Legence: The Best Quarter It Ever Printed, Sold Off 52%
Revenue +85%, backlog +92%, a first-ever buyback — and a stock down 52% in ninety days. The gap is guidance duration, a margin guide that steps down, and a data-center tape that stopped paying for backlog.
LGN · Earnings Call · 2026-11-12
The beat the tape refused to bank
On paper, Legence's Q3 2026 was the best quarter in the company's short public life. Revenue of $1.310 billion rose 85% year-over-year, adjusted EBITDA of $177 million roughly doubled, adjusted EBITDA margin of 13.5% expanded about 100 basis points, and backlog growth pushed consolidated backlog and awards to $5.9 billion, up 92% with a 1.2x quarterly book-to-bill. Jeff Sprau opened with the tally almost defiantly: record revenue, record backlog, record EBITDA, with just under half the growth organic.
The tape has rendered a very different verdict. Legence shares are down 52.5% over the last ninety trading days and sit 65.8% below the May 6 peak of $102.27 — a de-rating so complete that Craig Irwin of ROTH felt obliged to name it on the call:
So sometimes it's best to talk about the elephant in the room, right, one way to move it off to the side. And your stock is trading at a multiple where there's clearly some skepticism about the backlog.
That is the whole story in one sentence. Legence is not being asked whether the quarter happened; it is being asked whether the backlog means what management says it means. Nothing in this report resolves that — but the transcript contains the answer to why the market flinched, and it is not the quarter.
The guidance is the tell, not the quarter
Read the two numbers side by side. Backlog and awards grew 92%. The 2027 revenue guide is $5.3–5.6 billion against a roughly $4.9 billion implied 2026 (Q1 $1.038B, Q2 $1.262B, Q3 $1.310B, plus a Q4 guide of $1.26–1.30 billion) — call it 8% to 14% growth. Management was explicit about the reconciliation, and about the duration that causes it:
“As our project sizes, our average project sizes have increased over the last year. There is a bit of elongation on that backlog... you, of course, have some of that backlog burning well into 2028 and even some into 2029.” — Steve Hansen, Chief Operating Officer · 2026-11-12
The backlog is real; it is just long. And a project that starts in 2028 does not defend a multiple today. This is the same elongation the company flagged last quarter — “we are seeing our backlog stretch into further out periods than we had historically because they're aware, too, right, that they need the resources to get their builds completed” — Steve Hansen, Chief Operating Officer · 2026-08-13 — so the pattern is now two quarters old and the market has stopped rewarding it.
The margin arithmetic is the sharper tell. Q3 delivered 13.5%. The Q4 guide of $160–170 million on $1.26–1.30 billion implies roughly 12.7%–13.1%; the 2027 EBITDA midpoint of $710 million on revenue of about $5.45 billion implies roughly 13.0%. Guidance embeds a margin below the quarter just reported. Steve Hansen conceded why: “in this quarter, again, we had exceptional project execution, which we don't model that level of execution and going forward” — Steve Hansen, Chief Operating Officer · 2026-11-12. Investors have heard "exceptional execution" before — it topped this company's own keyword list in 2026Q2 as Strong project execution — and are now discounting it as non-recurring.
Three things that are genuinely new
The freshest item is the buyback program, the company's first, authorized the same morning it announced a deal. That juxtaposition is the strategy in miniature: Stephen Butz framed it as another tool in the capital-allocation kit, "enabling us to opportunistically repurchase shares when we believe doing so represents an attractive use of capital relative to other opportunities." A first buyback at a 52% drawdown is management making a statement with the only language a beaten-up multiple understands. Note also that pro forma net leverage is down to 1.2x — about half the year-ago level — after three notches of ratings upgrades, so the balance sheet can fund it.
Second is acquisition of CTI, a Central Texas mechanical contractor for roughly $100 million ($90 million cash plus ~260,000 shares). Less than 10% of Legence's data center work sits in Texas today; CTI buys unionized process-piping labor and a semiconductor-sector customer book in Austin. Sprau's rationale — "we've wanted to expand our presence in Texas for quite some time" — is backed by a genuinely new emphasis on semiconductor fabs, which now ranks among the company's top keywords for the first time. Semiconductors grew over 50% last quarter from a small base; they are being positioned as the second leg to data centers.
Third, and least comfortable, is Loudoun County — a brand-new keyword tied to a brand-new analyst question about post-midterm, state-level data center moratoriums. Steve Hansen's answer was categorical: “We've got a substantial amount of work there, about 40% in that area, and we have not seen any of the projects that we're doing there, pause or delay in any way.” — Steve Hansen, Chief Operating Officer · 2026-11-12 Sprau added that post-election “conversations with our data center customers post election indicate a continuation of their plans to build” — Jeffrey Sprau, Chief Executive Officer · 2026-11-12. Forty percent concentration in one Virginia cluster, in a quarter when Virginia politics entered the question flow, is precisely the kind of disclosure that gets capitalized into a 65% drawdown.
What quietly left the story — and what the stock price is doing to GAAP
What management stopped emphasizing matters. Life science and health care, once a top-five company keyword, have fallen off; Hansen admitted the backlog build there is weaker — large hospital projects "haven't replaced those yet." Education delivered softer revenue but strong program-and-project-management wins, and the education end market is now the offset story rather than the growth story. Sustainability consulting has simply vanished from the discussion.
Meanwhile, the collapsing share price is mechanically flattering the reported P&L, via profit interest marks — the single highest-momentum keyword in this print. The legacy profit interest units swung from a $3.8 million expense to an $8.9 million benefit in gross profit and a ~$33 million gain inside SG&A, because the mark-to-market liability falls with the stock. The GAAP picture underneath remains thin: gross margin of 17.4%, down 4.0 points year-over-year, and operating income of -$23 million. The adjusted numbers are the right lens here — but the gap between adjusted and GAAP is itself part of why the multiple compressed.
The tape versus the order book
Legence's own price action matches a sector-wide unwind rather than a company-specific failure. The global theme tape shows the entire AI-infrastructure complex de-rating over thirty days: AI data centers is the worst thirty-day cluster on the board, and data center capacity is down nearly 30% on a ninety-day view. Legence sells fabrication and craft labor, not GPUs — but it trades as a derivative of that sentiment anyway.
Which is why the retrofit argument deserves more attention than it got. Asked what happens when the new build cycle rolls off, Hansen pointed out Legence has been building data centers since the late 1990s and is refurbishing them today. Butz made the identical point a year ago — “we're bullish about the retrofit market for data centers. There's a lot of data centers that are probably getting long in the tooth” — Stephen Butz, Chief Financial Officer · 2025-11-17 — and the incumbency loop Sprau describes (design, install, maintain, break-fix) is the durable part of the model. It is also the part nobody is paying for in a drawdown.
Sprau and Hansen spent much of the Q&A insisting the business is flexible — “We've always wanted this growth to be an and versus an or” — Jeffrey Sprau, Chief Executive Officer · 2026-08-13, from last quarter, recurred almost verbatim this quarter. It is a genuine operating philosophy and also, at this point, a slogan. The more useful close came from Hansen: “we can't control the narrative or sentiment or stock price for that matter. What we can control is taking care of our customers.” — Steve Hansen, Chief Operating Officer · 2026-11-12
The controllable just posted its best quarter. The uncontrollable marked it down by half. Until Loudoun headlines clear and the 2028–2029 backlog starts converting into a growth rate closer to 92% than 10%, that standoff is the investment case — and the buyback is management's first real bet on its own side of it.