SPX Bet on the Ramp — and the Ramp Sent a Bill

A 20% EPS quarter, a trimmed high end, and a $25 million self-inflicted margin hole — what a -32% drawdown looks like from inside the growth story

SPXC · Earnings Call · 2026-10-29

The bill for the ramp

SPX Technologies delivered what on the surface looks like another clean quarter: revenue up 20.5%, adjusted EBITDA up 21.7% to $165.7 million, adjusted EPS of $2.21 for a fifth consecutive quarter of 20%-plus growth. HVAC revenue rose a record 35.5% (22.3% organic), and segment backlog jumped 69% organically year-over-year and 15% sequentially to roughly $1.1 billion — the first time it has crossed a billion. The problem was underneath: HVAC segment margin fell about 140 basis points, and management cut the top end of its full-year EPS range to $8.50, for a $8.35 midpoint. Mark Carano put a number on it: about $25 million of second-half cost, the majority of it — call it $15 million — "growth-related inefficiencies," with the balance price/cost. “The majority of it, or more of it, is really related to these growth-related inefficiencies. Think about $15 million, and the balance, it's around price cost.” — Mark Carano, Vice President and Chief Financial Officer · 2026-10-29 Call it what it is: the company grew so fast at its data center cooling plants that it hired several hundred people at once, missed on productivity and turnover, and had to buy metal fabrication capacity from secondary suppliers at premium prices. Gene Lowe was unusually direct about the miss: “We did have higher turnover in the quarter. We also were off on productivity. So there were some variances there in our expectations as we've grown.” — Gene Lowe, President and Chief Executive Officer · 2026-10-29 And Carano on the outsourcing: “As we go to secondary suppliers with lead times that are appropriate for meeting this demand, just given the environment we are in, it is coming in at a higher cost than I think we would have originally thought.” — Mark Carano, Vice President and Chief Financial Officer · 2026-10-29 Management frames it as an intentional decision rather than a leak — "we made a decision, it was an intentional decision, to accelerate our hiring and secure additional suppliers to ensure that we can meet the demand that is out there." Perhaps. But it is still a variance against a guide that was set eleven weeks earlier, and it lands on a stock that had already been de-rated hard.

The tension with their own tape

What makes this quarter more than an ordinary cost blip is the contrast with what this management team said about supply chain for three straight calls. In February: “Looking across our supply chain, nothing jumps out as a material concern today.” — Mark A. Carano, Chief Financial Officer · 2026-02-24 In April, asked directly about data center supply chain delays: “Not for us.” — Eugene Lowe, President and Chief Executive Officer · 2026-04-30 And in July, the most confident version of all:

Before we take on a large order we actually have a very strong supply chain team. That will scrub every bill of material item and validate that we believe we can fulfill those items. So we are not flying blind.
Eugene Joseph Lowe · 2026-07-30

They were not flying blind — but the constraint turned out to be internal, not external: their own cut-and-punch steel lines, their own people. Note that when a competitor's heat-rejection peer flagged a coil shortage, SPX said it had none, and is expanding coil manufacture inside Madison. The bill of materials is theirs; the throughput wasn't. That is a different failure mode than the one investors were pricing, and it explains why the Q4 guide carries it forward rather than resolving it.

What the tape knew first

SPXC has fallen 29.5% over the last 90 trading days and sits 32% below its June 22 peak of $246.41. That is partly idiosyncratic and partly a downdraft in the whole complex: across the market's own theme tape, AI data center names are down 5.3% over 90 days with 49 of 56 tickers negative, and data center capacity is down 8.0% with seven of nine lower — despite both baskets still being strongly positive over 360 days. The rotation is fresh, not stale. Tellingly, SPXC's own keyword set shows where attention moved. Data center capacity was the company's single top keyword in the prior quarter; this quarter it fell out of the top thirty altogether. In its place: Input costs, Price cost, and steel. The narrative rotated from how much capacity to what the capacity costs — and the keyword capacity expansion registered as one of the quarter's sharpest declines. The inflation leg is genuinely sector-wide (Carlisle's input cost headwind, Sherwin-Williams' price cost balance), but the labor-ramp leg is SPX's own. The valuation gives the reaction its force. Price to revenue stands at 5.0x, up roughly 470% across eight years, against an operating margin of 16.9%. And the plant build is visible in the capex line: capital expenditure of $21 million in the latest reported quarter, up 174% year-over-year, or about 3% of revenue. None of that is alarming given interest coverage of 13.1x.

The part that isn't broken

Detection & Measurement told the opposite story: revenue down 7.7% on a tough prior-year comparison, but margin up 260 basis points, helped by mix and a one-time commercial resolution. Carano effectively retired the Investor Day 22-24% medium-term target: “That business has done a very nice job, I think, resetting the structural margin profile of D&M. I expect we'll update that perhaps as we roll out the guide for next year.” — Mark Carano, Vice President and Chief Financial Officer · 2026-10-29 Strip out ~175 basis points of this year's one-offs and the structural floor is higher than the published framework — a genuine positive that got buried under the HVAC noise. And on pricing, Lowe flatly rejected the industry worry that became its own keyword this quarter, pushback on pricing: “We haven't seen any pushback on pricing.” — Gene Lowe, President and Chief Executive Officer · 2026-10-29 Data center revenue should land modestly above the $430 million plan, and the $1.1 billion capacity target now has an upward bias. The most interesting new wrinkle is the leased facility model — Springfield pushes a couple hundred million of revenue through a facility that took a couple million of capital: “That is a leased facility that we added a very small amount of capital, a couple million dollars, and are pushing a couple hundred million of revenue through.” — Gene Lowe, President and Chief Executive Officer · 2026-10-29 That is an option on going past $1.1 billion without another Madison-sized cheque.

What to watch

The Q4 guide implies roughly 150 basis points of sequential margin improvement, two-thirds of it operating leverage, with the inefficiencies still bleeding through. The 2027 setup — "towards the high end of our medium-term target range" — depends on whether the new labour gets productive, whether Madison insources the outsourced fab, and whether price/cost turns net positive. If it does, $25 million reads as an investment. If it doesn't, it reads as the first crack in a de-rating that has already taken 32% off the equity. The market has voted once; management now has two quarters to prove the vote wrong.