Enerflex Books a Billion, Then Goes Looking for Engines
Record Engineered Systems bookings and a 1.75-gigawatt behind-the-meter backlog, against a dividend that still underwhelms and a capacity ceiling that doesn't lift until 2028.
EFXT · Earnings Call · 2026-10-29
A billion-dollar booking quarter, and a very quiet tape
Enerflex closed Q3 2026 with the kind of number that usually resets a story: “record bookings of approximately $1 billion” — Paul E. Mahoney, President and CEO · 2026-10-29 in Engineered Systems. Nine-month bookings of roughly $2 billion put the company on track to clear $3 billion for the year — against a prior annual high-water mark of about $1.5 billion set in 2018. Backlog sits at $2.1 billion, giving visibility into 2028. Free cash flow was $102 million versus $43 million a year ago, helped by a $51 million working-capital recovery and $62 million of proceeds from the Asia-Pacific divestiture that closed September 30. Net debt of $312 million against $227 million of cash took bank-adjusted leverage to roughly 0.6x.
And yet the gross margin line is the tell. Gross margin before D&A came in at $176 million, or 28% of revenue, against $206 million and 27% a year earlier — the prior-year period carrying an upfront gain from the DSACC expansion commencement and some Engineered Systems milestone recognition. Sequentially, 30% fell to 28%. The explanation was familiar: “primarily, it is mix. Project sequencing” — Preet Dhindsa, SVP and CFO · 2026-10-29. Return on capital employed of 15.8% was down from 16.9% a year ago. So: record demand on one side of the ledger, flat-to-down unit economics on the other, and a capacity build that management was candid will not show up in the numbers until 2028.
What is genuinely new: the data center order book
Last quarter, Paul Mahoney was explicit that the record bookings number excluded this theme entirely: “Q2 does not have data center elements in it” — Paul E. Mahoney · 2026-08-06. In Q3 it did. The company booked a contract to design, engineer, fabricate and assemble roughly 450 megawatts of behind-the-meter natural gas-fired power generation for a North American data center developer, taking the behind-the-meter power backlog past 1.75 gigawatts. That is a real step-change in the booked number, and it is company-unique: a compression and gas-processing fabricator repositioning itself as an integrated dispatchable-power contractor, with the promise of commissioning, O&M and installed base service revenue riding behind each unit, wrapped up in the ReliCore asset-performance platform.
What is not new is the funnel language. Mahoney has been running the same script for several quarters — 500 megawatts becoming "well over 1 gigawatt," then “we have reported many quarters now about our growing C in the market, which is at 7 gigawatts” — Paul E. Mahoney · 2026-08-06, and now the 7-gigawatt funnel has been deliberately winnowed: “we started to zero this down into 2 gigawatts of seriousness” — Paul E. Mahoney, President and CEO · 2026-10-29. The filtering criteria are investment-grade counterparties and regulatory line of sight, and it is a defensible discipline. But the honest read is that the "new" market is three years old as a conversation and one quarter old as revenue-bearing backlog.
I see probably an additional 1.5 to 2 in a very filtered manner, to the description that I laid out there. The gating item, obviously, is going to be equipment supply.
The constraint has migrated — from procurement to plant
For most of the last two years, the binding constraint in this business was lead times on large reciprocating engines — a top-ranked topic for the company in 2025, with management quantifying purchase obligations out to 2029 and warning that certain configurations stretched into 2028. That concern has now largely been internalised and, notably, has stopped dominating the call. What replaced it is the physical plant. Enerflex is spending $30–40 million in 2026 and $85 million in 2027 on Engineered Systems facilities, funded partly by cutting maintenance capex guidance to roughly $55 million from a prior $70–80 million and pushing PP&E/infrastructure to $30–40 million from $15 million. Mahoney was concrete about what that buys: “it is incremental acreage. it is incremental buildings” — Paul E. Mahoney, President and CEO · 2026-10-29.
When Tim Monachello of ATB pushed for levers to accelerate capacity sooner, the answer was candid and slightly deflating: “Nothing concrete at this point, but certainly, how do we accelerate with partners” — Paul E. Mahoney, President and CEO · 2026-10-29. And the margin payoff is explicitly deferred: “those margins are accretive to our current reported historical ES margins, but that is not going to hit until likely late 2027, call it Q4 into 2028” — Preet Dhindsa, SVP and CFO · 2026-10-29. Note too the recurring "historical average" refrain — Preet Dhindsa has been guiding to it since at least late 2025 (“we've been guiding towards historical Enerflex averages for a couple of quarters now” — Preet Dhindsa, CFO · 2025-11-06), and Q3 is the quarter it finally landed. The mix story is credible; the timing is not a surprise to anyone who has been listening.
The market is rotating beneath the story
Here is where the global tape gets interesting in a way that cuts against the narrative. Over the last 90 days the AI-power complex has been a source of losses, not gains: data center capacity is among the sharpest decliners, traditional data center names are down alongside it, and the great AI data center basket is negative across 49 of 56 constituents. Meanwhile the molecule trade has been working: Bcf per day advancers are positive across 15 tickers, natural gas production is firm, and Gulf Coast infrastructure names have held up. In other words, the market's preference has shifted from the power-hungry buildout story toward the demand for natural gas infrastructure underneath it — which is, arguably, Enerflex's actual franchise, and the reason the behind-the-meter pitch is coherent rather than opportunistic. That is a genuine strategic alignment. It is also a warning that the market may pay for the gas molecule before it pays for the megawatt.
The argument management keeps not winning
For the second time this year, an analyst pressed on capital returns. Aaron MacNeil of TD Cowen noted that a 17% dividend increase — the third consecutive raise, to $0.0305 per share quarterly — is large in percentage terms but immaterial in cash terms. On a ~$2.97 billion market cap that is roughly $0.12 annualised: a gesture. Dhindsa's response was fluent but familiar: “there is no reluctance, but there is a lot of assessment in different levers at different points in time” — Preet Dhindsa, SVP and CFO · 2026-10-29. The same list — growth capex, tuck-in M&A, debt paydown, buybacks, dividends — has been recited since the February call, when management promised to be "more prescriptive" once the strategy work concluded. The strategy work concluded in May. Capital allocation is still, by management's own framing, undecided.
There is a defensible logic: executing a $2.1 billion backlog with a 1.75-gigawatt power book requires working capital and facility spend, and preserving a 0.6x leverage profile while doing it is prudent. But it does mean the equity story is entirely a 2028 earnings story funded by today's cash flow, with almost nothing returned in the interim. Record bookings, record backlog, and a token dividend — that is the shape of this report. The next twelve months will be judged on whether the facility expansion converts, because the order book has already done its job.