assumes
In its favour
The price assumes growth barely above the economy's from today, against a record backlog of signed work that grew by nearly half in a year.
What could go wrong
Cuts both ways
A third of revenue rides on data-centre building, newer contracts pay a thinner margin, insiders sold near the high, and at the old average margin the value falls below the price.
Against competitors
Cuts both ways
It is the cheapest of its listed rivals, but the group fell together, so the gap says little by itself.
How its group is behaving
Cuts both ways
Contractors that build for data centres have been marked down as one group on fears the boom slows; that is the group's fact, and it has not passed.
The latest news and report
In its favour
The latest report beat forecasts with a record backlog and better cash; the full-year cash figure and the mechanical margin are still open.
Where this departs from the fair value
The margin of safety stands in for what a valuation cannot see. Here most of that is visible and sound: signed work, no debt, cash behind the profit, and a base case that already assumes a lower margin. A slightly thinner cushion is enough.
What moved it from Neutral
The view is now a judgement of the whole thesis, not the price band alone: the record backlog, the balance sheet and how little growth the price assumes outweigh a cushion slightly thinner than the rule's.
The evidence under this - 7 claims
The two end markets are not equivalent in difficulty and the substitution was helped by timing: semiconductor work rolled over gradually as projects reached completion while data centre demand was already accelerating, so the crews had somewhere to go. A simultaneous decline in both would test the mechanism and has not happened. Management also concedes it balances opportunities geographically rather than moving labour freely, and that some rural markets need prefabrication or non-union operations it does not usually run.
The value leans on the margin: at the 2021-2025 average of 7.35% and 6% growth it is about $446, well under the price, and data centre building is itself a cycle. The 12% starting growth is below the 22% of the last year and the 47% rise in work still to do, so the base case may understate a business whose mix has moved.
- 2026-11-09editedre-struck as a three-stage fair value after the one-stage normal year held growth at 4% from today
- 2026-11-09editedRe-struck on the upgraded method: every view read at 9%, the return on new capital set by the moat, and capital per dollar of sales read from the filings.
- 2026-11-09editedRe-struck at the 8% required return the owner chose after the roll-out screen: the market priced the median name on the roster at about 7.7%, so 9% on top of a 20% margin read most of it Unattractive.
- 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
- 2026-11-10editedre-valued on the owner's rules of 10 November: the return on new capital glides to its terminal level over years six to ten; growth the customers fund needs only the plant
Cheapest of five is not cheap: the whole cohort is exposed to the same data centre demand and fell together, so a discount to rivals priced on the same boom margins says nothing about whether the margins last. Against its own six-year medians EMCOR is dearer on every measure - 23.8 times trailing earnings against a 19.6 median and 14.9 times enterprise value to EBITDA against 11.9 at the September pack.
- 2026-09-26editedRe-struck on the research pack refreshed 2026-09-26 at the 25 September close of $762.21: 23.7 to 23.4 times guided earnings, 17.9 to 17.6 times EV/EBIT. Comfort Systems, held out of the EV/EBIT cohort at the thesis by our own extraction bug (not a vendor gap), is now priced at 30.8 times, so EMCOR is cheapest of five rather than four and the median moves from 19.9 to 23.2; the counter-evidence says why that wider gap is the rivals' premium and not a larger discount.
- 2026-11-08editedRe-struck under the valuation rules of 2026-11-08
- 2026-11-09editedRe-struck at the 6 November close; the rivals' 23.2 median was a 25 September reading and is carried to the same close, near 15.
The company
Every large building needs power distribution, cabling, cooling, piping and fire protection, installed by licensed tradespeople. EMCOR supplies them through about 100 operating companies, and then services what it installs.
It wins work by bidding, job by job, and its own filing admits the trade has few barriers to entry. What small firms cannot take is the largest jobs, $10M to over $200M: about 58% of its construction revenue in 2025.
The evidence under this - 4 claims
How it makes money
It is paid as each job is built, and again for servicing the systems afterwards. It kept 10.6 cents of each sales dollar as operating profit in the June quarter, against 9.2 cents in 2024 and 2025. More of the new data-centre work is priced as cost plus a fee, which caps what EMCOR earns on a job while leaving it responsible for delivery.
The evidence under this - 5 claims
Management put the guaranteed-maximum-price shift at 9% to 10% of the mechanical mix and framed it as prudent for both sides on complex first-of-a-kind systems, and the segment still earned 12.5%, which it says is in line with its own trailing twelve and twenty-four month averages. Lower markup on pass-through materials and subcontractors is not the same as lower profit per hour of EMCOR's own labour.
A capped fee also caps the downside, and management argues the shift is prudent on first-of-a-kind mechanical systems where a fixed price would be priced with a large contingency or lost. On unfamiliar work a cost-plus form can be worth more than a fixed price with the risk mispriced, and EMCOR's mechanical segment margin remains above its own twenty-four month average.
Where the money comes from, stream by stream
| Stream | Share of sales (Q2 2026) | How it earns it |
|---|---|---|
| US mechanical construction | 44.6% | HVAC, piping, plumbing, fire protection and process systems installed in new and renovated buildings, including data centres, mostly on bid projects. |
| US electrical construction | 32.3% | Power distribution, lighting, low-voltage and fibre cabling in buildings; network and communications, which holds the data centre work, was 58% in the quarter. |
| US building services | 16.3% | Maintaining, repairing and retrofitting mechanical systems and running facilities for building owners and government sites, mostly recurring service work. |
| US industrial services | 6.9% | Field and shop services for refineries and petrochemical plants: turnarounds, heat exchangers and maintenance. |
Who runs it
Tony Guzzi has run EMCOR since 2011 and was its president from 2004; Jason Nalbandian, who rose through the finance team, became finance chief in 2024. Insiders own under 1% of the company, so these are hired managers rather than owners. The yearly bonus pays on earnings per share and on how much operating profit arrives as cash; the long-term award pays on three years of earnings per share. Nothing pays on return on capital, which matters as more cash goes into acquisitions.
The evidence under this - 3 claims
Why it wins, and who could beat it
Why customers come back
Big jobs need a surety bond, an insurer's guarantee the work gets finished, and crews to staff them. EMCOR carried about $3.03B of bonds at the end of 2025. Crews are the scarcer part: most states license electricians, most train as apprentices, and the trade cannot grow fast.
The evidence it works: in the June quarter EMCOR's data-centre revenue grew 67.3%, while data-centre building across the country grew 34.1%. It is taking share.
The evidence under this - 5 claims
Part of the growth was bought: EMCOR recorded $169.2m of incremental acquisition revenue across the whole company in the quarter, some of it in these segments, so the organic share gain is smaller than the gap suggests. The sector definition is also wider than data centres, and if fibre and cabling grew faster than data centre work the gap would say nothing about data centre share at all.
The threat
Comfort Systems bids for the same jobs and has leaned harder into them: technology customers were 58.7% of its June-quarter revenue, and it grew 50.3%. Quanta, several times EMCOR's size, now says in its annual report that it does this work inside data centres too.
The evidence under this - 2 claims
The rivals: FIX, FIX, PWR, ECG, MYRG, IESC, TPC, APG.
Who else does this, rival by rival
| Ticker | Rival | How far along | Priced against it |
|---|---|---|---|
| FIX | Comfort Systems USA, Inc.Priced against it: Yes | On sale | Yes |
| Comfort Systems USA as a bidder for the same private contractors | On sale | ||
| PWR | Quanta Services, Inc.Priced against it: Shown, not counted | On sale | Shown, not counted |
| ECG | Everus Construction GroupPriced against it: Yes | On sale | Yes |
| MYRG | MYR GroupPriced against it: Yes | On sale | Yes |
| IESC | IES HoldingsPriced against it: Yes | On sale | Yes |
| TPC | Tutor PeriniPriced against it: Shown, not counted | On sale | Shown, not counted |
| APG | APi GroupPriced against it: Shown, not counted | On sale | Shown, not counted |
How big it could get
How big this could get
Americans built data centres at a $66.1B yearly rate from April to June 2026, up 34.1% on a year earlier. That market is now a third of EMCOR's revenue.
The limit is people, not demand: the country had 821,000 electricians in 2025, and the government expects about 0.9% more a year.
The evidence under this - 3 claims
Who has signed
No customer is named, and none is 10% of sales. What is booked is $17.14B of work still to do, up 43.9% in a year. Less of it is due soon, though: 76.0% within a year, down from 81.6% in December and a norm near 85%.
The evidence under this - 3 claims
EMCOR's own risk factors say amounts in remaining performance obligations may not result in actual revenues, that many contracts are subject to cancellation or suspension on short notice at the discretion of the client, and that cancellation risk generally increases during economic slowdowns. The 10-Q says the cancellation penalty makes the risk very low, but that is the company's own statement.
- 2026-10-29editedRe-sourced to the September 10-Q: $18.57bn against $17.14bn at June.
- 2026-10-29editedRe-sourced to the September 10-Q one-year table: 77.3%, a small recovery from June's 76.0% that is still below the year-end share.
The results
The numbers
Sales grew 22% over the twelve months to Sep 2026, faster than the 17% in the year to Dec 2025.
Operating margin is 10%, up from 7.0% in the year to Dec 2023.
Running costs take 9¢ of each sales dollar, about where it was (10¢) in the year to Dec 2021.
Each dollar put into the business earns 40% a year after tax, up from 16% in the year to Dec 2021.
It turned 7.5% of sales into free cash, $1.48B over the last twelve months.
Yearly sales growth has ranged from +12% to +17% over the last 5 years.
The share count fell 19% in 4.8 years.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-11-10. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Sep 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $9.90B | $11.1B | $12.6B | $14.6B | $17.0B | $19.8B |
| Sales growth | +13% | +12% | +14% | +16% | +17% | +22% |
| Profit | ||||||
| Gross margin | 15% | 14% | 17% | 19% | 19% | 20% |
| Operating margin | 5.4% | 5.1% | 7.0% | 9.2% | 10% | 10% |
| Net margin | 3.9% | 3.7% | 5.0% | 6.9% | 7.5% | 7.8% |
| What growth costs, per sales dollar | ||||||
| Selling, general and admin | 9.8% | 9.4% | 9.6% | 9.7% | 10% | 9.7% |
| Capital spending | 0.4% | 0.4% | 0.6% | 0.5% | 0.7% | 0.6% |
| Cash | ||||||
| Cash from operations | $319M | $498M | $900M | $1.41B | $1.30B | $1.61B |
| Free cash flow | $283M | $449M | $821M | $1.33B | $1.19B | $1.48B |
| Balance sheet | ||||||
| Cash and short-term investments | $821M | $456M | $790M | $1.34B | $1.11B | $775M |
| Debt | $262M | $247M | $5.30M | $0 | - | - |
| Net cash | $560M | $209M | $784M | $1.34B | - | - |
| Returns and shares | ||||||
| Return on capital | 16% | 17% | 27% | 36% | 38% | 40% |
| Shares (diluted) | 54.3M | 50.1M | 47.6M | 46.8M | 45.1M | 44.2M |
What growth costs
Growing does not make EMCOR cheaper to run: overhead has stayed close to a tenth of sales for six years while sales doubled. Its profit margin did widen, but because each job earned more, not because overhead was spread thinner, and job margins can shrink again, as one division's did this year.
The evidence under this - 2 claims
Management put the guaranteed-maximum-price shift at 9% to 10% of the mechanical mix and framed it as prudent for both sides on complex first-of-a-kind systems, and the segment still earned 12.5%, which it says is in line with its own trailing twelve and twenty-four month averages. Lower markup on pass-through materials and subcontractors is not the same as lower profit per hour of EMCOR's own labour.
The last report (29 Oct)
Quarter ended September 2026, filed figures.
The September quarter beat forecasts on profit and sales. Work still to do reached a record $18.57B, up 47.3% in a year, and data-centre-type network work roughly doubled to $2.03B in the two construction segments. Profit margin rose to 9.8%, or 10.5% without a one-off write-down.
Cash caught up: in nine months the business turned 98.0% of net income into operating cash, from 41% at the half, and management expects $1.5B to $1.7B for the year. The mechanical margin fell again, to 12.1%, but is said to hold near 12% to 12.5%. The full-year forecast rose a third time, to $20.6-21.0B of sales, and the buyback grew by $500M.
What it did to the thesisIt supported the case: work still to do and cash both improved. Two things stay open, the full year finishing level with profit and the mechanical margin still falling, so the view and its confidence hold.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $9.60 against $8.37 expected.
The evidence under this - 7 claims
EMCOR's own risk factors say amounts in remaining performance obligations may not result in actual revenues, that many contracts are subject to cancellation or suspension on short notice at the discretion of the client, and that cancellation risk generally increases during economic slowdowns. The 10-Q says the cancellation penalty makes the risk very low, but that is the company's own statement.
- 2026-10-29editedRe-sourced to the September 10-Q: $18.57bn against $17.14bn at June.
Conversion is still just below the stated rule of at least 100%, and changes in operating assets and liabilities still took out $220.6m. Third-quarter cash is large partly because the first half was so low; one quarter does not make a year, and the fourth quarter has to hold it.
Management put the guaranteed-maximum-price shift at 9% to 10% of the mechanical mix and framed it as prudent for both sides on complex first-of-a-kind systems, and the segment still earned 12.5%, which it says is in line with its own trailing twelve and twenty-four month averages. Lower markup on pass-through materials and subcontractors is not the same as lower profit per hour of EMCOR's own labour.
EMCOR has raised guidance three times and beaten eight quarters running, which is as consistent with conservative guiding as with strengthening demand (C22 counter). The call on the same morning, where the second-half margin bridge would be put to management, was not yet held.
- Management said the net contract liabilities rose with advance billings and that slower growth would turn working capital into cash, so third-quarter cash is helped by timing (S17).
- The cash range is a rule-based estimate, not a filed result, and the first half converted only 41% (C29).
What management did with the money
It agreed to buy five electrical contractors for $750M, 7.1 times their yearly cash profit, while the market values EMCOR at 16.0 times. Its share count fell 23.8% from 2018 to June 2026, and it has no borrowings.
Insiders in 2026Q2: 13 purchases and 7 sales, from their Form 4 filings.
The evidence under this - 4 claims
The figures come from a slide and the call and are not in the 10-Q, which reports only the four smaller companies bought in the first half for $99.8m. Schmidt and Connelly, about 75% of the total, had not closed at the time of the call, so both the price and the trailing financials are management's account of a transaction not yet on the balance sheet. The trailing EBITDA is also unaudited and pre-integration.
The gap is not free money. The acquired backlog is written up and amortised, which management says will limit the earnings contribution for the remainder of 2026, and Comfort Systems is bidding for the same assets - it paid $206.0m for Hunt Electric in May 2026 and $109.8m for Feyen Zylstra in October 2025. A private electrical contractor with data centre exposure is no longer an unbid asset, and the multiple paid should rise. The earnout of up to $90m would also take the price to 8.0 times if fully earned.
- 2026-09-26editedEMCOR's own cohort EV/EBITDA re-struck on the research pack refreshed 2026-09-26 at the 25 September close: 16.2 to 16.0 times. The 7.1 times paid is unchanged.
Cash fell $149.6m in the quarter while operating cash flow was strong, so the balance sheet is being spent on acquisitions and buybacks, not hoarded.
- 2026-10-29editedRe-sourced to the September 10-Q: cash $774.8m, down from $924.4m.
The price, and where we disagree
In short. Fair. About 82% of our base-case value, which already assumes the record margin falls back.
Since the May high the shares have fallen by about a third while the company raised its forecast, and its close rivals fell as much or more. Buyers are paying less for profit they suspect is at a peak.
Underneath, data centres are the only private building market still growing: the rest fell 8.2% in the June quarter.
The evidence under this - 3 claims
A falling multiple on rising guidance is only an opportunity if the guidance is durable; it is the correct market response to earnings the market believes are cyclical. The stock also rose 19.3% on the day of the second raise before giving most of it back, which is the market re-testing and re-rejecting the same case.
Cheapest of five is not cheap: the whole cohort is exposed to the same data centre demand and fell together, so a discount to rivals priced on the same boom margins says nothing about whether the margins last. Against its own six-year medians EMCOR is dearer on every measure - 23.8 times trailing earnings against a 19.6 median and 14.9 times enterprise value to EBITDA against 11.9 at the September pack.
- 2026-09-26editedRe-struck on the research pack refreshed 2026-09-26 at the 25 September close of $762.21: 23.7 to 23.4 times guided earnings, 17.9 to 17.6 times EV/EBIT. Comfort Systems, held out of the EV/EBIT cohort at the thesis by our own extraction bug (not a vendor gap), is now priced at 30.8 times, so EMCOR is cheapest of five rather than four and the median moves from 19.9 to 23.2; the counter-evidence says why that wider gap is the rivals' premium and not a larger discount.
- 2026-11-08editedRe-struck under the valuation rules of 2026-11-08
- 2026-11-09editedRe-struck at the 6 November close; the rivals' 23.2 median was a 25 September reading and is carried to the same close, near 15.
Public nonresidential construction is a separate series and is not falling, and EMCOR's institutional revenue - schools, colleges, public sector - grew strongly in the same quarter. So the private series overstates the contraction in EMCOR's whole addressable market, though not in the private part of it.
How we value it
- Method
- Cash the business will produce, in three stages
- Required return
- 8%
- Long-run growth
- 4%
- Moat
- An average moat
| Bear | Base | Bull | |
|---|---|---|---|
| Sales growth, years 1 to 3 | 6% | 12% | 18% |
| Normal operating margin | 7.3% | 9% | 10% |
| Return on new capital after year ten | 8% | 12% | 15% |
| Value a share | $446 | $739 | $1,013 |
- Price
- $608
- Priced on
- 6 Nov
- Price as a share of the base value
- 82%
- Starting growth the price pays for
- 4.4%
- Yearly return the price offers on the base case
- 9%
- Risk check: valued at 10%, the rate for a cyclical or indebted business
- $516
Where each figure comes from
| Figure | Source | Where exactly |
|---|---|---|
| Sales, the last twelve months | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the four quarters to that date. |
| Operating margin today | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the four quarters to that date. |
| Tax rate | Our judgement | Our reading of the evidence. |
| Sales per dollar of new capital | Filed report | Annual report (10-K) for the period to 31 Dec 2025, filed 26 Feb 2026: the year's sales over its operating capital, above the cap of 6 because its customers fund its growth. |
| Shares | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the diluted share count. |
| Cash and investments | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the balance sheet on that date. |
| Debt | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 29 Oct 2026; the balance sheet on that date. |
| Sales growth, years 1 to 3 | Our judgement | Our reading of the evidence; each case below names it. FMP's consensus of 6 to 10 analysts read 16.2%, 10.6%, 2%. |
| Normal operating margin | Our judgement | Our reading of the evidence; each case below names it. |
| Return on new capital after year ten | Our method's rule | Set by our reading of the moat: the 8% rate with none, 12% with an average one, 15% with a strong one; the bear case earns the rate. |
| The bear and bull cases | Our judgement | Our reading of the evidence; each case below names it. |
| Price | Market price | FMP's closing price on 6 Nov 2026. |
What each case rests on
Bear caseData centre building turns: growth starts at 6%, the margin falls back to the 2021-2025 average of 7.35%, and new capital earns only the 9% required.
Base caseGrowth starts at 12%, below the 14% a year of 2021-2025 and the 47% rise in work still to do; the margin settles at 9.0%, its 2024-2025 level, as fee-based data centre contracts grow.
Bull caseGrowth starts at 18%, near the 22% of the last year, and the margin holds at 10.0%, close to today's 10.4%, because the new mix is the new normal.
What moves the base value
| Lower | Higher | |
|---|---|---|
| Growth in years 1 to 3, a quarter lower or higher | $685 | $797 |
| The normal margin, 15% lower or higher | $642 | $837 |
| Capital per dollar of growth, a quarter more or less | $738 | $740 |
| Return on new capital after year ten, at the rate or one step up | $611 | $791 |
| The rate, a point higher or lower | $606 | $961 |
| Long-run growth a point lower | $665 | - |
Value a share as growth and margin change
| Normal margin ↓ · growth → | 50% | 75% | 100% | 125% | 150% |
|---|---|---|---|---|---|
| 7.2% | $524 | $565 | $609 | $656 | $706 |
| 8.1% | $579 | $625 | $674 | $727 | $782 |
| 9% | $634 | $685 | $739 | $797 | $859 |
| 9.9% | $689 | $745 | $805 | $868 | $936 |
| 10.8% | $745 | $805 | $870 | $939 | $1,013 |
Growth is shown as a share of the base case's. Green: the price is at or under 80% of that value. Red: at or over 120%.
How we read the multiple
Counting its cash, the company costs about 13 times its profit from the business over the last year, earned at a record margin of 10.4 cents a dollar.
We value it in three steps: sales growth starts at 12% and slows toward 4% over ten years, the margin settles at 9 cents, and a buyer wants 8% a year. That gives about $739 a share. The $608 price assumes growth starting near 4%. On the old average margin the value is about $446; if the record margin holds, about $1,013.
Against its own history. Dearer than its normal on every measure. It also earns far more on what it owns than it did, so much of that gap is a better business.
The evidence under this - 4 claims
Cheapest of five is not cheap: the whole cohort is exposed to the same data centre demand and fell together, so a discount to rivals priced on the same boom margins says nothing about whether the margins last. Against its own six-year medians EMCOR is dearer on every measure - 23.8 times trailing earnings against a 19.6 median and 14.9 times enterprise value to EBITDA against 11.9 at the September pack.
- 2026-09-26editedRe-struck on the research pack refreshed 2026-09-26 at the 25 September close of $762.21: 23.7 to 23.4 times guided earnings, 17.9 to 17.6 times EV/EBIT. Comfort Systems, held out of the EV/EBIT cohort at the thesis by our own extraction bug (not a vendor gap), is now priced at 30.8 times, so EMCOR is cheapest of five rather than four and the median moves from 19.9 to 23.2; the counter-evidence says why that wider gap is the rivals' premium and not a larger discount.
- 2026-11-08editedRe-struck under the valuation rules of 2026-11-08
- 2026-11-09editedRe-struck at the 6 November close; the rivals' 23.2 median was a 25 September reading and is carried to the same close, near 15.
The value leans on the margin: at the 2021-2025 average of 7.35% and 6% growth it is about $446, well under the price, and data centre building is itself a cycle. The 12% starting growth is below the 22% of the last year and the 47% rise in work still to do, so the base case may understate a business whose mix has moved.
- 2026-11-09editedre-struck as a three-stage fair value after the one-stage normal year held growth at 4% from today
- 2026-11-09editedRe-struck on the upgraded method: every view read at 9%, the return on new capital set by the moat, and capital per dollar of sales read from the filings.
- 2026-11-09editedRe-struck at the 8% required return the owner chose after the roll-out screen: the market priced the median name on the roster at about 7.7%, so 9% on top of a 20% margin read most of it Unattractive.
- 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
- 2026-11-10editedre-valued on the owner's rules of 10 November: the return on new capital glides to its terminal level over years six to ten; growth the customers fund needs only the plant
Priced against its rivals
The competitors its own annual report names, each confirmed against the rival's filing. Measured on company value for each $1 of operating profit, cheapest first:
| Ticker | Company | company value for each $1 of operating profit |
|---|---|---|
| EME | EMCOR Group | 17.6× |
| ECG | Everus Construction Group | 18.6× |
| MYRG | MYR Group | 20.2× |
| IESC | IES Holdings | 26.2× |
| FIX | Comfort Systems USA | 30.8× |
| TPC | Tutor Perini not in the median | 14.8× |
| APG | APi Group not in the median | 34.0× |
| PWR | Quanta Services not in the median | 51.0× |
The middle of the other 4 is 23.2. Tutor Perini Corporation, APi Group Corporation, Quanta Services, Inc. compete with it in one product line inside a much bigger company, so their price says little about this business and they are not counted in the middle. EMCOR Group, Inc. is cheapest of 5.
Rivals' figures come from a data provider, so read the spread of the group rather than any single pair.
The cheapest of its rivals, but they all fell together, so the gap says little about whether the margin lasts.
Also checked and set aside: Dycom Industries (a substitute, not the same product); MasTec (a substitute, not the same product).
Where we disagree with the market
EMCOR is an AI-boom stock at the top of a cycle, so each rise in profit deserves a lower price for each dollar.
A machine that moves bonded crews to whichever market grows, priced about fair on growth that fades and a margin a little below today's.
The market has reasons. Nine of fourteen analysts say buy, but only two publish a 2028 estimate: few will vouch for the years the price depends on.
Our evidence is the switch already made: factory work fell from 23% of the mechanical business to 10% while data centres rose from 13% to 35%, and revenue kept growing. Against us, that switch was helped by timing, the new work pays less, and if the margin returns to its old average the price is well above the value.
What analysts say, and the evidence
The evidence under this - 5 claims
- 2026-09-26editedRe-struck on the research pack refreshed 2026-09-26: two more FINRA settlements landed (912,626 shares at 31 August, 942,104 at 15 September), so 'fallen for two consecutive settlements' became three falls and a 3.2% rise, at the same 2.15% of float. Ratings unchanged at nine buy and five hold.
The two end markets are not equivalent in difficulty and the substitution was helped by timing: semiconductor work rolled over gradually as projects reached completion while data centre demand was already accelerating, so the crews had somewhere to go. A simultaneous decline in both would test the mechanism and has not happened. Management also concedes it balances opportunities geographically rather than moving labour freely, and that some rural markets need prefabrication or non-union operations it does not usually run.
Management put the guaranteed-maximum-price shift at 9% to 10% of the mechanical mix and framed it as prudent for both sides on complex first-of-a-kind systems, and the segment still earned 12.5%, which it says is in line with its own trailing twelve and twenty-four month averages. Lower markup on pass-through materials and subcontractors is not the same as lower profit per hour of EMCOR's own labour.
The value leans on the margin: at the 2021-2025 average of 7.35% and 6% growth it is about $446, well under the price, and data centre building is itself a cycle. The 12% starting growth is below the 22% of the last year and the 47% rise in work still to do, so the base case may understate a business whose mix has moved.
- 2026-11-09editedre-struck as a three-stage fair value after the one-stage normal year held growth at 4% from today
- 2026-11-09editedRe-struck on the upgraded method: every view read at 9%, the return on new capital set by the moat, and capital per dollar of sales read from the filings.
- 2026-11-09editedRe-struck at the 8% required return the owner chose after the roll-out screen: the market priced the median name on the roster at about 7.7%, so 9% on top of a 20% margin read most of it Unattractive.
- 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
- 2026-11-10editedre-valued on the owner's rules of 10 November: the return on new capital glides to its terminal level over years six to ten; growth the customers fund needs only the plant
What could go wrong, and how we would know
The risks
- The boom slows. About a third of revenue rides on data-centre building, and a sharp slowdown would stop the growth however well the company is run.
- Contracts that pay less. The move to cost-plus and capped-price data-centre work cut the mechanical margin by 1.1 points in the June quarter.
- Cash still a hair behind profit. The nine months turned 98.0% of net income into cash, against a rule of 100%.
- Insiders selling near the high. The finance chief, the controller and two directors sold within a fortnight of the May high, none under a plan set in advance.
The evidence under this - 6 claims
Management put the guaranteed-maximum-price shift at 9% to 10% of the mechanical mix and framed it as prudent for both sides on complex first-of-a-kind systems, and the segment still earned 12.5%, which it says is in line with its own trailing twelve and twenty-four month averages. Lower markup on pass-through materials and subcontractors is not the same as lower profit per hour of EMCOR's own labour.
The $525.9m is the net of the working capital block, not a single outflow, and receivables grew 20.2% against revenue growth of 19.7%, so days outstanding stretched only modestly. Contract liabilities of $2.61bn still exceed contract assets of $0.40bn, so customers fund the work overall. Management says cash flow is back-half weighted every year and that the first quarter is always weakest.
The algorithm held in 2024 at 1.40 times net income and roughly held in 2025 at 1.02 times, so it has a two-year record. The gap is a timing statement until the fourth quarter closes, and 2025's own first half converted at 56% before finishing at 102%.
- 2026-10-29editedNine-month cash conversion filed: 98.0%, from 41% at the half; the algorithm is now one quarter's cash from being met, not met.
Forty further Form 4 rows in the window are grants, vestings, withholding and gifts - compensation, not a view. The largest sale by far is now explained: a plan adopted three months before it executed, and before the guidance this thesis argues about, carries no information. What survives is smaller and sharper - a cluster of discretionary sales near the high - and an absent plan flag is the absence of a claimed defence rather than proof of discretion.
What we still do not know
- Who the data-centre customers are. No filing names one, so a change in their plans would show first in bookings.
The evidence under this - 1 claim
What would change our view
- Attractive→NeutralThe price reaches our value
It needs:
- ○The share price rises to about $739, what we think the business is worth.about ≈$745.98 today, -2% above the last close
- Attractive→NeutralThe work still to do starts to shrink
It needs:
- ○Booked work still to do falls on a year earlier, or less than 70% of it is due within a year.checked at the next results, due by 26 Feb
- Reopens the researchThe record margin proves to be the new normal
It needs:
- ○The margin, leaving out one-off gains, stays above 10% through 2027 while booked work keeps growing.checked at the next results, due by 26 Feb
- Attractive→UnattractiveThe price runs 20% above fair value
It needs:
- ○The price rises to about $887, 20% over our value of about $739.about ≈$896.01 today, 18% above the last close
We would also be wrong if
- Its data-centre revenue grows more slowly than data-centre building nationally for two quarters running.
- Full-year operating cash comes in below net income.
- It pays more than 10 times cash profit for a sizeable acquisition, or pays in shares.
What is coming
- February 2027 Fourth quarter and full-year 2026 results, with 2027 guidance: Whether 2026 lands inside twice-raised guidance, whether the bonus's cash ratio is met, and whether first 2027 guidance assumes data centre work keeps growing.
Since we wrote this
- 13 NovNewsEMCOR buys P1 Construction for cash; undisclosed price leaves the acquisition test unscored. View heldEMCOR agrees to buy P1 Construction, a Midwest mechanical and electrical contractor.
- 10 NovThesis reviewJudged Attractive on the whole picture, at 82% of our valueThe view is now a judgement of the whole thesis, the fair value its main evidence and no longer the decider.
- 9 NovThesis reviewNeutral: priced at 84% of a value near $728Re-valued in three stages at an 8% return, with growth slowing fastest early: the price sits inside the 20% margin of safety.
- 29 OctEarningsWork still to do hit a record; cash nearly caught upNine-month cash reached 98% of profit and booked work rose 47%; the mechanical margin fell again, so the view held.
- 29 OctEarningsQ3 2026 call: full-year cash and mechanical margin framedThe Q3 2026 earnings call
- 5 OctThesis reviewThe long case now sits under the verdictAfter the five-name pilot, the owner asked for the long case on every thesis.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
- 26 SepThesis reviewPrice route now measured on guided earnings, one multiple throughout. View heldThe route back to Neutral was labelled as one multiple but carried the figures of another. It now reads 30 times guided earnings per share throughout, from 23.7 at the thesis.
- 26 SepThesis reviewComfort Systems joins the price comparison: cheapest of five. View heldComfort Systems, the rival EMCOR names first, is now in the price comparison. EMCOR is the cheapest of five at 17.6 times operating profit, but its two nearest rivals are still priced much like it.
- 9 SepThesis writtenAttractive, medium convictionThe biggest bonded contractor in its trade, priced for its plan to be merely met, with cash lagging profit as the open question.
Also checked 4 times with nothing new to act on.
Sources 19
Sources - 19
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S1primarycompany filing2026-02-26Item 1 business description and the named competitor list, the market-sector revenue tables for 2023-2025, remaining performance obligations, the surety exposure, and the $144.9m gain on the United Kingdom disposal with its stated 85 basis point effect on operating margin.backsC01C02C03C07C08C09C13C19C24C25C32C39C41C46C47C48C51C52C58
- S10primarycompany filing2025-11-20A competitor's own description of the field it competes in: large national or regional competitors alongside small, privately owned contractors who generally have limited access to capital.backsC12
- S11primarycompany filing2026-02-25A competitor's segment split and its statement that its commercial work leverages deep expertise in the data center submarket, with union labour named as its delivery model.backsC42
- S12secondaryvendor data2026-09-26Research pack for EME, frozen at content/EME/research-pack.json (refreshed 2026-09-26)content/EME/research-pack.jsonPrice, market capitalisation, enterprise value, the trailing multiple series against its own six-year medians, the competitor cohort's multiples, analyst coverage and estimates, insider filings, short interest and the measured price reactions.backsC18C23C31C32C35C36C37C38C40
- S13primarycompany filing2026-09-15Each insider transaction with its Rule 10b5-1 affirmative-defence flag and, where one was claimed, the plan adoption date from the footnote - the field the vendor insider feed does not carry.backsC38
- S14primarycompany filing2026-04-21Security ownership of management and of holders above 5% at 7 April 2026; the Compensation Discussion and Analysis - the 2025 Annual Incentive Program matrix on adjusted earnings per share and the cash flow ratio, the LTIP's restricted stock units and three-year earnings-per-share cash award, and what was paid for 2025.backsC48C49C50
- S15primarycompany filing2026-10-29The quarter's and nine months' income statement and cash flow statement, the market-sector revenue disaggregation by segment, segment operating margins, remaining performance obligations with their one-year split and the cash balance.backsC20C21C29C30C35C53C54C55C58
- S16primarycompany filing2026-10-29The third raise of 2026 guidance with the previous range beside it, the non-GAAP reconciliation, and the $500 million increase to the share repurchase programme.backsC22C56
- S17primarytranscript2026-10-29Q3 2026 earnings call transcript (FMP)data/cache/transcripts/EME-2026Q3.jsonmanagement's account of full-year cash, mechanical margin and data-centre mix; analysts' questions are a separate origin from the answers
I think the level we're running at today is where we're going to be running for probably the next 12 months or so, right?
backsC57 - S18primarycompany filing2026-11-09Revenue and GAAP operating income for each fiscal year 2021 to 2025 and the trailing four quarters, the diluted share count and the cash balance at 30 September 2026.backsC58
- S19tertiaryvendor data2026-11-09Completed-session closes on 25 September and 6 November 2026 for the subject and its four filed rivals.backsC35
- S2primarycompany filing2026-07-30The quarter's and half's income statement, balance sheet and cash flow statement, the market-sector revenue disaggregation by segment, remaining performance obligations with their one-year split, the share count reconciliation and the acquisitions note.backsC03C04C08C09C14C19C20C21C23C25C28C29C30C31C33C34C39C45C46
- S3primarycompany release2026-07-30The raised full-year 2026 guidance and the previous guidance it replaced, segment operating income and margins, and the capital returned in the half.backsC22C23C27C36C43
- S4primarycompany release2026-02-26The initial 2026 guidance issued in February, full-year 2025 segment revenues and margins, and the after-tax gain on the United Kingdom sale.backsC22C33
- S5primarytranscript2026-07-30The acquisition batch's price and trailing financials, organic growth excluding acquisitions and the divestiture, the change in how fast remaining performance obligations burn, the guaranteed-maximum-price mix shift, and management's statement that the quarter's margin is not in guidance.backsC09C17C18C20C21C26C27C29C34C43C44
- S6primarygovernment data2026-09-01The data center line of private construction put in place and every other private nonresidential category beside it, monthly, at a seasonally adjusted annual rate.
Value of Private Construction Put in Place - Seasonally Adjusted Annual Rate (Millions of dollars. Details may not add to totals since all types of construction are not shown separately.) ... Office | Jul 2026p 123318 | Jun 2026r 119328 | May 2026r 115159 | Apr 2026 111220 | Mar 2026 107164 | Jul 2025 101666 | 3.3 | 21.3 ... Data center | 75166 | 70755 | 65688 | 61859 | 58121 | 47810 | 6.2 | 57.2 ... Nonresidential | 755169 | 752375 | 749441 | 745266 | 745125 | 781119 | 0.4 | -3.3 ... Manufacturing | 167805 | 169106 | 171381 | 175033 | 178488 | 214353 | -0.8 | -21.7
backsC05C06C14 - S7primarygovernment data2026-09-09The size of the United States electrician workforce, its projected ten-year growth, annual openings and the licensing and apprenticeship route into the trade.
Quick Facts: Electricians ... Number of Jobs, 2025 821,000 ... Job Outlook, 2025-35 9% (Much faster than average) ... Employment Change, 2025-35 75,900 ... About 72,700 openings for electricians are projected each year, on average, over the decade. ... Most electricians learn through an apprenticeship, but some start out by attending a technical school. Most states require electricians to be licensed.
backsC10C11 - S8primarycompany filing2026-07-24A direct competitor's own revenue by type of customer, showing technology customers at $1,917.4m and 58.7% of its quarterly revenue, and its two recent electrical-contractor acquisitions.backsC15
- S9primarycompany filing2026-02-19A $27bn competitor stating in its own overview that it provides low voltage electrical, mechanical, plumbing and process infrastructure for data centers - EMCOR's product, declared by an entrant.backsC16