Everus Prints a Record Quarter and Wears a 52% Drawdown

On November 4, Everus Construction Group delivered the kind of quarter that usually gets rewarded: record third-quarter revenue of $1.38 billion, up 40%; EBITDA of $140.2 million, up 57.5%; EBITDA margin of 10.1%, up 110 basis points; and a total solid backlog of $5.48 billion, up 86% year over year. Management raised full-year 2026 guidance to $5.0–5.1 billion of revenue and $485–495 million of EBITDA. Net leverage was 0.7x, roughly half the low end of the company's own 1.5x–2.0x target.

The stock, meanwhile, closed the session 52.9% below its May 6 peak of $169.16 and down 52% over the last ninety trading days. That gap — not the revenue line — is the story.

A quarter that should have worked

The operating detail is genuinely good. Excluding acquisitions, revenue grew 36% organically. Electromechanical revenue rose 48% to $1.14 billion with segment EBITDA margin of 10.5%, up 180 basis points; T&D rose 14% to $255.6 million at a steady 15% margin. Roughly $742 million of backlog arrived with SE&M and Epsilon. Free cash flow for the first nine months was $221.2 million versus $74.8 million a year earlier — though Max Marcy was explicit that some of that was working-capital timing, not a new run rate.

The filed numbers confirm the direction rather than the headline. In the last 10-Q (Q2, filed August 5), total revenue was $1.23 billion, up 34% year over year, with operating margin at 9.1%, up 1.2pp. Capital intensity stayed low: capex of $20 million against that revenue base. Interest coverage reached 24.2x.

The tape disagrees with the backlog

Why would a company with a record book get cut in half? Because the largest single component of that book is data centers, and the market has spent the last quarter liquidating the AI-infrastructure complex. In the global price tape, AI data center is a 30-day decliner at -21% with fifty-six negative tickers against eleven positive, and data center capacity names are down nearly 39% over the same window. Everus is not being valued on its execution; it is being valued as a levered proxy on hyperscaler construction sentiment.

Analysts asked precisely this, and Jeffrey Thiede pushed back without a data point: “We have not seen any change of behavior from our customers or our outlook for data centers is strong.” — Kurt, Analyst · 2026-11-04 He then walked the diversification ledger — hospitality, warehouse, semiconductor, government, healthcare, renewables, substation and transmission all up year over year in backlog.

The problem is that this concentration has been disclosed in identical language for at least three quarters. In August: “Data centers, as we've said before in previous quarters, and it still is the case. It's the largest part of our backlog.” — Jeff Thiede, Chief Executive Officer · 2026-08-05 A diversification strategy that leaves the largest end market unchanged is a hedge, not a pivot — and the tape is now pricing the unhedged part.

Fifty-five basis points nobody wants to annualize

Buried in the quarter was a one-time net gain of $7.6 million from settling outstanding billings on a past project — about 55 basis points of margin. An analyst did the arithmetic: 10.1% less 0.55 is 9.6%, so is that the 2027 starting point?

Well, we are not quite ready to provide 2027 guidance yet... some of it is through execution. Right? So there is still some execution benefits that we saw during the third quarter. So I would not call execution benefits, you know, normalized margin.

That is the honest answer, and it is also the tell. The company's own keyword history shows execution benefits ranking near the top of this quarter's themes — the language of a windfall being described as a capability. Prior calls show the same pattern: in the November 2025 Q&A, Thiede called the margin upside “really hard to forecast” — Jeff Thiede, CEO · 2025-11-05, and by February he was crediting “exceptional margin upside in 2025” — Jeff Thiede, Chief Executive Officer · 2026-02-25 across four different markets. Each year the upside is real; each year the base rate is the question.

What management quietly stopped selling

The loudest new theme is acquisition-driven: offsite construction and the acquisitions of Epsilon dominated this call, with Epsilon described as 100% offsite and a source of cross-sell into automotive and healthcare. There is a genuine wrinkle here: modular work converts more slowly. “The modular activity that Epsilon has typically has a longer backlog burn than we see in our business.” — Maximillian J. Marcy, CFO · 2026-11-04 That cuts both ways — a longer tail of revenue visibility, but a slower conversion than the backlog burn rhythm of roughly 80% in twelve months that management has reciting for five straight quarters.

What faded is more interesting. new geography — the Mountain States semiconductor foothold that management spent 2025 and the first half of 2026 promoting — spiked hard as a company theme and then fell further than any other Everus keyword this quarter. On the August call it got a full paragraph (“we've talked about our additional location satellite office that we have in the Mountain States. That project is ramping.” — Jeff Thiede, Chief Executive Officer · 2026-08-05); on this call it is a subordinate clause. The organic growth engine narrative has been swapped for an M&A narrative, and the market has not paid for either.

Meanwhile the long-term targets — 5–7% organic revenue growth, 7–9% EBITDA CAGR — went unchallenged this quarter, having been called "way too conservative" in August, when Marcy replied that “this can be a cyclical market, and we're looking at what we can deliver over the long term here.” — Maximillian Marcy, Chief Financial Officer · 2026-08-05 The cycle he was referring to appears to have arrived in the share price before it arrived in the order book.

The bottom line

Everus is executing at a level that its balance sheet and its history do not fully explain: 40% organic-plus-acquired growth, margin expansion, sub-1x leverage, a $500 million-plus capital deployment year. None of that is in dispute. What is in dispute is duration. The backlog is up 86%, but 80% of it burns within a year, and the end market driving the increment is the one the tape is aggressively de-rating. February's 2027 guide — and whether data center bookings extend beyond the planning stages management keeps referencing — is the only thing that resolves it. Until then, this is a well-run company priced as a momentum short.