SPX Technologies

Phase A — Understand the business

Lens 1 · Company Overview

SPX Technologies is a Charlotte, NC-based diversified industrial that designs, manufactures, installs and services highly-engineered infrastructure equipment through two reportable segments: HVAC and Detection & Measurement (D&M). ~4,700 employees, operations in 16 countries, sales in 100+ countries, 462 patents. It is the successor to Legacy SPX Corporation (founded 1912 in Muskegon, MI as Piston Ring Co.; NYSE-listed since 1972), reorganized as a holding company in 2022 after the 2015 spin-off of SPX FLOW and the 2022 sale of its asbestos-liability subsidiaries (the "Canvas" sale — an under-appreciated de-risking that removed all consolidated asbestos liabilities, uncapped indemnity from the buyer).

How it makes money (FY2025, total revenue $2,265.1M):

• HVAC — $1,518.2M (67% of revenue):
• Package & process cooling + engineered air movement/handling — $932.9M. This is the crown jewel: cooling towers and fluid coolers (Marley, Recold, SGS, Cincinnati Fan, Air Enterprises, TAMCO, Ingénia) sold to industrial markets "including data center and power generation." Sold direct, via independent reps, and distributors.
• Hydronic + electric heating & ventilation — $585.3M. A roll-up of ~18 brands (Berko, Qmark, Weil-McLain, Patterson-Kelley, INDEECO, Williamson-Thermoflo, Sigma, Omega, Thermolec…) serving residential/commercial/industrial/institutional.
• Detection & Measurement — $746.9M (33%):
• Underground locators, pipe inspection & rehab, robotics — $255.9M (Radiodetection, Pearpoint, CUES, ULC Robotics, Sensors & Software).
• Communication technologies, aids-to-navigation, transportation systems — $491.0M (TCI, ECS, KTS; Flash Technology, Sealite, Sabik; Genfare transit fare systems).

Customers / concentration: No customer or common-control group >10% of consolidated revenue in any period presented. End-customers: utilities, telcos, defense agencies, municipalities/transit, hyperscalers, industrial OEMs, power generators. This is genuinely diversified demand — a strength relative to single-end-market peers.

Contract structure / payment terms: Mostly point-in-time recognition — $2,026.2M of $2,265.1M (89%) recognized at a point in time, only $238.9M (11%) over time. HVAC equipment ships in 1–3 months, terms 15–60 days; larger cooling-tower construction projects run 3–9 months and recognize over time (percentage-of-cost). This is a book-and-ship industrial, not a long-cycle backlog-burner — backlog converts fast (HVAC ~83% of YE2025 backlog recognized within 2026). No take-or-pay; modest recurring (services <10% of either segment, though D&M is adding SaaS via Genfare transit and ULC).

Verdict on the model: A classic "good business bought well, repeatedly" — niche #1/#2 positions in unglamorous engineered-equipment categories, decentralized operating brands, bolted together by an M&A-led holding company. The model's quality is rising as the mix shifts toward the cooling franchise riding the AI/data-center capex wave.

Lens 2 · Supply Chain

Map: raw materials → SPX manufacturing → channel → end customer.

• Upstream inputs (named in filings): petroleum-based products (FRP/plastics for cooling fill & casings), aluminum, steel, copper — the core commodities for heat-exchange and air-movement equipment. SPX states it is "not significantly dependent on any one or a limited number of suppliers," uses numerous sources given product/geographic dispersion, and has historically passed commodity inflation through pricing. Occasional long-term supplier contracts increase pricing exposure. This is a deliberately de-risked, multi-source input base — no single-source chokepoint disclosed, which is unusual and favorable.
• Internal: Manufactures most components in-house; selectively outsources sub-assemblies "where strategically and economically beneficial." Tangible long-lived assets concentrated in the US ($419.4M) and Canada ($88.2M) — the M&A has added Canadian capacity (Ingénia, Thermolec, Sigma & Omega all Canadian air-handling/heating). Capacity is the binding constraint, not supply — SPX is mid-build on a ~$700M multi-year capacity-expansion program (cooling + the new Tennessee dampener facility) that doesn't reach full production until 2028, with Q1-2026 HVAC margin dinged by "start-up costs and inefficiencies" from that build.
• Channel: Direct to customer + independent manufacturing reps + third-party distributors + (for heating) retailers.
• End customers / buyers: Hyperscale data-center operators (the swing buyer — one hyperscaler has locked multi-year OlympusMAX demand commitments ), power generators, industrial process plants, utilities, telcos, transit authorities, defense (via KTS), municipalities.

Chokepoints: The real chokepoint is SPX's own manufacturing capacity for data-center cooling, not an upstream input. Tariff exposure (2025 US tariffs + retaliation) is flagged as a watch item but "did not have a direct material impact" in FY2025 and is managed via pricing + sourcing actions; ~80% of revenue is US-recorded, limiting import dependence. Names-or-it-didn't-happen check: passed (specific brands, commodities, the hyperscaler-commitment fact, the Tennessee facility).

Lens 3 · Competitive Advantages (moats)

Cooling (the franchise that matters): SPX's Marley/Recold sits in the top tier of the global cooling-tower oligopoly — the top 5 players (Baltimore Aircoil/BAC, SPX/Marley, EVAPCO, Johnson Controls, Hamon) hold ~48–52% of the global cooling-tower market. Crucially, BAC and EVAPCO are private — SPX is effectively the only listed pure-play way to own large-scale evaporative/dry cooling, which is part of why it has become the market's data-center-cooling proxy.

Durable moats:

• Engineering + spec-in + installed base. Cooling towers and fluid coolers are engineered-to-spec, performance-warranted, and physically integrated into a plant or data center — switching mid-design is costly. The new Marley OlympusMAX modular dry/adiabatic fluid cooler (launched Apr 29 2026) with a factory-or-field bolt-on adiabatic module is a genuine product-differentiation moat for hyperscale thermal predictability.
• Brand + decades of reference installations. Marley is a ~century-old brand; reference base and reliability matter in a category where downtime is catastrophic for a data center.
• Scale + capacity as a moat right now. With hyperscale cooling demand outrunning supply, the firm that can build capacity fastest wins share — SPX's $700M capacity program is itself a competitive weapon (and a barrier: a new entrant cannot conjure this).
• Margin proof of pricing power: HVAC segment margin expanded 20.9% → 23.7% → 24.5% (2023→24→25) — operating leverage + favorable mix + pricing, not just volume.

D&M: Moats are narrower but real — Radiodetection (utility locating) and CUES (pipe inspection) have strong niche positions; Genfare (transit fare) and aids-to-navigation are relationship/regulatory-driven. KTS adds defense tactical-networking IP. D&M margin 19.2% → 22.1% → 23.6% (2023→24→25) — also expanding.

Bargaining power: Strong over suppliers (multi-sourced commodities, pass-through pricing). Over customers: strong in cooling right now (demand > supply; hyperscalers committing multi-year), more balanced in heating/D&M. Net: a wide-enough moat in cooling, narrower-but-defensible elsewhere — a B+ moat overall, trending up with mix.

Lens 4 · Segments

All figures (FY) and (Q1).

By segment — revenue, segment income, margin, backlog:

Growth decomposition (FY25 vs FY24): HVAC +11.2% (organic +6.1, acquisitions +5.1, FX ~0); D&M +20.6% (organic +6.3, acquisitions +13.8 from KTS). Both segments are growing organically mid-single-digit and inorganically via the roll-up — the model is working on both axes.

Latest quarter (Q1-2026): HVAC rev $394.0M (+22.0%: organic +9.6%, acq +11.5%, FX +0.9), seg income $88.6M, margin 22.5% (down from 22.9% on capacity start-up costs). D&M rev $172.8M (+8.3%: organic +3.0%, acq +3.9%), seg income $46.7M, margin 27.0% (up from 22.9% on mix + higher-margin transit SaaS).

By geography (FY25): US $1,812.6M (80%), Canada $193.4M (9%), China $71.7M (3%), UK $95.5M (4%), Other $91.9M. ~80% domestic, ~3% China — minimal direct geopolitical/China-decoupling exposure, a genuine differentiator vs. globalized industrials.

Trend read: HVAC is accelerating (organic 6.1% FY25 → 9.6% Q1-26, explicitly "data center demand"); D&M is steady mid-single-digit organic with a sharp margin step-up. The backlog tells the cleanest story: HVAC backlog $451.3M (Q1-25) → $755.3M (Q1-26), +67% — even stripping ~$130M of acquired backlog (Crawford/Thermolec/Sigma), organic backlog $625M vs $451M (+38%). Demand is real and visible, not a forecast.

---

Phase B — Measure performance

Lens 5 · Earnings Result (Q1-2026, the latest print)

Headline (GAAP, three months ended March 28 2026):

• Revenue $566.8M, +17.4% YoY (organic +7.4, acq +9.0, FX +1.0).
• Gross profit $230.6M, gross margin 40.7% (vs 40.6% — stable at a healthy level).
• Operating income $87.7M, +31.7% (op margin 15.5% vs 13.8% — strong leverage).
• Continuing-ops diluted EPS $1.27 (+15.5% from $1.10); net income $59.9M.
• Interest expense fell to $8.4M (from $12.3M) — the Aug-2025 equity raise paid down the revolver.

Non-GAAP / Street (FY-managed metrics): Adjusted EPS $1.69 vs $1.56 consensus (+8% beat); adjusted EBITDA $126.1M, +22.9%. The GAAP-to-adjusted gap is large and structural — driven by intangible amortization ($23.3M in the quarter, $87–91M/yr) and acquisition/integration costs. This is the single most important number-quality caveat for this name (see Lens 10).

What drove it: HVAC organic +9.6% on data-center cooling demand + capacity ramp; D&M margin +410bps on transit SaaS mix. Guidance was raised, not just reaffirmed.

Guidance (raised with Q1): FY2026 revenue $2.575–$2.645B (+~15% midpoint), adjusted EPS $7.75–$8.15 (+~18%), adjusted EBITDA $600–$625M (+~21%) — up from the Feb-2026 initial guide of $2.535–$2.605B rev / $7.60–$8.00 EPS / $590–$620M EBITDA. The raise was explicitly attributed to data-center cooling demand. Data-center revenue target lifted from $300M to $350M (vs $200M in FY25, ~$150M FY24), ~12% of total sales.

Balance-sheet flags: Inventory built to $342.4M (from $302.2M) — deliberate, to support the $755M backlog; contract liabilities/down-payments rising (a good sign — customers pre-funding). No negative working-capital surprise. Net debt remains minimal (see Lens 10).

Market reaction: Stock rose on the print; sits up ~56% over the trailing year. The market is rewarding the data-center narrative and the guidance raise.

Unusual vs. own history: The HVAC margin dip (22.5% vs 22.9%) is the one "negative" — but it's a quality problem (start-up costs from building ahead of demand), and management framed it as transitory leverage timing. Discontinued-ops loss of $4.5M = the Crawford non-core (aerospace/defense/marine) divestiture (see Lens 9).

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on the research-layer shelf (transcripts=0) — this lens is ``-grounded from call summaries.

• Trajectory of tone (FY25 → Q1-26): Management raised FY2025 adjusted-EPS guidance four times through 2025 ($6.00–$6.25 initial → $6.35–$6.65 Q2 → $6.65–$6.80 Q3 → delivered $6.76, +21%). That cadence of serial raises is the clearest sentiment signal there is — management has been consistently conservative-then-beats.
• What management is focused on: (1) Data-center cooling — the headline; they raised the DC growth outlook intra-year from 50% to 70% and the DC revenue target from $300M to $350M; (2) capacity expansion (the $700M program, Tennessee facility) as the gating item; (3) disciplined M&A (Thermolec, Crawford, with immediate divestiture of Crawford's non-core); (4) margin expansion toward a ~25% EBITDA-margin structural target by 2028.
• Recurring phrases / what's new: "OlympusMAX," "data center," "capacity," "record backlog." What they've added is explicit hyperscaler-commitment language (multi-year demand locked) — a step-up in confidence vs. a year ago when data-center was a smaller, less-specified bucket.
• Sentiment read: Unambiguously positive and rising, but now bordering on consensus-bullish — the risk is that the call narrative and the Street narrative have fully converged, leaving less room for upside surprise on the multiple (see Lens 12/13).

Lens 7 · Comps

Peer set = listed industrials with HVAC/data-center-cooling/thermal or roll-up DNA.

Read: SPXC's ~28x forward P/E and ~24.6x EV/EBITDA put it mid-pack — cheaper than AAON (~47–66x) and Comfort Systems (~41x), roughly in line with nVent (~28x) and Modine. It is not the cheap data-center-cooling option, but it is not the most expensive either. The bull argument is that its margin-expansion path (22.4% → ~25% EBITDA by 2028 ) and the pure-play cooling exposure justify a premium-to-history multiple; the bear argument is that mid-20s EBITDA for a ~15% grower already prices a lot of the data-center story (Lens 12/13). Two closest physical peers (BAC, EVAPCO) are private — there is genuine scarcity value in SPXC as the listed cooling-tower vehicle, which supports a structurally higher multiple than a generic industrial.

Lens 8 · Stock-Price Catalysts (what moves the stock)

Mostly ``; the pattern over the last ~2–3 years (the relevant window — the company's character changed post-SPX FLOW spin and post-asbestos sale):

• The dominant catalyst is the data-center cooling narrative + guidance raises. Stock +~56–60% over the trailing year, driven by: each quarter's adjusted-EPS beat-and-raise; the intra-2025 lift of the data-center revenue target ($150M→$200M→$300M→$350M); the OlympusMAX launch (Apr 2026) and the hyperscaler multi-year commitment; the $755M record HVAC backlog print.
• M&A announcements move it secondarily — KTS ($340M, Jan 2025), Sigma & Omega, Thermolec ($140M), Crawford ($299M, Feb 2026 — "reset the 2026 growth floor" per ). The market rewards SPX's M&A because the track record of accretive integration is strong.
• The Aug-2025 equity raise (3.059M shares at $188, ~$551M net) was a notable event — dilutive but de-levering; the stock absorbed it because it funded the growth/M&A engine.
• Macro/rate sensitivity is present but secondary — as a US-centric, short-cycle book-and-ship industrial with pricing power, it tracks industrial-capex and AI-capex sentiment more than rates.

What the pattern reveals: The market now reacts primarily to data-center cooling data points (revenue target, backlog, hyperscaler wins) and the beat-and-raise cadence — i.e., SPXC trades as an AI-infrastructure derivative with an industrial floor. The risk symmetry: a single soft data-center print or a backlog stall would hit it harder than a generic industrial, because the multiple embeds the narrative.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO Gene Lowe — President & CEO since September 2015 (~10-year tenure), joined SPX 2008; ex-Bain & Company, Lazard Technology Partners, Milliken. Critically, he ran SPX's Evaporative Cooling business before the top job — the very franchise now driving the data-center thesis. He is the architect of the post-2015 transformation: took a sprawling, asbestos-encumbered industrial conglomerate, spun SPX FLOW (2015), sold the asbestos liabilities (2022), and rebuilt SPX into a focused two-segment HVAC + D&M compounder.
• Track record (quantified): Revenue $1,741M (2023) → $2,265M (2025), +30% in two years; operating income $221.9M → $350.4M, +58%; HVAC margin 20.9% → 24.5%; adjusted EPS +21% in 2025 alone; serial guidance beat-and-raise. This is a genuinely strong operating + capital-allocation record, not financial engineering — margins and organic growth both improved.
• Capital-allocation history: The model is disciplined M&A + organic capacity investment, no dividend. ~$1.3B deployed on acquisitions 2023–early-2026 (ASPEQ $421.5M, Ingénia $292M, KTS $340M, Sigma & Omega $143M, Thermolec $140M, Crawford $299M), funded by debt then de-levered with the 2025 equity raise. The Crawford deal is the tell on quality: bought the air-handling crown jewel, immediately carved out and sold the non-core aerospace/defense/marine units for $60M within ~7 weeks of closing — buy-what-you-want, flip-the-rest discipline that most roll-ups lack. ROE ~13.5% FY25 is optically modest but understated by the mid-year equity raise and the heavy intangible/goodwill base.
• Skin in the game: Insider ownership figures n/a (no insider-transactions.csv on the shelf; would need a proxy pull). Long-term incentive comp design was made more aggressive in 2024 (max payout 150%→200% of target) — a mild governance watch-item (richer upside for execs), though tied to performance shares.
• Red flags: Few and minor. (1) An external research note (Paragon Intel) has flagged "execution failures and leverage" risk and questioned whether Lowe can pivot from a high-leverage M&A playbook to debt reduction — but the Aug-2025 equity raise already de-levered the balance sheet, blunting that critique. (2) A planned D&M segment leadership transition announced June 19 2026 — routine succession, worth monitoring but not alarming. (3) Richer LTI payout cap. No related-party deals, no promotional behavior, clean accounting history.
• Archetype: Professional manager / operator-acquirer (not founder). For this stage — a sub-$3B-revenue compounder scaling a data-center franchise — this is the right archetype: disciplined, repeatable, integration-focused. The key-man risk is real (Lowe is the strategy and the cooling pedigree), but a 10-year bench has been built.

Lens 10 · Forensic Red Flags

Acting as a forensic analyst. All figures `` unless noted.

Income statement:

• The GAAP-vs-non-GAAP gap is the #1 thing to watch. Adjusted EPS ($6.76 FY25) sits well above GAAP continuing-ops diluted EPS ($5.06) — the bridge is dominated by intangible amortization ($91.3M FY25, guided ~$105M FY26) plus acquisition/integration costs ($28.9M) and the KTS retention-comp amortization ($24.2M). This is legitimate for a serial acquirer (amortization of acquired customer relationships is a real non-cash charge), but it means the company looks materially more profitable on the metric management guides than on GAAP — the gap will persist and widen as the roll-up continues. Not fraud; a quality-of-earnings discount.
• SBC is modest and well-disclosed (long-term incentive comp $16.7M FY25) — not a flattering-the-non-GAAP issue.
• A $23.0M gain on the Filtran equity security flowed through "Other income" in FY25 (driven by Parker-Hannifin / Donaldson acquiring the underlying businesses) — a real, cash-backed mark, but a non-operating, non-recurring boost worth normalizing out.

Balance sheet:

• Intangibles + goodwill = ~$1.91B at YE25, rising to ~$2.30B post-Crawford/Thermolec (Q1-26) — that is >60% of total assets ($3.6B). This is the structural signature of a roll-up: a large, soft asset base that must be supported by the acquired businesses' forecasts. Management explicitly flags KTS ($104.4M goodwill), Sigma & Omega, ASPEQ trademarks ($51.5M), and ULC as fair-value-approximates-carrying — i.e., little impairment cushion; a forecast miss at any of these triggers an impairment charge. There is also $511M of historical accumulated goodwill impairment (legacy SPX) on the books — evidence the company has impaired before. This is the single biggest accounting risk: not today's cash flows, but the durability of the acquired-business forecasts underpinning $2.3B of intangibles.
• Receivables: AR net $357.2M, contract assets jumped to $65.0M (from $11.3M) — but the increase is explained (Q4 project progress at cooling/aids-to-nav businesses where billing milestones hit Q1-26) and contract liabilities (customer down-payments) also rose, so receivables are not outrunning revenue in a worrying way. DSO is reasonable for the mix.
• Inventory $302.2M → $342.4M (Q1-26) — building for the $755M backlog; raw materials are the bulk ($235.5M). Deliberate, not a demand-air-pocket signal.
• Company-owned life insurance (COLI) — $60.3M cash-surrender value in "Other assets," borrowable against. A legitimate but slightly idiosyncratic asset/liquidity source; cleanly disclosed, not a red flag, just unusual.

Cash flow:

• Cash conversion is solid. Operating cash flow (continuing) $335.6M FY25 vs net income $244.0M — OCF > net income (the right direction; D&A and working-capital discipline more than offset). FCF (continuing) ≈ $243.5M. Capex tripled ($38.0M → $92.1M) on the capacity build — a use of cash that is investing for growth, not a leak. Interest paid $41.7M, taxes $57.3M — both reconcile cleanly.
• No evidence of earnings outrunning cash; no aggressive capitalization (capitalized software trivial at $7.2M; R&D fully expensed at $54.0M).

Leases / related parties / contingencies: Operating leases small ($75.8M liability). No material related-party transactions. Contingencies: legacy environmental + the DBT/Kusile South Africa power-project discontinued op — but this was fully settled in 2023 (Settlement Agreement with MHI, mutual release) and is now an immaterial wind-down ($14.1M residual liabilities). The asbestos liabilities were sold off entirely in 2022. The big legacy tail risks have been actively cleaned up — a positive.

Leverage: Total debt $501.6M (term loan $499.1M; revolver $0, trade-receivables facility $0) vs cash $364.0M → net debt ~$137.6M. That is roughly 0.3x EBITDA — essentially de-levered after the Aug-2025 equity raise. Weighted-avg rate 5.1%; term loan matures 2030; covenant max leverage 3.75x (4.25x post-acquisition) with $1,489.5M revolver capacity undrawn. Enormous dry powder for the roll-up to continue — the balance sheet is a strength, not a risk.

Regulatory findings (required sub-section):

• SEC Litigation Releases / AAERs: None. "No LR found" and "No AAER found" for SPX Technologies, 2021-06-29 to 2026-06-29, per SEC EDGAR EFTS.
• Non-SEC enforcement (FTC/DOJ/FDA/CFPB): No material enforcement hits surfaced in web review for SPX Technologies as an industrial parent. (Note: this is `` negative-evidence, not an exhaustive clearance.)
• 10-K Item 3 (Legal Proceedings): SPX's own disclosure — "We are also subject to legal proceedings and claims that arise in the normal course of business. We believe these matters are either without merit or of a kind that should not have a material effect individually or in the aggregate". Standard ordinary-course language; no material litigation disclosed.
• Conclusion: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K Item 3 as of 2026-06-29.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 / FY2027 / FY2028)

Built bottom-up from the latest actuals + management guidance. All outputs `` with arithmetic shown; no forecast.ts logged (watchlist/unattended). Note SPXC guides on adjusted EPS (the metric the Street uses); I anchor there and flag the GAAP gap.

Anchors: FY2025 adjusted EPS $6.76 actual; FY2026 guidance $7.75–$8.15 (mid ~$7.95) on revenue $2.575–$2.645B / adj EBITDA $600–$625M. Diluted share count ~50.5M post-raise.

Arithmetic (base FY27): FY26 $7.95 × 1.15 rev growth × ~1.005 margin-mix uplift ÷ ~1.005 dilution ≈ ~$9.10. (base FY28): $9.10 × 1.15 × margin uplift ≈ ~$10.40. These are deliberately tied to the disclosed margin-expansion arc and the company's demonstrated ~15% total growth, not heroic assumptions.

GAAP caveat: GAAP continuing-ops diluted EPS will run ~$1.50–$2.00/yr below these adjusted figures (intangible amort ~$105M FY26 ≈ ~$1.60/sh after-tax) — so a "GAAP P/E" looks far richer (~45x TTM) than the "adjusted P/E" (~28x fwd). Hold both in view.

Forecast to track (not logged): Base case → SPXC FY26 adjusted EPS ≥ $7.75 (the bottom of guidance) resolves favorably with high probability (~85%) given Q1 beat-and-raise; the more falsifiable bet is FY26 adjusted EPS ≥ $8.15 (top of guide), p ≈ 0.40.

Lens 12 · Bull vs Bear

Bull case. SPXC is the listed pure-play on hyperscale data-center cooling — its two closest physical competitors (BAC, EVAPCO) are private, so it carries scarcity value as the tradeable vehicle. The cooling franchise (Marley/Recold/OlympusMAX) is riding a multi-year AI-capex thermal-rejection wave: data-center revenue ~$150M (2024) → $350M (2026E) → with the $700M capacity program not fully online until 2028, the runway extends years past current guidance. HVAC backlog of $755M (+67% YoY) is visible, ordered demand, not a forecast. Underneath the narrative is a genuinely high-quality compounder: both segments expanding margins (HVAC 20.9%→24.5%, D&M 19.2%→27.0% in Q1), a beat-and-raise management team with a 10-year track record, a de-levered balance sheet (~0.3x EBITDA) with ~$1.5B of undrawn revolver to keep rolling up niche leaders, ~80% US revenue (decoupling-proof), and disciplined M&A (the Crawford buy-and-flip is the proof of taste). Structural EBITDA-margin expansion to ~25% by 2028 compounds the EPS growth. At ~28x forward / ~24.6x EBITDA it is cheaper than AAON and Comfort Systems for arguably better mix.

Bear case (2–3 things that could permanently impair or de-rate):

1. Data-center capex digestion. The entire re-rating rests on hyperscale thermal demand. If AI-capex enters a digestion phase (one or two hyperscalers pause), the cooling order book stalls — and because the multiple embeds the narrative, the de-rate is violent. This is a cyclical-priced-as-secular risk.
2. The multiple has done the work. Stock +56% in a year; much of the $700M capacity program's revenue is already in guidance, leaving "limited re-rating potential until 2027/2028 revenue becomes visible". Mid-20s EBITDA for a ~15% grower is not cheap on an absolute basis.
3. Roll-up / intangible fragility. ~$2.3B of goodwill+intangibles (>60% of assets) carried at fair-value-approximates-carrying — no impairment cushion on KTS, Sigma & Omega, ASPEQ. A few acquisitions underperforming their deal models = a non-cash impairment + a credibility hit to the M&A engine that justifies the premium.

Pre-mortem (18 months out, thesis broke — what happened?): Most likely path: a hyperscaler capex air-pocket in H2-2026/2027 froze cooling orders just as SPX's new capacity came online, turning the start-up costs into stranded fixed cost; HVAC margin compressed; the beat-and-raise streak ended; the multiple compressed from ~28x to ~20x forward, and the stock fell 30–40% despite still-positive EPS — a classic "great company, priced for perfection, demand blinked" outcome. A secondary path: a botched/over-priced acquisition forces an impairment and breaks the M&A halo.

Are multiples too high? Not egregiously — but fully fair, not cheap. The premium-to-history is defensible if data-center demand compounds as guided; it is unforgiving if growth merely normalizes.

Contrarian view (what the market is refusing to see): The market is treating SPXC as an AI-derivative and obsessing over the data-center revenue line — but the under-discussed story is the D&M segment's margin transformation (19.2% → 27.0% in Q1) driven by transit SaaS (Genfare) and defense (KTS), a higher-quality recurring-ish mix that the cooling narrative completely overshadows. If D&M re-rates as a software-ish industrial rather than a project shop, there's value the consensus isn't modeling. Conversely, the market may be under-pricing the cyclicality risk in cooling precisely because it has fallen in love with the secular framing.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• Where revenue is concentrated and what breaks it: ~12% of sales now ride data-center cooling, and that 12% is the entire reason the multiple expanded ~2x. The other 88% is good-but-pedestrian short-cycle industrial growing mid-single-digit. If you strip the data-center premium, you're paying a data-center multiple for a low-double-digit industrial grower. A single soft hyperscaler-capex quarter re-rates the whole thing.
• Why the moat may be weaker than bulls think: Cooling towers are engineered but not proprietary in the way bulls imply — BAC and EVAPCO (both private, both well-capitalized) are equally capable of pouring capacity into the data-center wave, and hyperscalers multi-source deliberately. SPX's "scarcity as the listed pure-play" is a stock-market moat, not a business moat. OlympusMAX is months old; the "$50M first-year bookings" is small relative to a $12.3B market cap.
• Most dangerous competitor bulls underestimate: Baltimore Aircoil — private, focused, the historical #1 in evaporative cooling, and not constrained by quarterly EPS optics; it can undercut on price or out-invest on capacity without a public shareholder watching margins.
• Worst capital-allocation / accounting concerns: The relentless M&A means earnings quality is permanently flattered by non-GAAP — adjusted EPS is ~33% above GAAP, and that gap grows with every deal. ~$2.3B of intangibles with no impairment cushion is a slow-burning landmine; the company has impaired $511M of goodwill before. The LTI payout cap was raised to 200% in 2024 — richer comp into a rising tape.
• Assumptions that must hold for today's price: (1) data-center cooling compounds >40% multi-year; (2) the $700M capacity build comes online on time/on budget and is absorbed, not stranded; (3) the beat-and-raise cadence continues; (4) no acquisition blows up; (5) margins reach ~25% EBITDA. All five must roughly hold.
• If growth disappoints 20–30%: FY27 adj EPS drops toward ~$8.10 (bear) vs ~$9.10 (base), and — more damaging — the multiple compresses from ~28x toward ~18–20x as the secular framing collapses to cyclical. Combined, that's a plausible 35–45% drawdown from a ~$240 price.
• Single scenario that permanently impairs: A structural shift in data-center cooling architecture (e.g., a rapid industry move to direct-to-chip liquid cooling / immersion that displaces large evaporative tower demand at hyperscale faster than SPX can pivot) would strand the $700M capacity build and impair the thesis. Plausibility: moderate — liquid cooling is rising, but evaporative/hybrid rejection remains essential for heat rejection to atmosphere even in liquid-cooled facilities, so this is a real-but-not-imminent risk, not a 2026 event.

Lens 14 · Management Questions (ordered by information value)

1. Of the $350M FY26 data-center revenue target, how much is contracted/committed (e.g., the multi-year hyperscaler commitment) vs. pipeline — and how many distinct hyperscaler customers does it represent?
2. As liquid/direct-to-chip cooling scales at hyperscale, what share of a liquid-cooled facility's thermal budget still requires your evaporative/dry heat-rejection equipment — and how does OlympusMAX position you if the architecture shifts?
3. The $700M capacity program reaches full production in 2028 — what utilization and incremental margin do you underwrite on it, and what is the downside if data-center order intake slows before it's absorbed?
4. With ~$2.3B of goodwill+intangibles at fair-value-approximates-carrying, which acquired units are closest to an impairment trigger, and what forecast assumptions would have to break?
5. Post the 2025 equity raise you're near-de-levered — what leverage are you willing to run back up to for M&A, and how does the pipeline look at current private-market multiples (HVAC deals at ~11x EV/EBITDA)?
6. D&M margin hit 27% in Q1 on transit-SaaS mix — is that a structural step-up or project-timing, and how large can the recurring/software revenue base in D&M become?
7. What is the realistic organic growth rate of the non-data-center HVAC business (heating + the rest of cooling) absent the AI tailwind?
8. How much of HVAC's record $755M backlog is data-center vs. power-generation vs. general industrial, and what's the margin differential across those?
9. On the Crawford buy-and-flip — what is your standing framework for which acquired businesses you keep vs. divest, and should we expect more of that?
10. How are 2025–2026 tariffs flowing through cost and pricing, and what's the net margin impact you're absorbing vs. passing through?
11. What is the succession/key-person plan given how central your evaporative-cooling pedigree is to the data-center strategy — and what does the June-2026 D&M leadership transition signal?
12. Why no dividend or buyback at this stage, and at what point does capital return enter the allocation framework?
13. What's the right normalized GAAP-to-adjusted-EPS bridge investors should model as the roll-up continues — does the amortization gap keep widening?
14. What would cause you to slow the capacity build — what leading indicators of hyperscale digestion are you watching?
15. Where could a well-capitalized private competitor (BAC, EVAPCO) take share from you over the next three years, and what's your defense?

---