Carrier Global Corporation (NYSE: CARR) — The Volume Came Back, The Pricing Power Didn’t

Date: August 15, 2026 · Follow-up update to the June 14, 2026 report (see “Changes since June 14, 2026” immediately after Claude’s Take) Price reference: $62.78 (close 2026-08-14); market cap ~$52B; enterprise value ~$63B; ~824M shares out Fiscal year: December · Sector: Industrials — Building Products & Equipment (Intelligent Climate & Energy Solutions) · CIK: 0001783180

Standing disclaimer (applies to sections 1–15, the analytical body of this article): The body of this report takes no investment recommendation and sets no price target. It discusses valuation only as embedded expectations and scenario analysis. The single, deliberate exception is the Claude's Take block immediately below, which is fenced off as a subjective view.

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information only and is not investment advice. The analytical body that follows (sections 1–15) carries no recommendation and no price target.

Verdict: HOLD — and I am downgrading the business, not the price. (The call is nominally unchanged from June 14; the reasoning underneath it has changed materially, and not in Carrier’s favour.) The stock did exactly what the prior report said it would — it fell ~10% from $69.91 to $62.78, and ~17% from the June 25 peak of $75.73 — so the “avoid-adding at $70” half was right. But it fell for a reason I did not have in June, and that reason is worse than the one I did have. Constructive accumulation zone ~$48–54 (lowered from $50–55): roughly 17–19× a tax- and amortization-honest ~$2.75–2.90 of normalized earning power, i.e. where you stop paying a compounder multiple for a business earning a 6% honest ROIC. Fair-to-fully valued ~$60–70 (lowered from $65–75) — a band I have set deliberately between my two methods, which disagree: the scenario work puts the base case at $65–75, while a peer-multiple sum-of-the-parts reaches only $45.7–58.7B of enterprise value against a market $62.7B. The gap between those two numbers is, almost exactly, the value of the Viessmann question (see the sum-of-the-parts discussion under Valuation), and I am not going to paper over it. At $62.78 the stock sits in the lower half of my band and is genuinely no longer expensive — but it is not cheap either, and the thing I would be buying is a worse business than I thought I was analysing eight weeks ago. Conviction: medium, unchanged.

Here is the change. In June the bear case was European: Carrier overpaid $14.2B for a low-margin heat-pump business at a subsidy-inflated cycle top. That bear case is now half-falsified and replaced by something larger. European heat pumps recovered — EHPA units +13% in 2025 with Germany +50%, Q1-2026 +17% — and Carrier’s CSE segment still earns 7.2%, is running “pricing promotions,” has just replaced its president, and is launching a cheaper value-tier product against Chinese OEMs growing 13–50% in Europe. Volume returned; price did not. That converts Europe from a bad cycle into a bad industry. And then Q2 revealed the same disease in the crown jewel: CSA took +$120M of sales and lost $56M of operating profit — a -47% incremental margin — while Carrier announced an 8% list price increase, guided 6–7% realization, and delivered 3%. Gitlin explained why on the call: “we want to at least maintain share. So that’s kind of a balancing act.” A business with genuine customer captivity does not face that trade. Four of five HVAC majors lost margin in the same quarter, including Trane on a record backlog — so this is not Carrier failing uniquely, which is precisely the problem: the 2022–25 margin expansion I read as dealer-channel pricing power was a shared, transient, regulation-enabled episode (SEER2, A2L, the R-410A pre-buy), not a barrier to entry. After-tax ROIC of 7.05% against Trane’s 25.1% and Lennox’s 29.1% is what that looks like in the financials, and on Greenwald’s own bands, 6–8% means advantages are absent.

So why HOLD and not AVOID? Because the price has already done real work, and three things genuinely cut the other way. The de-rate was multiple compression, not a numbers cut — no broker downgraded, Bernstein raised to $78, estimates went up while the price went down. The margin hit has a benign and self-liquidating mechanism: Section 232 on HVAC was cut 25%→15% effective June 8, Carrier had pre-committed to hand price back, and it did so with high-cost LIFO inventory still in COGS — a timing artifact with a stated 2027 sunset, and Carrier fully mitigated the 2025 tariff round. And management’s capital discipline at the margins is real: 1H buybacks at $62.33 were struck near today’s price, with Carrier deliberately standing down when the stock hit $72. Against that: the guide is a Q4 hockey stick (implied Q4 adjusted operating profit of $829–929M against $455M last year, +82% to +104%), the “adjusted” EPS is itself tax-flattered (1H adjusted ETR 15.1% vs 22.1%, worth $0.12–0.16), capex guidance was raised 53% while FCF guidance was not, 37.98M Viessmann shares (~$2.4B) remain freely saleable with a demonstrated clearing price of $62.01, and no insider has bought a share since the collapse despite an open window. This is a fairly-priced average business with a cyclical recovery in front of it and a structural returns problem underneath it — not a compounder on sale. Bullish trigger: Q3 delivers the ~16.5% adjusted operating margin and CSA price realization re-accelerates toward the announced 8%, proving the price give-back was tariff timing rather than share defence. Bearish trigger: the CSE goodwill fails its July-1 annual test (result lands in the Q3 10-Q, ~late October), or Q4 misses the required ~$829M+ and the FY2026 guide is cut — confirming the raise was a hockey stick.

Tag: “The volume came back; the pricing power didn’t.”

Changes since June 14, 2026 (this is a follow-up update)

This report is a nine-week follow-up to the June 14, 2026 report (itself a six-day update to June 8). Per update discipline, durable analysis — industry structure, the segment architecture, the Viessmann deal history, the financial scaffolding — is carried forward and re-verified; this subsection isolates what moved. Bottom line: Q2-FY2026 (reported 2026-07-28) delivered the volume inflection the prior report was waiting for, and the market sold it anyway. The stock fell 13.6% over two sessions on a raised guide. The prior report’s price call was vindicated; its business-quality verdict was not, and is downgraded here.

What moved (FACT):

1. Price: $69.91 (2026-06-12) → $62.78 (2026-08-14), -10.2%. The stock first rallied to an interim high of $75.73 on 2026-06-25, then fell -8.9% on 2026-07-28 (the Q2 print) on 16.36M shares (~2.5× the 90-day average) and a further -5.1% on 2026-07-29 — a -13.6% two-day decline on a raised full-year guide. Peak-to-trough -20.9%. Market cap fell to ~$52B, EV to ~$63B.
2. Q2 delivered the volume half of the inflection and failed the margin half. Organic sales +3% — the first positive organic quarter after two years of flat-to-negative — with CSA residential +9% and light commercial +10%. But every one of the four segment margins fell year-over-year, adjusted operating margin fell 190bp to 17.2%, and adjusted EPS fell 7% to $0.86.
3. Guidance was RAISED, not cut: FY2026 to ~$23B sales, ~$3.5B adjusted operating profit, ~$2.90 adjusted EPS, and the data-center revenue target from ~$1.5B to ~$2.0B (exit rate $2.5B). Orders +~40%, commercial HVAC +~65%, data centers +>300%, backlog >$8B.
4. The mechanism behind the margin miss is a tariff cut, not a tariff rise. Section 232 on HVAC equipment was cut 25%→15% (proclaimed 2026-06-01, effective 06-08, sunsetting 2027-12-31). Carrier had pre-committed to its channel that it would hand price back — and did, while high-cost inventory was still flowing through COGS. Realized residential price was ~3% against an announced 8% list increase and 6–7% guided.
5. Two disposals closed and one acquisition was made. Riello sold to Ariston, closed 2026-07-01 for ~€289M EV, with a $46M impairment — sold below carrying value. NORESCO closed 2026-08-03 (the 10-Q still described it as pending). 75F, a building-management-software bolt-on, closed 2026-07-23, price undisclosed.
6. CSE leadership was replaced. On 2026-06-17 — three days after the prior report — Thomas Heim, who had run the Viessmann integration for nine years, was replaced as CSE president by Thomas Donato, an external hire (ex-Bosch Power Tools CEO). No 8-K was filed.
7. The momentum/factor regime flipped hard. Six-month relative strength went from +32.0% to -5.74%; every risk-adjusted horizon inside a year is now negative. Twelve-month: CARR -3.1% vs SPY +21.7%, TT +13.1%, JCI +48.2%.
8. Own-history valuation percentiles fell: composite from ~72nd to 60.4th, P/S from ~90th to 74.9th. The stock genuinely de-rated against its own history.

What was falsified, confirmed, or newly discovered (INTERPRETATION unless marked):

• The prior report’s moat verdict is DOWNGRADED from “narrow but genuine — demand-captivity-plus-scale in the NA dealer channel plus the transport-refrigeration duopoly” to weak/narrow. The June-2026 quarter is a natural experiment: four of five HVAC majors failed to pass through an input-cost move in the same quarter. What the prior report called dealer-channel pricing power was a shared, regulation-and-cycle-enabled episode, not a barrier to entry. Full argument in the Competitive Position section.
• The European bear pillar is half-falsified — and replaced by a worse one. European heat-pump volumes recovered strongly (see Industry Dynamics); CSE margins did not. Europe is now a bad industry finding rather than a bad cycle finding.
• NEW: an antitrust overhang not in the prior report. Carrier is a named defendant in consolidated HVAC price-fixing class actions (E.D. Mich.), with amended consolidated complaints filed 2026-07-10; treble damages, class period from 2020-01-01, nothing accrued, Viessmann’s US subsidiary separately named (FACT, Q2-FY2026 10-Q). See Risk Analysis.
• NEW: the first arm’s-length mark on European thermal assets. Ariston disclosed the Riello price Carrier did not: ~0.72× sales / 8.3× EBITDA, against the ~3.0× sales / ~17.1× unsynergized EBITDA Carrier paid for VCS. See Capital Allocation.
• Two prior-report errors corrected. (i) Riello was not a Viessmann asset — UTC acquired ~70% of it in 2016; it sits in CSE only because CSE is the Europe segment. (ii) The “FCF -20%” reading of the 1H cash flow statement is wrong: Carrier’s reported FCF is computed off total operating cash flow including discontinued operations, contrary to its own published definition. On continuing operations 1H FCF rose 22% ($742M vs $608M). See Financial Quality.
• The CSE goodwill impairment has NOT occurred and remains live. The $234M goodwill decline was 99% foreign-currency translation; Riello carried only $3M of CSE goodwill. The $7,586M CSE block is fully intact and untested since 2025-07-01. The FY2026 annual test date is July 1, 2026, so the result lands in the Q3-FY2026 10-Q, expected late October 2026 — a dated, identifiable catalyst.
• The call is unchanged at HOLD, with a lowered valuation zone ($48–54 accumulate, $60–70 fair, vs $50–55 / $65–75). The price fell about as much as the business-quality assessment did; the two roughly cancel.

📈 Stock Price Action — Five-Year Event Map

Over the trailing sixty months Carrier has traded a full round trip and then some: from ~$54 in August 2021 down to a ~$32.15 low (2022-10-20), up ~150% to an all-time high of $80.43 (2024-10-14), then through two separate ~35% drawdowns to a 52-week low of $49.79 (2025-11-20), a ~52% recovery to $75.73 (2026-06-25), and a ~21% post-earnings give-back to the last close of $62.78 (2026-08-14). The 52-week range is $49.79 – $75.73; the stock sits ~17.1% below its 52-week high and ~21.9% below its all-time high. Five years of ownership have produced a ~+3.7% annualized total return for ~32% annualized volatility (FactorsToday, 2026-08-15). Price moves below are FACT; the attributed causes are INTERPRETATION.

1. From the August-2021 high of ~$54.33 the stock lost ~41% into October 2022 — the Fed’s tightening cycle, the European energy shock that hit the future Viessmann end-market, and the unwinding of the pandemic HVAC demand pull-forward. 2. The October-2022 low began a ~79% rally to $57.41 (2023-08-01), interrupted on 2023-04-24 by a -7.3% session as the Viessmann talks leaked; Carrier confirmed on 2023-04-25 both the acquisition and its intention to exit Fire & Security and Commercial Refrigeration. The market initially disliked paying up for a European heat-pump asset, then re-rated the “climate pure-play” story before giving back ~22% as the 10-year yield approached 5%. 3. The strongest phase of the five years: Viessmann closed 2024-01-02, ~$10B of divestiture proceeds funded buybacks, and the US A2L refrigerant transition pulled residential demand forward. The stock made its all-time closing high of $80.43 on 2024-10-14. 4. Nine days later the Q3-2024 print cut -8.8% out of the stock as the European heat-pump market collapsed and guidance came down; the April-2025 tariff announcement completed a -32% drawdown. 5. The tariff pause on 2025-04-09 (+9.4%) and a strong Q1-2025 print on 2025-05-01 (+11.6%) produced a ~46% round trip back to within ~1% of the all-time high. 6. The Q2-2025 print broke it: -10.6% in a session on residential destocking and a cut guide, then a four-month grind to the $49.79 low. The CEO bought in the open market near that low — the only insider open-market purchase in five years. 7. From the November low the stock rallied ~52% over seven months on the Q4-2025 and Q1-2026 prints and data-center order momentum. The path was not clean: a -19% air-pocket in March 2026 and an April whipsaw. 8. Q2-2026 beat and raised the full-year guide — and the stock fell 8.9% that session, then a further 5.1% the next as Lennox’s own print (-21%) repriced the entire residential-HVAC complex. The two days are analytically distinct and the distinction matters (see Variant Perception): on day one the S&P was up 0.24% and the HVAC peer mean was -1.84%, so Carrier’s -8.9% was ~7pp of idiosyncratic decline — the market’s independent verdict on Carrier’s own quarter. On day two Carrier’s -5.1% outperformed a peer mean of -8.84%. The stock has since recovered ~4.8% on volume roughly half the 90-day average.

1. Executive Summary

Nine weeks ago the prior report described Carrier as “a good business mid-renovation, priced as if the renovation is already done,” and concluded that the stock was fully valued at ~$70 pending a residential inflection that had not arrived. The inflection arrived. The stock fell anyway. Understanding why is the whole content of this update.

Q2-FY2026, reported 2026-07-28, was on its face the quarter the bull case ordered: organic sales +3% (the first positive print in two years), CSA residential +9% and light commercial +10%, orders +~40% with commercial HVAC +~65% and data centers +>300%, backlog above $8B, and a raised full-year guide — ~$23B sales, ~$3.5B adjusted operating profit, ~$2.90 adjusted EPS, with the data-center revenue target lifted from ~$1.5B to ~$2.0B. The market’s response was to sell the stock -8.9%, then -5.1% the following session: a -13.6% two-day decline on a raised guide, and a -20.9% fall from the June 25 peak of $75.73 to $59.92.

The market was right, and the reason is a pricing failure, not a demand failure. Every one of Carrier’s four segment margins fell year-over-year — CSA 27.0%→24.4%, CSE 7.9%→7.2%, CSAME 15.3%→11.8%, CST 17.6%→16.0% — and adjusted operating margin fell 190bp. The arithmetic in the crown jewel is the single most important number in this report: CSA grew sales by $120M and lost $56M of operating profit, a -47% incremental margin. Consolidated, Carrier took +$238M of revenue and earned $71M less adjusted operating profit doing it. Management announced an 8% residential list-price increase, guided 6–7% realization, and delivered ~3% — and the CEO explained why on the call: “We’ve actually been doing very well on share… we want to at least maintain share. So that’s kind of a balancing act.” A firm with genuine customer captivity does not have to make that choice.

The competitive read is the material change from the prior report, and it is a downgrade. The tempting hypothesis — that Carrier specifically fumbled while better operators held price — does not survive contact with the peer data. Four of the five HVAC majors lost margin in the June-2026 quarter: Carrier -190bp adjusted, Trane -203bp gross (its first decline in six years, on a record +70% backlog and 123% book-to-bill), Lennox ~-222bp operating ex a one-time tariff refund with its CEO “surprised by the price competition,” and Daikin -0.8pt. Only Johnson Controls expanded (+243bp) — and JCI sold its residential business to Bosch. That simultaneity is worse for Carrier than a company-specific miss would have been. When four oligopolists all fail to pass through an input-cost move in the same quarter, the binding constraint is competitive, not cyclical. In Greenwald’s terms, the 2022–25 margin expansion the prior report treated as proof of dealer-channel captivity was a shared, transient, regulation-and-cycle-enabled phenomenon (SEER2, the A2L transition, the 2024 R-410A pre-buy, post-COVID inflation pass-through) — the cooperative arm of a prisoner’s dilemma, not a barrier to entry. The financial signature is unambiguous: after-tax ROIC of 7.05% (FY2025), against Trane’s 25.1% and Lennox’s 29.1%. Greenwald’s own bands read 15–25%+ as advantages present and 6–8% as advantages absent.

Two mechanisms genuinely mitigate the margin story, and this report gives them full weight. First, the proximate cause was a tariff cut, not a hike: Section 232 on HVAC equipment was reduced 25%→15% (proclaimed 2026-06-01, effective 06-08, sunsetting 2027-12-31); Carrier had publicly pre-committed to hand price back if it fell, and did so while higher-cost inventory was still flowing through COGS under LIFO. That is a timing artifact with a stated expiry, and Carrier has form — the 10-Q states it “fully mitigated the 2025 impact of tariffs implemented in 2025.” Second, Carrier alone among the majors booked no IEEPA tariff refund while Lennox recognized $30M and Daikin ~¥10.5B, a pure accounting-timing distortion that made an identical set of facts look worse at Carrier and leaves an unrecognized contingent asset. Neither mitigation, however, touches the structural finding: Goris conceded that even fully-recovered tariff pricing is “slightly a margin-dilutive impact” — dollar-neutral pass-through compresses margin percentage permanently.

The raised guide is a Q4 hockey stick. 1H-FY2026 adjusted operating profit was $1,689M, down 16% year-over-year at a 14.4% margin. The ~$3.5B full-year target therefore requires ~$1,811M in 2H (+42% YoY) at a ~16.0% margin, against 12.3% in 2H-2025 and 15.5% in 2H-2024. Working management’s own Q3 guide (~$5.95B revenue, 16.5% margin, $0.75 EPS) through the arithmetic isolates the problem in the fourth quarter: Q4 implied adjusted operating profit of $829–929M against $455M actual last year — +82% to +104% — at a 15.5–17.3% margin versus a post-spin Q4 best of 13.2%. Q4 adjusted EPS implied is $0.72 against $0.34, +112%. Revenue is equally back-loaded: 1H organic was +1%, so the full-year guide needs 2H organic of +10.4% to +14.6%. The comp is genuinely easy — Q4-2025 was the destocking trough — and the >$8B backlog gives unusual visibility. But management’s own residential math undercuts the quality of the growth: of a guided +20% second half, “10 of that will come from the absence of destocking,” leaving mid-single-digit real movement.

Earnings quality deteriorated on a dimension the prior report missed. Adjusted operating profit now sits 56% above GAAP, with add-backs up 26% year-over-year to $606M in 1H. More consequentially, the 1H adjusted effective tax rate was 15.1% against 22.1% — the $99M Swiss valuation-allowance release and an $18M state settlement were not excluded from adjusted EPS, worth $0.12–0.16/share. Tax-normalized, 1H adjusted EPS fell 17–19%, not the reported 9%, and the FY2026 guide is really ~$2.74–2.78. On an amortization-honest basis (charging the ~$850M/yr of acquired-intangible amortization that recurs through 2030), FY2026E EPS is ~$2.10–2.13; charging recurring restructuring as well, ~$1.89–1.92. TTM adjusted EPS of $2.45 is already below FY2025’s $2.59.

The cash and returns picture is mixed, and one prior-report claim is corrected in Carrier’s favour. The headline “1H FCF -20%” is an artifact of Carrier computing reported FCF off total operating cash flow including discontinued operations, contrary to its own definition; on continuing operations 1H FCF rose 22% to $742M, with working capital consuming $381M less cash than a year earlier. That is a genuine positive. Against it: capex guidance was raised ~53% to ~$600M — taking capex/sales from 1.80% to ~2.61% for the new U.S. data-center plant — while FCF guidance was not raised. The asset-light thesis is now structurally qualified. Returns deteriorated outright: GAAP ROIC 5.5% (from ~7.0%), amortization-and-restructuring-honest 6.0%, company-adjusted 9.5%, ROE 8.4%. Even a fully delivered FY2026 guide only lifts honest ROIC to ~8.6% — still short of a ~9% WACC. Net debt/EBITDA rose to 3.53x GAAP (from 2.55x), driven by a -27% TTM EBITDA collapse rather than borrowing. Tangible book is -$9.55/share.

Capital allocation moves from “qualified negative” to negative on the central decision, with real competence at the margins. The prize is a price: Ariston disclosed what Carrier did not — Riello changed hands at ~0.72× sales / 8.3× EBITDA, the first arm’s-length clearing price for a European thermal asset since Carrier paid ~3.0× sales / ~17.1× unsynergized EBITDA for VCS. The 2023 underwriting promised €200M of synergies with “the vast majority by year three” (2026), a European heat-pump market “expected to triple to $15 billion by 2027,” and a return to pre-transaction leverage in ~two years. None of it happened; no quantified synergy scorecard appears in any SEC filing; CSE’s SG&A load is unchanged at ~21% of sales; CSE restructuring totals $251M since close, 62% of all segment restructuring; and the executive who ran the integration for nine years was replaced on 2026-06-17 without an 8-K. Offsetting this: 1H buybacks were struck at $62.33 — essentially today’s price — with Carrier deliberately standing down to 643k shares in June at $72.00 after buying 2.38M in April at $60.52. That is genuine discipline. But distributions are not self-funded — 1H returned $1,210M against $795M of FCF, plugged with $361M of commercial paper — and the 2026 proxy contains zero occurrences of ROIC, ROE, or economic profit, while every paying metric is mechanically increased by a debt-funded, goodwill-heavy acquisition.

Cross-section verdicts. Business: a scaled climate company whose quality is concentrated in a North American residential franchise that just demonstrated it cannot hold price, and whose growth is migrating into structurally lower-margin commercial and data-center work. Competitive position: weak/narrow, downgraded — one genuine duopoly (transport refrigeration, ~11% of sales), a distribution/access advantage in NA residential that wins volume and share but demonstrably not price, nothing defensible in Europe or China, and no moat at all in the segment driving all the growth. Industry: still structurally above-average, but materially less so than in June — the profit pool has moved from pricing to volume, and Carrier’s growth is mix-dilutive into it. Capital allocation: negative on the Viessmann decision, competent on buyback execution, structurally misaligned on incentives. Valuation: the stock has genuinely de-rated — own-history composite from ~72nd to ~60th percentile, P/S from ~90th to ~75th — and is no longer expensive; but with Value zeroed in every factor model and honest ROIC below WACC, it is fair, not cheap. The swing variables are now two: whether Q4 delivers the required ~$829M+, and whether the untested $7,586M CSE goodwill survives its July-1 test when the result surfaces in late October.

2. Business Overview

2.1 What Carrier is — carried forward and re-verified

[FACT] Carrier was spun off from United Technologies on April 3, 2020. Over 2023–2025 it executed a wholesale portfolio transformation: it acquired Viessmann Climate Solutions (VCS) on January 2, 2024 for total consideration of $14.2 billion, and divested its Fire & Security businesses and its Commercial & Industrial Fire and Commercial Refrigeration units for roughly $10B. The result is described by management as a climate “pure-play.” That description is accurate as to scope; the Financial Quality section shows it has still not produced a cleaner or higher-quality earnings stream.

[FACT] Operations are classified into four segments, all climate (Q2-FY2026 10-Q, Note 16): Climate Solutions Americas (CSA), Climate Solutions Europe (CSE), Climate Solutions Asia Pacific, Middle East & Africa (CSAME), and Climate Solutions Transportation (CST). Brands include Carrier, Viessmann, Toshiba, Automated Logic, Bryant, Day & Night, Heil, CIAT, and Carrier Transicold/Sensitech.

2.2 The segment picture as of Q2-FY2026 — where the money actually is

[FACT] Q2 and 1H-FY2026 segment results (Q2-FY2026 10-Q, Note 16 and MD&A Segment Review):

[FACT] On the half: CSA $5,873M at a 20.4% margin (vs 24.9%, -450bp); CSE $2,617M at 7.0% (vs 8.4%); CSAME $1,751M at 10.8% (vs 15.0%); CST $1,451M at 15.1% (vs 16.3%). Total segment margin 15.3% vs 18.8%.

[INTERPRETATION] Three structural facts fall out and govern the rest of the report. First, CSA remains the entire business: ~53% of Q2 sales and ~72% of segment operating profit. Any thesis on Carrier is a thesis on North American residential and light-commercial HVAC. Second, the margin decline is universal, not localized — it is not a European problem, a China problem, or a freight problem; all four segments compressed simultaneously on growing volume. Third, CSE remains structurally broken at 7.0–7.2%, less than a third of CSA’s margin, on a business that cost $14.2B. The prior report’s framing — that Carrier bought the lowest-margin segment in its own portfolio — is confirmed and, if anything, understated: CSE’s margin has fallen further while the European end-market has recovered (see Industry Dynamics).

2.3 Growth is migrating out of the profitable part of the portfolio

[FACT] CSA Q2 organic +4% decomposed: residential +9%, light commercial +10%, commercial -8% (management attributes the commercial decline to “timing of customer deliveries”). CSE: residential/light-commercial up high-single-digits, commercial down mid-single-digits. CSAME: double-digit growth in India, the Middle East, Southeast Asia and Australia, offset by China residential/light-commercial down 13–14%. CST: container +~40%, global truck and trailer down low teens — both company-sourced; I could not independently corroborate the container figure against reefer-container build data.

[FACT] CFO Patrick Goris stated on the Q2-FY2026 call that CSA commercial margins run in the mid-to-high teens against a segment average of 20–21%. Data-center revenue is guided to ~$2.0B in FY2026 (~8.7–9% of sales), of which ~$500M landed in 1H — meaning ~75% must arrive in 2H. Backlog exceeds $8B, ~68% commercial, of which ~40% is data centers (≈$2.2B).

[INTERPRETATION] This is the most under-appreciated fact in the Carrier story and it reframes the growth narrative. The margin pool is residential replacement; the volume pool is commercial and data centers. Every point of mix shift toward the order book management is celebrating is a point of margin dilution to the segment that carries the company. The +65% commercial-HVAC and +>300% data-center order growth is therefore not straightforwardly good news — it is growth that structurally lowers Carrier’s blended margin, in a business where the blended margin is already the problem. Corroboration from the peer set points the same way: AAON cut its full-year gross-margin guidance 200–300bp with ROIC falling 24.2%→9.3% (filing-sourced), and Modine reportedly grew data-center sales +90% with a ~960bp gross-margin decline (secondary/vendor-sourced — not traced to a Modine filing, and weighted accordingly) (see Industry Dynamics).

2.4 The aftermarket claim, re-tested and materially weakened

[FACT] Q2 service sales grew to $717M from $636M (+12.7%) against product sales +2.9%, with double-digit service growth in all four segments. Aftermarket/parts-and-service was 28% of FY2025 net sales.

[FACT — and this is the new finding] Carrier’s own disclosed service gross margin is 24.4%, below its product gross margin of 27.6%. The comparable figures at Trane are 43% service vs 32.6% product.

[INTERPRETATION] The prior report, and the sell-side generally, treat the growing aftermarket as a quality upgrade — a recurring, higher-margin annuity that justifies a premium multiple. Carrier’s own numbers do not support that at this company. At Trane the aftermarket is genuinely accretive: service earns ~10 points more than product, so mix shift raises blended margin. At Carrier service earns ~3 points less than product, so aftermarket growth is mildly dilutive. This does not make the service business worthless — recurring revenue is more stable and strengthens channel relationships — but it removes a load-bearing plank from the quality argument. The genuine, defensible new metric is engagement rather than margin: ~55,000 monitored channel-partner technicians, +35% year-over-year, which is a real distribution asset (see Competitive Position).

Verdict: Carrier is a scaled, focused climate business whose profitability is concentrated in one segment (CSA, ~72% of segment profit) that has just demonstrated it cannot hold price; whose $14.2B European acquisition still earns 7%; and whose growth engine — commercial and data-center — carries structurally lower margins than the business it is growing away from. The portfolio is coherent in scope and deteriorating in quality.

3. Industry Dynamics

The durable structural framing from the June report carries forward: HVAC is a consolidated, oligopolistic, replacement-driven industry with high aftermarket attach and meaningful regulatory tailwinds. What changed in nine weeks is where inside that industry the profit sits. The one-line version: in June the industry was good because it could price; today it is good only because it has volume.

3.1 Tariffs — a regime change, not a rate change

[FACT] The Supreme Court struck down the IEEPA tariffs 6–3 on 2026-02-20, and the Court of International Trade ordered CBP refunds (estimated ~$175B, Penn Wharton). The administration pivoted to more durable authorities: a Section 122 10% stopgap (150 days from 02-24) and Section 232 expanded on 2026-04-02/06 — 50% on HS chapters 72/73/74/76 (steel, iron, copper, aluminium), 25% on derivative products with >15% metal content, now assessed on full customs value rather than metal content alone.

[FACT] Proclamation 11032, effective 2026-06-08 and running through 2027-12-31, cut residential HVAC equipment to 15% and left commercial HVAC unrelieved.

[FACT] Copper reached an all-time COMEX high of $6.77/lb on 2026-08-07, with a refined-copper Section 232 determination still pending. Copper and aluminium are the dominant raw-material inputs in HVAC equipment.

[INTERPRETATION] Three consequences, and the second is the one most commentators have inverted.

First, the cost shock is mostly symmetric. Trane manufactures >95% domestically and still lost 203bp of gross margin. This is an industry event, not a Carrier sourcing failure.

Second — and this inverts the intuition — Carrier’s Q2 margin damage came from a tariff cut, not a tariff increase. Carrier had publicly committed to its channel in April that it would hand price back if Section 232 came down. It came down on June 8. Carrier gave the price back while higher-cost inventory was still working through COGS on LIFO. Gitlin confirmed the sequence verbatim on the call: an announced ~8% list increase, cut back when 232 fell “from 25% to 10–15%,” yielding ~3% realized price in Q2 and ~4% for the year. This is a genuine timing artifact with a stated sunset (2027-12-31), and Carrier has form: the 10-Q states it “fully mitigated the 2025 impact of tariffs implemented in 2025.”

Third, the residual asymmetries run against Carrier. (i) Only ~29% of Carrier’s significant properties are in the U.S. (ii) The relief went to residential; Carrier’s growth is in the unrelieved commercial band. (iii) Most consequentially for like-for-like comparison, Carrier recorded zero IEEPA refund while Lennox booked $30M and Daikin ~¥10.5B — a pure accounting-timing difference that made Carrier’s Q2 look materially worse than peers on identical underlying facts, and that leaves Carrier holding an unrecognized contingent asset not reflected in the guide. Any peer margin comparison that ignores this overstates Carrier’s underperformance.

3.2 U.S. residential — a restock, quite possibly a pre-buy, not yet a recovery

[FACT] AHRI sell-in data for 2026: April +5.1%, May +6.7%, June +22%. But the trailing twelve-month average is -17%; management characterizes the end market as flat at ~7.0–7.5M units; and field inventory is down ~25%. Existing-home sales fell to a 4.05M SAAR and single-family starts declined three consecutive months.

[FACT] Lennox disclosed Section 232-related list pricing effective 2026-07-01.

[INTERPRETATION] Only one story satisfies all four data points simultaneously: distributors are restocking a depleted channel into a flat end market. Sell-in (+22% in June) is running far ahead of sell-through (flat), which is the definitional signature of a restock, and inventory down 25% explains why it can run for several quarters. Worse for the quality of the June number specifically: a sell-in spike in the month immediately before a disclosed July 1 price increase is a pre-buy signature. Carrier’s residential +9% is therefore in line with, not ahead of, the industry, and some portion of it is likely borrowed from Q3. Management’s own guidance concedes the mechanism — of a guided +20% second-half residential number, “10 of that will come from the absence of destocking.” The genuine underlying movement is mid-single-digit.

3.3 Data-center cooling — real demand, late in the capital cycle, dilutive on the way in

[ASSUMPTION — third-party estimates, not a primary source] Published hyperscaler capex estimates for 2026 cluster in the ~$690–725B range (Futurum, CreditSights, FactSet and press compilations differ), with >$1T commonly cited for 2027E. Treat the level as directionally reliable and the precise figure as unverified. Carrier raised its data-center revenue guide from ~$1.5B to ~$2.0B with a $2.5B exit rate, against a market it sizes at ~$21B — roughly 9–10% share. Backlog >$8B, ~68% commercial, ~40% of that data centers (≈$2.2B).

[FACT] Capacity is being added simultaneously across the industry: Carrier’s new Texas/Alabama plant, Trane quadrupling applied capacity, nVent’s third liquid-cooling expansion in three years, AAON capex +101%, a $183M Munters order, Eaton’s $9.55B Boyd acquisition, Schneider committing >$700M, Vertiv at 2.5× capex.

[FACT / SECONDARY] Margin compression is already visible before that capacity lands: AAON cut full-year gross-margin guidance 200–300bp with ROIC falling 24.2%→9.3% (filing-sourced). Modine is reported to have grown data-center sales +90% with a ~960bp gross-margin decline — this figure is secondary/vendor-sourced, not traced to a Modine filing, and is treated as supporting rather than decisive. Goris put Carrier’s CSA commercial margins at mid-to-high teens against a 20–21% segment average.

[INTERPRETATION] Applying the Marathon capital-cycle lens: demand is genuine and still accelerating, but this is a late up-leg, not an early one. High returns have attracted capital from every direction, the capacity is contracted and being built now, and — the tell that matters — margins are compressing while demand is still accelerating. That is what the top of a capital cycle looks like from the inside. For Carrier specifically the implication is sharper than “competition will eventually arrive”: the data-center backlog is dilutive to Carrier’s blended margin on the way in, not just at the end. Every dollar of the celebrated order book converts at mid-to-high teens into a segment averaging 20–21%. Management is underwriting the new U.S. plant on going from ~10% to 15–20% share — an explicit share-gain bet in a market where seven well-capitalized competitors are simultaneously adding capacity.

3.4 Europe — the volume recovered, which makes the margin worse news

[FACT] European heat-pump unit sales (EHPA): 2025 +13% to 2.9M units, with Germany +50% (~299k units, roughly half of all new heating systems); Q1-2026 +17%, Germany +34%. Germany scrapped the 65%-renewable heating mandate (effective ~2026-07-01) but retained the 70% BEG subsidy through 2029.

[FACT] Against that recovering market, Carrier’s CSE segment earned a 7.2% margin in Q2 (7.0% on the half), ran “pricing promotions” (the phrase appears twice in the 10-Q, and only for CSE), and is launching a cheaper Vitocal 200 value tier. NIBE’s Q1 was flat; Daikin Europe +2%. Chinese OEMs are growing 13–50% in Europe.

[INTERPRETATION] This is the most important change in the industry section, and it half-falsifies the prior report’s central bear pillar while replacing it with something worse. The June report argued that Viessmann was bought at a subsidy-and-gas-price cycle top and that European heat-pump demand was structurally deflating. The demand half of that claim is now falsified: the market grew 13% in 2025 and 17% in Q1-2026, Germany is booming, and the subsidy survives to 2029. Carrier is earning 7.2% into a market growing 17–34%, while discounting. A cyclical problem resolves when the cycle turns. This cycle turned, and the margin did not follow — which reclassifies Europe from a bad cycle to a bad industry for Carrier: a structurally fragmented, subsidy-mediated, Chinese-entrant-exposed market in which Viessmann’s brand commands no durable premium. That is a permanent finding, not a timing one, and it is the strongest single piece of evidence bearing on whether the $14.2B was well spent (see Capital Allocation).

3.5 Transport refrigeration and China

[FACT — company-sourced] CST: container volumes +~40%, global truck and trailer down low teens. Not independently corroborated against third-party reefer-container build data. Container is the lower-margin of the two, which is why CST margin fell 160bp on flat organic sales. ACT Research data show tractor demand recovering ahead of trailers, implying a 2027 turn for the trailer cycle. The Carrier Transicold / Thermo King duopoly structure is unchanged.

[FACT] China: domestic air-conditioning shipments -5.6% in 1H-2026; Xiaomi is pricing ~7% below market; industry profits -6%. Carrier’s China residential/light-commercial fell 13–14%. Gitlin has publicly questioned the future of the China RLC business. Chinese overcapacity is exporting into Europe.

[INTERPRETATION] CST is the one segment where the margin decline is honestly explained by mix rather than price — no pricing commentary appears anywhere in its disclosure, which is consistent with a genuine duopoly holding price in a quarter when every contested market did not (see Competitive Position). China is a structural share problem, not a cyclical one, and it is being exported into Carrier’s second-largest segment.

3.6 A new structural overlay: antitrust

[FACT] Carrier is a named defendant in consolidated HVAC price-fixing class actions in the Eastern District of Michigan (Berg v. Bosch, 2:26-cv-10949 and related), with amended consolidated complaints filed 2026-07-10. The complaints name Carrier and six peers said to control >90% of the U.S. market, seek treble damages, and specify a class period beginning 2020-01-01. Viessmann’s U.S. subsidiary is separately named. Carrier has accrued nothing (Q2-FY2026 10-Q, Legal Proceedings).

[INTERPRETATION] The claim is unproven and industry-wide, and the report does not assume liability. But the timing is analytically uncomfortable in a way worth stating plainly: the alleged class period (2020–present) maps almost exactly onto the window in which the industry expanded margins in unison — the same window the Competitive Position section concludes was a shared, regulation-enabled pricing episode rather than a set of individual competitive advantages. Both readings cannot be flattering. Either the coordinated margin expansion was competitive and fragile (this report’s view, argued under Competitive Position), or it was something plaintiffs will argue was not competitive at all. Sized in the Risk Analysis section.

Verdict: HVAC remains a structurally above-average industry — consolidated, replacement-driven, regulation-supported — but it is materially less attractive than the June assessment, and the deterioration is specific rather than general. The profit pool has migrated from pricing to volume; the growth pockets everyone is celebrating (commercial, data centers) carry lower margins than the businesses they are growing away from; European volume recovery has failed to produce European profit; and the industry’s ability to pass through input costs — the single best evidence of structural attractiveness — was tested in June 2026 and failed across four of five majors.

4. Competitive Position

This section contains the material change from the prior report. The moat verdict is downgraded from “narrow but genuine” to “weak/narrow.”

4.1 The natural experiment of June 2026

The June-2026 quarter offered something rare: a simultaneous, exogenous input-cost shock applied to every major HVAC manufacturer at once. That is as close to a controlled experiment on pricing power as public markets provide, and the result is unambiguous.

[FACT] June-2026 quarter, peer margin comparison:

*Lennox reported figures include $30M of one-time IEEPA tariff refunds, all recognized in Q2 with none remaining for 2H. Ex-refund, Lennox Q2 operating margin was ~21.0% vs 23.0% (-222bp), with HCS segment margin ~21.1% vs 23.7% (~-400bp on -12% volume).

[INTERPRETATION — the central finding of this report] Four of the five majors lost margin. The tempting hypothesis — that Carrier fumbled while better operators held price — is refuted. And the refutation is worse for the thesis than confirmation would have been. If one oligopolist fails to price, that is an execution problem at one company. When four fail in the same quarter — including Trane, holding a 123% book-to-bill and a record +70% backlog, and Lennox, whose CEO says he “was surprised by the price competition” — the binding constraint is competitive, not demand-driven.

In Greenwald’s framework the conclusion is direct: the 2022–2025 margin expansion that the prior report read as evidence of dealer-channel customer captivity was a shared, transient, regulation-and-cycle-enabled phenomenon — SEER2, the A2L refrigerant transition, the 2024 R-410A pre-buy, and post-COVID inflation pass-through. It was the cooperative arm of a prisoner’s dilemma, not a barrier to entry. Pricing power that exists only while an enabling regulatory condition lasts is not a moat; it is a windfall with a schedule.

[FACT] Carrier’s relative standing inside that industry-wide failure is middling, not worst. Trane grew organic +9% — three times Carrier’s +3% — while losing roughly half Carrier’s margin. Lennox’s ex-refund incremental margin (-42%) was worse than Carrier’s (-30%). JCI expanded 243bp but has no residential exposure. The prior report’s characterization of Carrier as a strong #2-quality operator ranked below Trane survives the quarter unchanged. What does not survive is the claim that the industry’s structure conferred durable pricing power on any of them.

[INTERPRETATION — resolving an apparent contradiction in the evidence, because it matters] On a reported basis Lennox looks like a counter-example: gross margin +43bp and operating margin +13bp, which would suggest at least one peer did hold the line and that the “everyone had this quarter” defence is unavailable to Carrier. That reading does not survive the one-time items. Lennox’s reported Q2 includes $30M of IEEPA tariff refunds, all recognized in the quarter with none remaining for the second half — while Carrier booked no refund at all (see Industry Dynamics). Ex-refund, Lennox’s operating margin was ~21.0% versus 23.0%, -222bp — worse than Carrier’s -190bp — on organic growth of +3%, identical to Carrier’s. The like-for-like comparison therefore confirms rather than refutes the industry-wide finding.

What the peer set does legitimately show is that the differentiator in the quarter was volume leverage, not pricing. Trane suffered a larger gross-margin hit than Carrier (-203bp vs -173bp) yet held its operating-margin decline to 99bp, because +10.6% volume gave it operating leverage to absorb the cost. Carrier and Lennox, both growing organic at +3%, had no such cushion and both lost ~190–222bp. Nobody priced through the shock; the companies that grew volume fastest simply diluted its effect. That is an important refinement, and it is not flattering to the moat thesis: it means the variable separating winners from losers this quarter was end-market exposure and volume, not customer captivity or brand.

4.2 Timing versus moat — the answer differs by segment

The single most important analytical question in this report is whether Carrier’s margin loss is a timing lag that resolves or a competitive constraint that does not. The honest answer is that it is separable by segment, and mostly the former in the crown jewel.

The timing case — genuine, and this report gives it full weight. Section 232 was announced 2026-04-02 with mitigation pricing effective end-April; Goris: “the tariffs went into effect early April, and the pricing associated to mitigate some of that went into effect at the end of the month. And so as expected, during the second quarter, the impact of tariffs and pricing associated with tariffs was a net negative.” The 10-Q lists “improved price” as an offset in CSA, not a shortfall. Carrier fully mitigated the 2025 tariff round. And the June 8 tariff cut — with Carrier pre-committed to hand price back — mechanically explains the gap between an 8% announced increase and ~3% realized. All of this is real and self-liquidating.

The moat case — stronger, and it survives the timing explanation. Three pieces of evidence do not resolve with time:

1. The share/price trade-off is admitted. Gitlin: “We’ve actually been doing very well on share… we want to at least maintain share. So that’s kind of a balancing act.” A firm with genuine customer captivity does not choose between price and share; that is the definition of not having captivity.
2. Full pass-through still dilutes margin. Goris on Q3 pricing: it “is there to offset tariffs… it’s actually slightly a margin-dilutive impact.” Dollar-neutral recovery of a cost compresses margin percentage arithmetically. Even perfect execution leaves a permanent residue.
3. CSE is a straightforward pricing failure with no tariff involvement. The 10-Q says “pricing promotions” twice, only for CSE. The segment president was replaced. Gitlin: “we should have been more aggressive on cost.” The mix drag is battery and solar — a commodity adjacency Viessmann brought with it. None of this is a tariff-timing story.

[INTERPRETATION] CSA is predominantly timing plus a permanent mix residue; CSE is a moat failure; CSAME shows explicit price deflation in China; and CST is the one clean pass — its margin fell purely on container/truck-trailer mix with no pricing commentary anywhere, which is exactly what a real duopoly looks like in a quarter when every contested arena failed to price.

4.3 The Greenwald tests, re-run

Test 1 — market-share stability. [FACT] Gitlin states Carrier’s U.S. commercial HVAC share is going from ~10% to “15% to 20%” — the explicit basis for underwriting the new plant. [INTERPRETATION] Greenwald’s threshold is that share shifts of more than ~5 points over a decade indicate the absence of barriers to entry. A company planning a 5–10 point share gain is describing a barrier-free market. Management is telling us, in the language of a growth story, that its fastest-growing market has no moat.

Test 2 — returns on capital. [FACT] Consolidated after-tax ROIC (FY2025): Carrier 7.05%, Trane 25.1%, Lennox 29.1%. On the latest TTM Carrier is 6.28% (ROIC.ai), and this report’s amortization-and-restructuring-honest calculation puts it at 6.0% (see Financial Quality). [INTERPRETATION] Greenwald’s bands: 15–25%+ indicates competitive advantages are present; 6–8% indicates they are absent. This is the test that settles the section. Whatever Carrier’s channel position is worth qualitatively, it does not show up as excess returns on capital — and a moat that cannot be tied to a financial outcome is not a moat.

Test 3 — barriers to entry. The North American two-step distribution channel is a genuine asset: ~55,000 monitored channel-partner technicians, +35% year-over-year, is a real and growing distribution advantage. But it demonstrably delivers access, share and volume — not price. That is a distribution advantage, which is worth something, rather than customer captivity, which would be worth a great deal more.

4.4 The growth engine has no moat at all

[INTERPRETATION] The most uncomfortable synthesis in this report: Carrier’s defensible economics sit in the businesses that are shrinking or flat (residential replacement at 20–21% segment margins; transport refrigeration at ~11% of sales), and its growth sits in businesses with no barriers at all (commercial HVAC, explicitly barrier-free by management’s own share math; data centers at mid-to-high-teens margins against seven well-capitalized entrants adding capacity simultaneously). Carrier is growing its mix out of the captive part of the portfolio into the contested part — and the record backlog everyone is celebrating is the mechanism by which that happens.

Verdict: DOWNGRADED to a weak/narrow competitive advantage. What Carrier genuinely has: (i) one real duopoly — Carrier Transicold in transport refrigeration, ~11% of sales, the only segment that held price this quarter; (ii) a genuine distribution/access advantage in the North American residential dealer channel that wins volume and share but, on the June-2026 evidence, not price; (iii) scale in procurement and a real installed base. What it does not have: pricing power that survives a cost shock, any defensible position in Europe or China, a margin-accretive aftermarket (see Business Overview), or any barrier whatsoever in the segment generating all of its growth. The prior report’s “narrow but genuine moat” was too generous. A 7% after-tax ROIC against peers at 25–29% is not a moat that failed to show up in the numbers; it is the numbers telling us the moat is largely industry-structural and shared, not company-specific and owned.

5. Growth History and Forward Opportunities

5.1 The historical record — three years of no organic growth

[FACT] Continuing-operations revenue: FY2022 $17,288M; FY2023 $18,951M; FY2024 $22,486M; FY2025 $21,747M (-3.3%, -1% organic); 1H-FY2026 $11,692M (+3.2% reported, +1% organic). The FY2024 step-up was Viessmann, not growth. On an organic basis Carrier has produced roughly nothing across three years while adding $14.2B of acquired revenue and ~$7B of goodwill.

[FACT] Q2-FY2026 broke the streak: organic +3%, the first positive quarter in two years.

[INTERPRETATION] The inflection is real and should be credited as such — it arrived a quarter earlier than the prior report’s bull case required. But the Industry Dynamics section establishes that its residential component is substantially a channel restock into a flat end market, possibly amplified by a pre-buy ahead of Lennox’s disclosed July 1 price increase; and the Business Overview section establishes that its commercial component converts at lower margins than the base business. This is low-quality growth in the precise technical sense: it does not carry incremental profit. Consolidated, Carrier grew revenue $238M in Q2 and earned $71M less adjusted operating profit.

5.2 The forward algorithm, and where it strains

[FACT] FY2026 guidance: ~$23B sales, ~$3.5B adjusted operating profit, ~$2.90 adjusted EPS; data centers ~$2.0B with a $2.5B exit rate; backlog >$8B (~68% commercial, ~40% of that data centers). Orders +~40% in Q2; commercial HVAC +~65%; data centers +>300%.

[FACT] 1H-FY2026 organic was +1%. Achieving the full-year revenue guide therefore requires 2H organic of +10.4% to +14.6%, with Q4 at approximately +12–15%.

[FACT] Of the guided +20% second-half residential number, Gitlin attributed “10 of that… to the absence of destocking.” Approximately 75% of the ~$2.0B data-center revenue must land in 2H (~$500M arrived in 1H).

[INTERPRETATION] The revenue side of the guide is the more credible half — a >$8B backlog is a real forward-visibility asset, and destocking comparisons genuinely do reverse mechanically. Morgan Stanley’s Chris Snyder pressed exactly this point on the call using the two-year stack, and the answer was the destocking arithmetic above. The honest characterization is that roughly half of the second-half residential acceleration is a base effect rather than end-demand. That is not fraudulent — base effects are real money — but it is not evidence of a demand inflection either, and it does not recur into 2027.

5.3 The three genuine growth vectors, sized and stress-tested

Data centers (~$2.0B FY2026E, ~8.7–9% of sales). Genuinely secular, genuinely order-corroborated, and structurally margin-dilutive to CSA (mid-to-high-teens versus a 20–21% segment average). Carrier holds ~9–10% of a ~$21B market and is underwriting a new plant on reaching 15–20% share against seven competitors expanding simultaneously (see Industry Dynamics). Verdict: real growth, poor incremental economics, late-cycle competitive setup.

Aftermarket / service (~28% of sales). Growing +12.7% in Q2 versus +2.9% for product, double-digit in all four segments, and supported by a genuine and growing digital footprint (~55,000 monitored technicians, +35%). But Carrier’s service gross margin of 24.4% sits below its product margin of 27.6% (see Business Overview), so aftermarket mix shift is mildly dilutive at this company — the opposite of Trane, where service earns ~43% against product’s ~32.6%. Verdict: a stability and relationship asset, not the margin-accretive annuity the narrative claims.

European heat-pump electrification (CSE, ~22% of sales). The end market recovered convincingly — +13% units in 2025, Germany +50%, Q1-2026 +17% — and the German 70% BEG subsidy runs to 2029. Carrier captured the volume (CSE organic +3% in Q2, RLC up high-single-digits) and none of the profit (7.2% margin, discounting, new segment president, cheaper value-tier launch, Chinese entrants growing 13–50%). Verdict: the growth is available; Carrier cannot monetize it. This is now the clearest evidence in the report that the Viessmann thesis was mis-underwritten (see Capital Allocation).

5.4 What is shrinking

[FACT] CSA commercial -8% in Q2 (management: delivery timing); China residential/light-commercial -13–14% with industry shipments -5.6% and Xiaomi pricing ~7% below market; global truck and trailer down low teens, with ACT data implying the trailer cycle turns in 2027; and the portfolio itself is shrinking by disposal — Riello (closed 07-01) and NORESCO (closed 08-03) both exit the 2H revenue base.

Verdict: Growth quality is poor and deteriorating in composition. The top-line inflection is real but is roughly half base effect and channel restock; the fastest-growing revenue converts at below-average margins; the aftermarket is dilutive rather than accretive at this company; and the one large end market that genuinely inflected (European heat pumps) produced volume without profit. Carrier is growing again, and the growth is not currently creating value — which is the precise definition of the condition that makes a business uninvestable at a premium multiple.

6. Financial Quality

6.1 The quarter and the half

[FACT] Q2-FY2026 (10-Q, Condensed Consolidated Statement of Operations): net sales $6,351M vs $6,113M (+3.9%); gross profit $1,728M vs $1,769M — gross margin 27.2% vs 28.9%, -173bp on higher sales; GAAP operating profit $825M vs $903M (-8.6%); adjusted operating profit $1,095M vs $1,166M (-6.1%), margin 17.2% vs 19.1% (-190bp); GAAP continuing diluted EPS $0.60 vs $0.70 (-14%); adjusted EPS $0.86 vs $0.92 (-7%).

[FACT] 1H-FY2026: net sales $11,692M (+3.2%); GAAP operating profit $1,083M vs $1,532M (-29.3%); adjusted operating profit $1,689M vs $2,014M (-16.1%), margin 14.4% vs 17.8%; gross margin -300bp; SG&A $1,672M vs $1,542M (+8.4%).

[FACT] The incremental economics: consolidated revenue +$238M, adjusted operating profit -$71M — an incremental adjusted operating margin of -29.8%; incremental segment operating margin -40.8%; incremental gross margin -17.2%. In CSA alone: sales +$120M, segment profit -$56M, i.e. $88M / 261bp of erosion and a -47% incremental margin.

[INTERPRETATION] This is the arithmetic the market repriced on July 28. A business that grows and earns less money doing it has, for that period, negative operating leverage — the opposite of the thesis a premium multiple rests on. The decomposition of the ~190bp adjusted-margin decline is approximately -185bp gross margin and -37bp lower joint-venture income, partly offset by ~+50bp of operating expense leverage. Note that the JV component (-$20M in Q2, -$33M on the half, attributed partly to the Middle East conflict) is not an operating failure and should not be charged against the moat argument.

6.2 The guidance bridge — a Q4 hockey stick

[FACT] Reconstructed adjusted operating profit by half (FY2025 10-K non-GAAP reconciliation p.37; Q3-FY2025 10-Q; Q2-FY2026 10-Q):

[FACT] The required 2H figure implies +41.7% YoY growth in adjusted operating profit off 2H-FY2025’s $1,278M. Management’s spoken framing on the call — 2H “up about 50%” — implies ~$1,911M and a 16.9% margin, +463bp YoY. Applying management’s own Q3 guide (~$5.95B revenue, ~16.5% margin, ~$0.75 adjusted EPS) isolates the burden in the fourth quarter: Q4 implied adjusted operating profit of $829–929M against $455M actual, i.e. +82% to +104%, at a 15.5–17.3% margin versus a post-spin Q4 best of 13.2%. Implied Q4 adjusted EPS is $0.72 versus $0.34, +112%.

[INTERPRETATION] The report should not caricature this. Three things genuinely support the guide: (i) the Q4-FY2025 comparison is exceptionally easy — $455M at a 9.4% margin was the destocking trough, not a normal quarter; (ii) FY2024 demonstrates this business can run a 15.5% second half, so ~16.0% is a stretch of ~50bp against a real precedent rather than an invention; and (iii) a >$8B backlog gives genuinely unusual revenue visibility, and announced price increases build through a year. Nor is the seasonal shape anomalous — the FY2026 guide implies a 48%/52% first-half/second-half split, close to FY2024’s 51%/49%; FY2025’s 61%/39% was the aberration. Do not argue the bear case from seasonality.

The bear case must be argued from the margin level and its driver. The load-bearing assumption is not revenue — it is that price/cost turns decisively positive in 2H. And management has already conceded that the tariff-offsetting pricing is “slightly a margin-dilutive impact” (see Competitive Position), that realized price is running ~3% against an 8% announcement, and that the Q3 guide itself steps CSA margin down from 24.4% to ~22%. The guide asks for the best second half in three years from a company whose most recent guidance step is downward and whose pricing is running at half of plan.

6.3 Earnings quality — worse than the prior report concluded

[FACT] Adjusted operating profit now stands 56% above GAAP operating profit; add-backs grew 26% YoY to $606M in 1H-FY2026 (restructuring $116M vs $55M; acquired-intangible amortization $426M; acquisition/divestiture costs $18M; the $46M Riello impairment).

[FACT — the new finding] The 1H-FY2026 adjusted effective tax rate was 15.1%, against 22.1% a year earlier. The $99M Swiss valuation-allowance release and the $18M state audit settlement were not excluded from adjusted EPS. The benefit is worth approximately $0.12–0.16 per share.

[INTERPRETATION] The prior report established that GAAP earnings were tax-flattered. This report establishes the sharper point: the “adjusted” number is tax-flattered too. Correcting for it, 1H adjusted EPS fell 17–19% year-over-year, not the reported 9%, and the FY2026 adjusted-EPS guide is worth approximately $2.74–2.78, not $2.90. This is a material haircut to the number every forward multiple in the market is computed on, and it is not disclosed as an adjustment.

[FACT] Amortization-honest FY2026E EPS — charging the ~$850M/yr of acquired-intangible amortization that recurs through 2030 (FY2025: $856M; Q2-FY2026: $213M) — is approximately $2.10–2.13. Charging recurring restructuring as well: approximately $1.89–1.92.

[FACT] TTM as of 2026-06-30: sales $22,108M; GAAP operating profit $1,723M (7.79%); adjusted operating profit $2,967M (13.42%); adjusted EPS $2.45 (below FY2025’s $2.59); GAAP continuing EPS $1.40; adjusted EBITDA $3,383M.

[INTERPRETATION] Every profitability line is below the FY2025 baseline the prior report used. The company is not recovering toward the consensus bridge; on a trailing basis it is still moving away from it.

6.4 Cash flow — a correction, and a genuine new qualification

[FACT — correcting the prior reading] Carrier’s reported free cash flow is computed off total operating cash flow including discontinued operations, contrary to its own published definition (“net cash flows provided by continuing operating activities less capital expenditures”). On the correct continuing-operations basis, 1H-FY2026 FCF rose 22% to $742M from $608M, and working capital consumed $381M less cash than a year earlier. The headline “-20%” comparison is an artifact.

[FACT] Q2 operating cash flow $927M; Q2 FCF $810M. Capital expenditures $211M in 1H versus $144M, with full-year capex guidance raised to ~$600M (+53%) for the new U.S. data-center plant. Capex/sales moves from 1.80% to ~2.61%. Free-cash-flow guidance was not raised alongside the profit guide.

[INTERPRETATION] The cash correction is a genuine point in Carrier’s favour and this report credits it: underlying cash conversion improved, and the working-capital release is real. But the capex inflection is equally real and structurally more important. The profit raise is being consumed by capital intensity. A ~45% increase in capital intensity to chase share in the lowest-margin part of the portfolio (see Industry Dynamics) is the single clearest financial expression of this report’s competitive conclusion — Carrier is spending more capital to grow into worse economics. The “asset-light compounder” description that supported a premium multiple is now materially qualified.

6.5 Returns on capital — the number that settles the business-quality question

[FACT] On average invested capital of $24,370M: GAAP ROIC 5.5% (prior report: ~7.0%); company-adjusted ROIC 9.5% (prior: ~10.7%); amortization-and-restructuring-honest ROIC 6.0%. ROE 8.4% (prior: ~10.5%). ROIC.ai independently shows a decline from 7.05% to 6.28% across three TTM periods. Peers: Trane 25.1%, Lennox 29.1% (FY2025 after-tax ROIC).

[FACT] Even a fully delivered FY2026 guide only lifts honest ROIC to ~8.6% — still below a ~9% WACC [ASSUMPTION].

[INTERPRETATION] This is the most important number in the report and it has moved the wrong way since June. Carrier does not earn its cost of capital on a GAAP or an honest basis, and clears it only on management’s own adjusted definition — which adds back the amortization created by the acquisition that diluted the returns in the first place. A company that earns ~6% on capital while its two closest comparables earn 25–29% does not have a competitive advantage; it has a scale position in a good industry that it is not converting into economic profit. The ~$7.6B of CSE goodwill earning a 7% segment margin is the arithmetic centre of the problem.

6.6 Balance sheet

[FACT] At 2026-06-30: cash $1,344M; short-term borrowings and current portion of long-term debt $1,638M (from $468M — decomposed as commercial paper $325M→$685M and current LTD $108M→$916M, i.e. mostly a maturity reclassification of the 2027 notes plus $360M of genuinely new commercial paper); long-term debt $10,314M; net debt $10,608M, up $330M in the half. Goodwill $15,267M; net intangibles $5,756M; assets held for sale $815M against liabilities held for sale $414M.

[FACT] Net debt / TTM EBITDA 3.53x GAAP, up from 2.55x a year ago — driven by a -27% TTM EBITDA decline, not by incremental borrowing; ~2.7–3.1x on adjusted definitions. Tangible book value -$7,876M, or -$9.55 per share.

[INTERPRETATION] Investment-grade and not distressed, but leverage is rising for the least attractive reason: the denominator is collapsing. A 3.5x GAAP leverage ratio on a business earning 6% on capital with negative tangible book leaves the equity with no asset floor and limited capacity to absorb either a goodwill impairment (non-cash but rating-relevant) or an adverse antitrust outcome.

6.7 Goodwill — no impairment, and the exposure is fully intact

[FACT] Goodwill moved $15,501M → $15,267M in 1H. The $234M decline was 99% foreign-currency translation (-$231M, of which -$219M in CSE), plus a $13M held-for-sale reclassification. CSE goodwill stands at $7,586M. Riello carried only $3M of CSE goodwill, so its disposal did not reduce the exposure. There was no impairment.

[FACT] The annual impairment test date is July 1 (10-Q Note 4). The FY2026 test therefore falls in Q3-FY2026 and its result will not be public until the Q3-FY2026 10-Q, expected late October 2026. The most recent public measurement remains the ~14% CSE fair-value cushion disclosed at FY2025 as a PwC Critical Audit Matter — untested in public since 2025-07-01.

[INTERPRETATION] The prior report’s flagged risk is neither realized nor removed; it is simply still pending, and it now has a date. Two developments since June bear on the likely test outcome and they point in opposite directions: CSE’s end market improved materially (see Industry Dynamics), which supports the cushion; but CSE’s margin fell further to 7.0–7.2%, it absorbed $89M of restructuring in one half, its president was replaced, and — most concretely — the Riello disposal required a $46M write-down to clear at ~0.72× sales / 8.3× EBITDA, the first arm’s-length mark on European thermal assets since the deal (see Capital Allocation). A discounted-cash-flow test is driven by forecast margins and discount rates, not by end-market unit growth. A cushion of ~14% is thin enough that a modest downward revision to CSE’s long-run margin assumption, or a modest rise in the discount rate, closes it.

Verdict: Financial quality deteriorated across essentially every dimension since the prior report, with one genuine correction in Carrier’s favour (continuing-operations FCF rose 22%, not fell 20%). Margins fell in all four segments on growing volume; incremental margins were negative; returns on capital fell to ~6% honest and remain below cost of capital even on a fully-delivered guide; leverage rose on a shrinking denominator; capital intensity increased ~45% without a corresponding FCF guide; and the “adjusted” earnings on which the market’s forward multiple rests are themselves flattered by ~$0.12–0.16 of undisclosed tax benefit. The economics do not currently improve with scale.

7. Capital Allocation

7.1 The portfolio reversal

[FACT] Since the prior report, Carrier has completed two disposals and one acquisition:

[FACT] Running portfolio tally since 2023: ~$14.2B acquired versus ~$10.42B divested — net ~-$3.8B.

[FACT — correcting the prior report] Riello was not a Viessmann asset. United Technologies acquired ~70% of the Riello Group in 2016; it sits inside CSE only because CSE is the Europe segment. The prior report characterized it as Viessmann-adjacent, which is wrong.

[INTERPRETATION] The correction does not weaken the finding; it sharpens it. Carrier bought $14.2B of European thermal exposure in January 2024 and by August 2026 is shrinking that footprint — divesting a legacy European thermal business at a loss while its acquired European business earns 7%. Whatever the strategic logic of pruning a non-core tail, the direction of travel is a retreat from the geography the company spent $14.2B to enter two years ago.

7.2 The Riello price — the first arm’s-length mark on the Viessmann-era thesis

[FACT] Ariston disclosed the transaction economics that Carrier did not: approximately €289M enterprise value on ~€400M of revenue and ~€35M of adjusted EBITDA — 0.72× sales and 8.3× EBITDA.

[FACT] Carrier paid €12B for Viessmann Climate Solutions at approximately 3.0× sales and ~17.1× unsynergized EBITDA.

[INTERPRETATION] This is the analytical prize of the update. Since January 2024 there has been no observable market price for a European thermal asset — every judgment about whether Carrier overpaid has been an inference from segment margins and goodwill cushions. Ariston has now supplied a clearing price, and it is a fraction of what Carrier paid. Riello and VCS are not identical assets — VCS is a stronger brand with a better product mix, and a premium is defensible — but a gap of 8.3× versus ~17.1× is not a quality premium; it is a different valuation regime. The Riello mark is a small, recent, cash, arm’s-length data point, and it points directly at the ~$7.6B of CSE goodwill still carried at a ~14% cushion (see Financial Quality).

7.3 The 2023 underwriting versus the 2026 outcome

[FACT] From the acquisition 8-K and associated materials (2023-04-25), Carrier underwrote: €0.7B of EBITDA on €4B of sales; €200M of synergies with “the vast majority by year three” — i.e. 2026; “the $5 billion European heat pump segment… expected to triple to $15 billion by 2027”; and a “return to pre-transaction leverage within approximately two years.”

[FACT] Outcomes as of Q2-FY2026: CSE runs a 7.0% segment margin; its SG&A load is 21.1% of sales versus 21.3% a year earlier — essentially unchanged; CSE gross margin is 29.7%, which is healthy; cumulative CSE restructuring since close is $251M — 62% of all segment restructuring — with 1H-FY2026’s $89M nearly matching all of FY2025; net leverage is 3.53x GAAP, above pre-transaction levels; and no quantified synergy scorecard appears in any SEC filing.

[INTERPRETATION] Two things stand out. First, the synergies were aimed at the wrong line. CSE’s gross margin of 29.7% is not the problem; its 21% SG&A load is, and that has not improved at all in a year despite $251M of restructuring spend. Second, the market-size premise was simply wrong — the European heat-pump market did not triple to $15B by 2027; it grew 13% in 2025 to 2.9M units and, more damagingly for the thesis, grew without margin (see Industry Dynamics). Carrier underwrote a volume explosion and got a modest volume recovery at commodity economics. The absence of any published synergy scorecard three years into a deal whose synergies were promised “by year three” is itself a disclosure choice worth noting.

[FACT] On 2026-06-17 — three days after the prior report — Thomas Heim, who had led the Viessmann integration for nine years, was replaced as CSE president by Thomas Donato, an external hire (ex-Bosch Power Tools CEO, previously Rockwell and ABB). No 8-K was filed. Six weeks later Gitlin stated publicly that “we should have been more aggressive on cost” and set a mid-teens margin target “over the next few years” — an ~800bp repair on a 7.2% base, with no dated milestone.

[INTERPRETATION] Replacing the integration leader is the clearest available admission that the integration has not worked. Setting an ~800bp margin-recovery target without a date is the second clearest.

7.4 Buybacks and distributions — genuine discipline, insufficient funding

[FACT] 1H-FY2026 repurchases: 12.0M shares at an average of $62.33 — against a 2026-08-14 close of $62.78, i.e. essentially breakeven, not under water. The pattern within the half shows deliberate opportunism: 2.38M shares bought in April at $60.52, stepping down to 643k shares in June at $72.00. Program-to-date average is $59.24, ~6% in the money. Diluted weighted-average shares fell to 836.5M from 866.3M (-3.4%).

[FACT] 1H-FY2026 total shareholder distributions were $1,210M against $795M of reported free cash flow, with the gap funded by $361M of new commercial paper (stated in the MD&A). Since YE2024: $4.81B returned versus $2.92B of FCF — a $1.89B shortfall that substantially accounts for the $2.22B increase in net debt.

[INTERPRETATION] The execution deserves credit and the prior report’s working assumption (that 2026 buybacks would be under water after the July collapse) is wrong — Carrier bought heavily at $60.52 in April and deliberately stood down at $72.00 in June. That is countercyclical behaviour, not price-insensitive absorption of stock. The criticism is not the price paid; it is that the buyback is debt-funded. Returning ~165% of free cash flow while margins are compressing, leverage is rising on a shrinking EBITDA base, and capital intensity is increasing ~45% is a financing choice, not a return of surplus capital.

7.5 The Viessmann family overhang

[FACT] A February 2026 shelf registration registered the entire 50.07M-share Viessmann stake for resale at Carrier’s expense. The family sold 12.1M shares at $62.01 in May 2026 (Form 4 / Schedule 13D/A, 2026-05-20). 37.98M shares — ~4.6% of the company, ~$2.4B at current prices — remain freely saleable.

[FACT] The S-3ASR filed 2026-07-28 is a routine reoffer of 715,279 shares under long-term incentive plans — not a capital raise.

[INTERPRETATION] The demonstrated clearing price for the family’s stock is $62.01, against a current price of $62.78. That is a concrete, sizeable, and price-anchored supply overhang from the single best-informed holder of the Viessmann asset — the people who sold it to Carrier. It is not a thesis-killer, but it is a real constraint on upside and a soft negative signal about how that holder values the business it built.

7.6 Insider behaviour

[FACT] Across 189 Form 4 filings over five years there is exactly one open-market purchase (transaction code P): CEO David Gitlin, 19,300 shares at $52.6203 on 2025-11-25 — up ~19.3% marked to $62.78. Everything else is grants (A), option exercises (M), tax withholding (F), and planned sales.

[FACT] No insider has purchased since the July collapse. The last Form 4 of any kind was 2026-07-24, four days before the print. The trading window has been open in the weeks since and no purchases have been filed.

[FACT] New Form 3s: Neil Barua (PTC CEO; appointed director 2026-07-24 to the Compensation and Technology & Innovation Committees) and Thomas Donato (CSE president). Schedule 13G/A filings show BlackRock at 7.6% and Capital Group at 12.8%.

[INTERPRETATION] The single 2025 purchase remains the strongest bullish insider tell in the file, and it was well timed. But its informational value decays: it was made nine months ago, ~16% below today’s price, at a very different price and a very different understanding of the margin structure. The absence of any follow-up purchase after a 21% peak-to-trough decline — with the window open — is a meaningful negative. If management believed the July sell-off was a tariff-timing misunderstanding of an intact franchise, that is the moment the file should show a purchase. It does not.

7.7 Incentives

[FACT] The 2026 DEF 14A contains zero occurrences of “ROIC,” “ROE,” or “economic profit.” The metrics that determine annual and long-term incentive payouts are Sales, Adjusted Operating Profit, Adjusted EPS (CAGR), Free Cash Flow, and Relative TSR.

[INTERPRETATION] Every one of the four financial metrics is mechanically increased by a debt-funded, goodwill-heavy acquisition: a purchased revenue stream raises Sales; purchased EBIT raises Adjusted Operating Profit; and three of the four are computed on a basis that adds back the ~$850M/yr of amortization the acquisition itself created. There is no charge for the capital employed anywhere in the plan. This is not an accusation of bad faith — it is a structural observation that the compensation design is tolerant of value-destructive growth, and that management, in doing the Viessmann deal, was paid to do exactly what it did. The absence of any return-on-capital hurdle at a company earning ~6% honest ROIC against peers at 25–29% is the governance finding of this report.

Verdict: NEGATIVE on the central capital-allocation decision, with genuine competence at the margins. The Viessmann acquisition was underwritten on a market-size premise that did not materialize, promised synergies that have not been disclosed as delivered, aimed its cost programme at the wrong line item, has consumed $251M of restructuring, has now cost the integration leader his job, and has just been marked — indirectly but concretely — by an arm’s-length European thermal transaction at less than half the multiple Carrier paid. Against that: buyback execution has been genuinely opportunistic and near-breakeven, the balance sheet remains investment-grade, and the portfolio pruning is defensible on its own terms. The offsetting competence is real but small relative to a $14.2B decision that has diluted returns on capital below the cost of capital.

8. Changes and Headwinds — Last Two Years

[FACT] Material developments, most recent first:

[INTERPRETATION] Do these strengthen or weaken the thesis? On balance, weaken — but with two genuine offsets.

Weakening. The Q2 margin outcome converted a cyclical question (“when does volume return?”) into a structural one (“can this company price?”), and the peer data answered it unhelpfully (see Competitive Position). The European volume recovery arrived and produced no profit, reclassifying Europe from bad cycle to bad industry (see Industry Dynamics). The Riello mark supplied a hostile external datapoint on European thermal valuations. The CSE presidency change is an admission of integration failure. The antitrust complaints add an unquantified tail. Capital intensity rose ~45% without an FCF guide.

Offsetting. The organic inflection is real and arrived early. The tariff mechanism is genuinely benign and self-liquidating with a stated 2027 sunset. Carrier’s zero-IEEPA-refund position understated its Q2 versus peers on identical facts and leaves an unrecognized asset. The order book and >$8B backlog are real forward visibility. Buyback execution was disciplined. And no broker cut numbers — estimates rose while the price fell.

9. Risk Analysis

[INTERPRETATION] The risk profile has changed shape since June. The prior report’s dominant risk was a single tail event — a CSE goodwill impairment — sitting on top of a cyclical timing question. That tail is unchanged and now has a date. But it has been displaced as the primary risk by a structural one: risks 1, 2 and 4 together describe a company that may be guiding to a margin level it cannot reach, in an industry that has just demonstrated it cannot price, while growing into its lowest-margin end market. Those are correlated — a pricing failure is precisely what would cause the Q4 miss — which means the downside scenarios cluster rather than diversify.

Catastrophic-loss risk is low. Carrier is investment-grade, generates $1.5–2.0B of free cash flow, holds a genuine #1/#2 position in most of its markets, and faces no solvency question. The realistic bear case is multiple compression and years of sub-cost-of-capital returns, not impairment of the enterprise.

10. Valuation Discussion (Embedded Expectations)

Market anchors (FACT, as of 2026-08-14): price $62.78; shares outstanding 824.3M; market cap ~$51.75B; net debt $10,608M; non-controlling interests $325M; enterprise value ~$62.7B. TTM (6/30/26): sales $22,108M; adjusted operating profit $2,967M; adjusted EBITDA $3,383M; adjusted EPS $2.45; GAAP continuing EPS $1.40; continuing FCF $1,831M. FY2026 guide: ~$23B sales, ~$3.5B adjusted operating profit, ~$2.90 adjusted EPS. Non-acquisition D&A $416M. [ASSUMPTION] WACC 9.0% (beta 1.21; sensitivity band 8.5–9.5%), terminal growth 2.5%.

10.1 Three corrections to the prior report’s valuation work

Update discipline requires re-testing the prior analysis, not just re-pricing it. Three of its conclusions do not survive scrutiny and are corrected here.

(i) The acquired-intangible amortization does not distort EV/EBITDA. The prior report implied that Carrier’s ~$850M/yr of amortization made EV/EBITDA comparisons with Lennox unfair. It does not: EBITDA adds back all depreciation and amortization by construction. The amortization distorts EV/EBIT and every earnings multiple — where it matters a great deal — but not EV/EBITDA. What genuinely distorts the trailing EV/EBITDA comparison is $285M of one-time items plus Carrier’s trough earnings base.

(ii) The Lennox premium widened by four points, not twenty-three. Raw trailing: CARR ~20.8× vs LII ~14.2× (+47%). Correcting only for the one-timers: 18.5× vs 14.2× — a +30% premium, against +26% in the prior report. The gap widened modestly. The prior report’s headline verdict — “CARR is not the value name in HVAC” — survives, but by a much narrower margin than a naïve reading of the current multiples suggests.

(iii) Carrier is now ~7% cheaper than Trane on an amortization-honest basis, not richer. This reverses the prior report’s most quotable claim (“~29–31× current-power earnings — richer than Trane”). At $62.78 the amortization-honest forward multiple is ~29.8× against Trane’s comparable ~32×. The prior claim was correct at $67–70 against a $2.80 consensus; it is not correct at $62.78 against a $2.90 guide.

10.2 Comparable companies at current prices

[FACT] Computed at 2026-08-14 closes. Note: ROIC.ai’s get_enterprise_value values each company off its 2026-06-30 quarter-end price and would overstate Carrier’s multiple by ~16%; the figures below are re-struck at current prices.

JCI’s ROIC-sourced EBITDA is implausible and is excluded rather than reported unreliably. Watsco (WSO) is the single most factor-similar name to CARR (similarity 0.957, ahead of TT and LII) and is retained in the comp discussion as the aftermarket-annuity benchmark.

[INTERPRETATION] Carrier sits materially below Trane and materially above Lennox on every measure — and its returns on capital are a quarter of either’s. The ~10-turn EV/EBITDA discount to Trane is deserved: Trane earns 25.1% on capital to Carrier’s ~6–7%, grew organic three times as fast in the quarter, and has a genuinely accretive service business. The ~2–3 turn premium to Lennox is the harder one to defend, since Lennox earns 29.1% on capital. The honest read is that Carrier is priced roughly where a #3-quality operator in a good industry should be priced — neither the bargain the forward P/E implies nor the egregious premium the trailing multiple implies.

10.3 The GAAP-to-adjusted decomposition ladder at $62.78

[INTERPRETATION] Two conclusions. First, the de-rate is real but it is entirely borrowed against the raised guide. On forward amortization-honest earnings the stock went from ~33.3× to ~29.8× — about 11% cheaper. But on trailing amortization-honest earnings it went from ~29–31× to ~33.0× — i.e. more expensive than in June, at a 7% lower price, because the earnings base deteriorated faster than the multiple. Every bit of the apparent cheapening depends on the Q4 hockey stick landing. Second, the “21.6× forward” that makes Carrier screen cheap is a stack of three optimistic adjustments: accept ~$850M/yr of real amortization as non-economic, accept an undisclosed ~$0.15 of tax benefit as recurring, and believe a Q4 that requires +82–104% growth. Strip all three and the number is ~33×.

10.4 Embedded expectations — what the market is actually underwriting

[FACT / ASSUMPTION] At EV $62.7B, a 9.0% WACC and 2.5% terminal growth, the current price requires approximately 6% free-cash-flow growth in perpetuity, or equivalently about 12.3% annually for a decade before fading, from a ~$1.9B base.

[FACT — correcting the prior report] The prior report’s scenario arithmetic contained an error: $1.9B growing at 6% for ten years and 2.5% thereafter discounts to ~$39B, not the $62–66B implied there. 6% is the required perpetual rate; it is not sufficient as a ten-year rate.

[INTERPRETATION] Requiring ~6% perpetual FCF growth from a business that has produced no organic growth in three years, earns ~6% on capital honestly measured, and is raising capital intensity ~45% is a demanding ask. It is not an absurd one — replacement demand is real and the installed base compounds — but it leaves no room for the pricing problem identified under Competitive Position to be structural.

The most interesting question this quarter: why did the market sell a raised guide? The answer is visible in the multiple. Carrier traded at ~18.7× forward adjusted EBITDA at the June 25 peak and ~15.9× today — on the same guide, subsequently raised. Estimates went up; no broker downgraded; Bernstein raised its target to $78 and Argus raised; Zacks moved the stock to Rank #1 as estimates rose while the price fell. The market therefore did not re-price the FY2026 number. It re-priced the quality of that number — 3% realized price against an 8% list increase, management conceding tariff pricing is “margin-dilutive,” a flagged 2027 headwind, and a 53% capex step-up with no FCF benefit. That is a multiple event, not an earnings event, and it is precisely what a market does when it decides a business is worth fewer turns rather than fewer dollars.

10.5 Scenarios

[ASSUMPTION-heavy — every input stated] Hinged on the two swing variables: the Q4 adjusted operating margin, and whether the CSE goodwill survives its July-1 test.

[INTERPRETATION] The price at $62.78 sits just below the base band, versus at the low end of the base band in the prior report — a modest, not decisive, improvement in asymmetry. Downside to the bear case is roughly -17% to -31%; upside to the bull case is roughly +47% to +74%. That looks attractive on raw arithmetic, but the probability weights are not symmetric: the bull case requires a Q4 that would be the best in the company’s post-spin history, delivered by a company whose pricing is running at ~40% of plan.

[FACT — correcting the prior report] A CSE goodwill impairment leaves tangible book value arithmetically unchanged (goodwill is already excluded from tangible book). The prior report implied it would “wipe out a slice of an already-negative tangible book,” which is incorrect. The damage from an impairment is informational, not balance-sheet arithmetic — it would be the auditor’s confirmation that the deal destroyed value — and this report estimates it at ~$10–12/share of value if it causes the market to re-mark CSE toward the Riello comparable.

[NOTE — an asymmetry in the guide worth flagging] A revenue shortfall makes the required 2H margin harder, not easier: holding the ~$3.5B adjusted operating profit target constant, a revenue miss lifts the implied 2H margin from ~16.0% to ~16.8%. The Riello and NORESCO disposals both remove 2H revenue.

10.6 Sum-of-the-parts — the sharpest test in this section

[FACT / ASSUMPTION] Valuing each segment on defensible peer multiples against the guide-achieved earnings base produces total enterprise value of $45.7–58.7B — below the market’s $62.7B even at the high end.

[INTERPRETATION] Running it in reverse is more revealing. Marking CSA, CSAME and CST at peer-appropriate multiples, the market is implicitly valuing CSE/Viessmann at roughly $11–22B, with a central estimate near $16B — at or above the $14.2B Carrier paid — for a segment earning a 7.0% margin, running pricing promotions, on its third strategic reset, and whose closest recent transaction comparable (Riello) cleared at ~0.72× sales / 8.3× EBITDA and required a write-down to sell. Marking CSE at anything close to that comparable removes roughly $10B of enterprise value, or ~$12/share.

[INTERPRETATION] There is genuine tension between this SOTP (which implies the equity is 7–20% overvalued) and the base scenario in the table above (which implies it is modestly undervalued). This report does not resolve it artificially. The SOTP is a cross-sectional, peer-multiple view that is unforgiving about CSE; the scenario table is a forward-earnings view that credits the guide. The gap between them is, almost exactly, the value of the Viessmann question — which is why the July-1 goodwill test result, landing in the Q3 10-Q in late October, is the single most informative scheduled disclosure on the calendar.

10.7 Own-history percentiles and shareholder yield

[FACT] AZI valuation_index, 2026-08-14 (n=3), with the 2026-06-12 reading in brackets: P/S 74.9 [90.0]; composite 60.4 [72.3]; P/E 91.3 [99.5]; P/B 14.9 [27.4]. Price $62.78; TTM EPS $1.5257; BVPS $15.9611; TTM sales/share $25.6042.

[INTERPRETATION] Read only against Carrier’s own post-spin history. P/S is the clean signal and it fell from the 90th to the ~75th percentile — a genuine de-rate, but still the expensive quartile of its own history. The P/E percentile (91.3) is corrupted by the depressed, low-quality TTM GAAP EPS denominator of $1.53 and should be ignored. The P/B percentile (14.9) is meaningless: book value is almost entirely goodwill and intangibles against -$7.9B of tangible equity, so a “cheap” P/B is an artifact, not a margin of safety. The composite’s fall from 72nd to 60th therefore overstates the improvement, because it is dragged down by the misleading P/B.

[FACT] Shareholder yield at $62.78: dividend 1.53% plus buyback ~2.9% = ~4.4–4.9% total, against an FCF/EV yield of 2.92%. Distributions are running at ~115% of free cash flow, part-funded by disposal proceeds and commercial paper while capex steps up 53%.

[INTERPRETATION] A ~4.5% shareholder yield is genuinely supportive and is one of the better arguments for the stock at this price. But it is not covered by free cash flow, and the gap is being financed. A yield funded from the balance sheet is a return of capital, not a return on it.

Verdict: Carrier has genuinely de-rated and is no longer expensive — own-history P/S from the 90th to the 75th percentile, forward amortization-honest P/E from ~33× to ~30×, and now a ~7% discount to Trane on that basis rather than a premium. But it is not cheap, and the cheapening is borrowed against a Q4 that requires +82–104% profit growth. On trailing honest earnings it is more expensive than in June. The SOTP does not clear the market price even at its high end, and the market is implicitly carrying CSE at or above what Carrier paid for it. The fair conclusion is that price and value have converged — which is a materially different statement from value having appeared.

11. Variant Perception

11.1 The consensus belief — and what the tape says about it

[FACT] The sell-side did not capitulate after the Q2 print. There were no downgrades. Bernstein raised its target to $78 (from $75); Argus raised; consensus remains Buy across 24 analysts with a median target of $78 and a range of $60–$90; Zacks moved Carrier to Rank #1 as estimates rose while the price fell. Institutional ownership is ~89.1%; short interest is 2.66% of float, up 14.8% month-on-month.

[INTERPRETATION] The consensus model is a quality cyclical mid-inflection: residential destocking ends, the >$8B backlog converts, data centers compound to $2.5B, price/cost turns positive in 2H as the tariff timing washes out, and buybacks shrink the count. The critical observation is that consensus and price have decoupled. Estimates went up and the stock went down 13.6%. That is not a market disagreeing about the number; it is a market re-rating the quality of the number — and it means the variant-perception question is not “will Carrier hit $2.90?” but “what is $2.90 of Carrier earnings worth?”

[FACT — the positioning read] The factor evidence sharpens this considerably. The ElasticNet model zeroes both Momentum and Value in all four nested specifications. Carrier was never a factor-momentum name even at the June peak — so July was not a momentum crash (the momentum factor is +10.3% over 252 days); it broke on its own news. And it is not yet a value name — Value is zeroed, and the own-history composite sits at only the 60th percentile. Its one distinctive loading is Industry: Home Construction at +0.431, its largest exposure after Market and its own sector — and that basket is the worst performer in the entire model over twelve months (-14.4%, z -1.46). Meanwhile the two factors working best — DividendYield (+18.1%, z +1.77) and Value (+13.7%, z +1.61) — are ones Carrier barely or does not load on. Alpha is -0.193; idiosyncratic volatility 25.8% on an R² of 0.53.

[INTERPRETATION] Carrier is in a factor no-man’s-land: a housing-linked cyclical that is not cheap, in a regime that is paying for cheap and punishing housing-linked cyclicals. There is no value bid waiting underneath it and no momentum cohort holding it up. That is the honest positioning answer, and it explains the low-conviction bounce — the recovery off the July 29 low has come on roughly half-normal volume (0.51× the 90-day average over the last five sessions), with Carrier bouncing +4.8% against SPY +6.4%, TT +7.5% and JCI +10.2%. No second distribution wave, but no accumulation either.

[FACT] The two-day sell-off decomposes cleanly and the decomposition matters. On 2026-07-28 — Carrier’s own print — CARR fell 8.90% while the HVAC peer mean was -1.84% and the S&P 500 was up 0.24%. That is ~7.1pp of idiosyncratic decline versus peers and ~9.2pp beta-adjusted versus the market. On 2026-07-29 — Lennox’s print — CARR fell 5.13% against a peer mean of -8.84%, i.e. it outperformed. Roughly 7 of the 13.6 points was the market’s independent verdict on Carrier’s own quarter; the remainder was the residential complex being marked down around it.

11.2 The strongest bull case

A scaled climate business at ~21.6× forward earnings and ~15.9× forward EBITDA — a ~10-turn EBITDA discount and a ~7% amortization-honest earnings discount to Trane — that has just returned to organic growth a quarter early, holds a record >$8B backlog with orders up ~40%, and is guiding data-center revenue to $2.0B with a $2.5B exit rate into a market growing toward $1T of hyperscaler capex. The margin problem has a specific, dated, self-liquidating cause: Section 232 was cut on June 8, Carrier had pre-committed to hand price back, and it did so with high-cost LIFO inventory still in COGS — with the relief scheduled to sunset 2027-12-31 and a demonstrated track record of fully mitigating the 2025 tariff round. Carrier also booked zero IEEPA refund while peers booked $30M and ¥10.5B, flattering their comparatives and leaving Carrier an unrecognized asset. The European end market has genuinely inflected (+13% units 2025, Germany +50%, Q1-2026 +17%) with the German subsidy secured to 2029, giving a 7.2%-margin segment enormous operating leverage if the new president executes toward mid-teens. Continuing-operations free cash flow rose 22% in the half. Buybacks were struck at $62.33, below today’s price, with management deliberately standing down at $72. Estimates are rising, no broker has cut, and the CEO bought stock at $52.62 with his own money. On this view, July was a timing quarter mistaken for a structural one.

11.3 The strongest bear case

The June-2026 quarter was a controlled experiment on pricing power and Carrier failed it — as did three of its four largest competitors, which is worse, because it means the 2022–25 margin expansion was a shared regulatory windfall (SEER2, A2L, the R-410A pre-buy) rather than any company’s moat. Carrier announced 8% price, guided 6–7%, delivered 3%, and the CEO explained that the company chose share over price. CSA grew sales $120M and lost $56M of profit — a -47% incremental margin. The financial signature is decisive: after-tax ROIC of 6–7% against Trane’s 25.1% and Lennox’s 29.1%, below a ~9% WACC on every honest measure, and still below it even if the guide is met in full. The guide itself is a Q4 hockey stick requiring +82–104% profit growth to a margin never achieved in a post-spin fourth quarter, and its EPS is flattered by $0.12–0.16 of undisclosed tax benefit inside the “adjusted” number. Growth is migrating from 20–21%-margin residential replacement into mid-to-high-teens commercial and data-center work, at the top of a capital cycle where AAON and Modine have already shown what that ramp does to margins. Europe took its volume recovery and produced 7.2% margins, pricing promotions, a fired segment president and a cheaper value-tier launch — converting the bear case from bad cycle to bad industry. The $7,586M CSE goodwill sits on a ~14% cushion untested publicly since July 2025, against a Riello transaction that cleared at 8.3× EBITDA and required a write-down. Capital intensity is up 53% with no FCF benefit; distributions run at ~115–165% of FCF funded with commercial paper; 37.98M Viessmann shares (~$2.4B) are freely saleable at a demonstrated $62.01; the comp plan contains no return-on-capital metric of any kind; and no insider has bought a share since a 21% drawdown, with the window open.

11.4 The 3–5 assumptions that actually matter

1. Price/cost turns decisively positive in 2H-FY2026. (Bull-critical, and the single load-bearing assumption in the guide.) Everything in the ~16.0% required 2H margin rests here. Management has already conceded that tariff-offsetting pricing is “slightly margin-dilutive” and has guided Q3 CSA margin down to ~22%.
2. The residential recovery is demand, not channel. (Both-critical.) AHRI June +22% against a -17% TTM average, a flat 7.0–7.5M unit market, inventory -25%, and a disclosed July 1 price increase all point to restock-plus-pre-buy. Management concedes half the 2H residential number is the absence of destocking.
3. The data-center backlog converts at acceptable incremental margins. (Bear-critical.) ~75% of $2.0B must land in 2H at mid-to-high-teens margins inside a 20–21% segment, while seven competitors add capacity. AAON and Modine are the live counter-examples.
4. CSE survives the July-1 goodwill test and the mid-teens margin target is credible. (Tail risk with a date.) A ~14% cushion is thin; the test result surfaces in the Q3 10-Q in late October.
5. The pricing failure was cyclical, not structural. (The whole thesis.) If four oligopolists genuinely cannot pass through input costs, every HVAC margin assumption for the next decade is too high — and Carrier, with the lowest returns on capital of the group, has the least cushion.

11.5 Falsifying evidence — what breaks each side

• Falsifies the BULL: Q3 misses the ~16.5% adjusted operating margin guide; or CSA price realization stays near 3% once the tariff timing has washed through; or Q4 comes in below ~$829M of adjusted operating profit and the FY guide is cut; or the CSE goodwill is impaired. Any one of these converts “timing quarter” into “structural reset.”
• Falsifies the BEAR: two consecutive quarters of CSA price realization re-accelerating toward the announced 8% with segment margin recovering above 25%; and/or the July-1 test showing a widening CSE cushion on improving European fundamentals; and/or data-center revenue landing the 2H ramp without segment-margin dilution. The tariff-cut mechanism, the zero-IEEPA-refund distortion, and the 22% rise in continuing FCF are the bear’s hardest rebuttals and are all genuinely favourable to Carrier.

Verdict: The variant perception has inverted since June. Then, the debate was about timing — would the resi inflection arrive, and was the market paying too much in advance for it? Now the inflection has arrived and the debate is about quality — what is a dollar of Carrier’s earnings worth when the company cannot price, earns 6% on capital, and is growing into a lower-margin mix? Consensus is still positioned for the timing story (Buy, $78 median, estimates rising, 2.66% short interest, 89% institutional). The market’s price action has already adopted the quality story; the analyst community has not. That gap — between a tape that de-rated 13.6% and an estimate sheet that went up — is where the variant perception now lives, and it is unusually well-defined: the Q3 print in late October carries both the margin test and the goodwill test in the same disclosure.

12. Fact vs. Interpretation

13. Open Questions

1. What is the actual dollar tariff exposure, and how much is structural? Carrier has never disclosed a dollar figure; the only precedent is FY2025’s “~$200M of incremental product pricing actions.” Without it, the 2H price/cost bridge cannot be independently verified.
2. What is the IEEPA refund Carrier is entitled to but has not booked? Lennox recognized $30M and Daikin ~¥10.5B. Carrier’s share is unquantified and sits outside the guide as an unrecognized contingent asset.
3. Will the July-1, 2026 goodwill test show the CSE cushion widening or narrowing? The result lands in the Q3-FY2026 10-Q (~late October 2026). This is the single most informative scheduled disclosure on the calendar.
4. What did Carrier receive for NORESCO, and what did it pay for 75F? Both undisclosed by both parties, with no 8-K for 75F. Together they materially affect the 2H revenue base and the portfolio tally.
5. Where are the Viessmann synergies? €200M was promised with “the vast majority by year three” (2026). No quantified scorecard appears in any SEC filing, and CSE’s SG&A load is unchanged at ~21% of sales after $251M of restructuring.
6. Is the residential recovery demand or channel? AHRI June +22% against a -17% TTM average and a disclosed July 1 price increase suggests restock plus pre-buy. Q3 sell-in data will separate them.
7. What incremental margin is the >$8B backlog actually carrying? Management gives a mid-to-high-teens range for CSA commercial but has not disclosed data-center-specific margins or the backlog’s blended conversion.
8. Does the mid-teens CSE margin target have a date? An ~800bp repair from 7.2% was announced with no milestone, by a president three weeks into the role.
9. How large is the antitrust exposure? Nothing is accrued; the class period runs from 2020; treble damages are sought. Unquantifiable at this stage but explicitly unreserved.
10. Will the Viessmann family sell the remaining 37.98M shares, and at what pace? The demonstrated clearing price is $62.01 against $62.78 today.

14. What Must Be True

14.1 Bull case — what must be true, with a falsification test per item

• Price/cost turns decisively positive in 2H-FY2026 and CSA realization re-accelerates toward the announced 8%. Falsification: Q3 realized CSA price stays near 3–4% once the June tariff cut has fully washed through, or Q3 adjusted operating margin misses the ~16.5% guide.
• Q4-FY2026 delivers adjusted operating profit of at least ~$829M (+82% YoY) at a 15.5%+ margin. Falsification: Q4 prints below ~$800M, or the FY2026 guide is cut at the Q3 report in late October.
• The residential recovery is end-demand, not channel restock. Falsification: AHRI sell-in decelerates sharply in Q3–Q4 after the July 1 price increase, or Carrier’s resi organic turns negative on the two-year stack while the end market stays flat at 7.0–7.5M units.
• The data-center backlog converts without diluting CSA segment margin. Falsification: CSA segment margin fails to recover above ~22% in 2H while data-center revenue ramps to its $2.5B exit rate — i.e. the mix drag proves permanent, as it did at AAON and Modine.
• CSE stabilizes, avoids impairment, and progresses toward mid-teens margins. Falsification: a CSE goodwill impairment in the Q3-FY2026 10-Q, or CSE margin failing to exceed 9% in FY2026 despite a European market growing 17%+.

14.2 Bear case — what must be true, with a falsification test per item

• The pricing failure is structural, not a tariff-timing artifact — Carrier cannot pass through input costs without losing share. Falsification: two consecutive quarters of CSA price realization at or above 6% with segment margin recovering above 25%.
• Returns on capital stay below the cost of capital, so growth destroys value. Falsification: honest (amortization- and restructuring-inclusive) ROIC clearing ~9% on a sustained basis — which on this report’s arithmetic requires roughly a full guide delivery plus another ~50bp of margin beyond it.
• Growth is structurally mix-dilutive — commercial and data-center revenue permanently lowers blended margin. Falsification: consolidated adjusted operating margin expanding YoY in a quarter where data-center revenue also grows double-digits, demonstrating the mix drag is absorbable.
• The $14.2B Viessmann deal destroyed value and the CSE carrying value is overstated. Falsification: the FY2026 annual test shows a widening cushion, and CSE margin progresses toward low double digits — removing both the impairment overhang and the mis-underwriting claim.
• Carrier is fully valued: the SOTP does not clear the market price and the market carries CSE at or above cost. Falsification: a de-rate that takes forward EV/EBITDA below Lennox’s ~14.2×, or delivery of the guide plus evidence of durable pricing, which would make ~15.9× demonstrably cheap on forward numbers.

15. Source Appendix

Full itemized citations with URLs and a data-integrity note appear in Appendix B — Source Appendix below. Primary sources of record: CARR Form 10-Q for the quarter ended 2026-06-30 (filed 2026-07-28) and its Exhibit 99 earnings release; the FY2025 Form 10-K (filed 2026-02-05); the Q3-FY2025 and Q1-FY2026 Forms 10-Q; the 2026 DEF 14A; the Form 3/4/5 corpus (189 Form 4s, five years); Forms 8-K dated 2026-07-24 and 2026-07-28; the Q2-FY2026 earnings call transcript (2026-07-28); and the February and July 2026 shelf registrations. The complete five-year SEC filing corpus (312 documents) was enumerated and reviewed. Market, price and factor data accessed 2026-08-15 (AZI price CSVs; FactorsToday; ROIC.ai). Industry data from AHRI, EHPA, ACT Research, USTR/Federal Register and COMEX as cited in-line. Peer financials come from the Trane, Lennox, Johnson Controls, Daikin, AAON and Modine filings and releases — with one third-party claim (“Carrier’s residential orders fell ~60%”) excluded as uncorroborable against Carrier’s own disclosures.

Sections 1–15 above carry no investment recommendation and no price target; the Claude's Take block near the top is a labeled exception representing the author’s own subjective view. Nothing here is investment advice.

APPENDIX A — Standard Diligence Questionnaire

Carrier Global Corporation (NYSE: CARR)

Prepared 2026-08-15 as a follow-up update to the 2026-06-14 questionnaire. All figures reconcile to primary filings — Q2-FY2026 Form 10-Q (period 2026-06-30, filed 2026-07-28) and its Exhibit 99 earnings release, the FY2025 Form 10-K (filed 2026-02-05), the Q3-FY2025 10-Q, the 2026 DEF 14A (filed 2026-03-03), and the five-year Form 3/4/5 corpus — unless flagged. GAAP is shown alongside every adjusted figure. Continuing-operations basis throughout. Stock-based compensation and recurring acquired-intangible amortization are treated as real expenses. Claims are labeled FACT / INTERPRETATION / ASSUMPTION / OPEN QUESTION. Management commentary is treated as a hypothesis, not evidence. Price reference $62.78 (close 2026-08-14). No recommendation; no price target — valuation appears only as embedded expectations. Where an answer has moved since June 14, it is flagged CHANGED. This questionnaire is supplemental to the main analysis and does not substitute for it.

General — What thoughtful questions have other investors asked about this company?

The Q2-FY2026 call (2026-07-28) is unusually useful here, because the analysts on it asked the right questions and, in three cases, went on to set the post-print marks. Six questions are genuinely live (INTERPRETATION, synthesized from the call and the news sweep):

1. “Why did the margin fall in a quarter you grew?” The first question of the call, from Scott Davis: “it really is the margin decline issue… the mix versus price/cost.” Management’s answer was tariff timing — Section 232 landed in early April, mitigation pricing at the end of the month, so Q2 carried a net-negative price/cost gap. This report credits that mechanism (see Industry Dynamics and Competitive Position) as genuinely self-liquidating. But it does not fully answer the question: all four segment margins fell, CSA took +$120M of sales and lost $56M of profit (a -47% incremental margin), and consolidated revenue rose $238M while adjusted operating profit fell $71M. The unanswered residue is mix and price realization, not tariffs.

2. “Why does $1 billion of extra revenue only produce $0.10 of extra EPS?” The sharpest arithmetic on the call, and probably the proximate cause of the selloff. Goris: “Compared to the prior guide, we’re raising $1 billion in sales and $0.10 of EPS. The reason why there is not a stronger… bigger fall-through is the investments… roughly in the $100 million range.” That is roughly an 11% incremental margin at a company whose own stated framework is “25% to 30% conversion for the total company.” Goris then supplied the structural qualifier: “If all of our growth comes from commercial, clearly the incrementals will be somewhat less.” Investors heard a raise funded by mix they do not want to pay 20× for.

3. “Where is the residential recovery on a two-year stack?” Christopher Snyder (Morgan Stanley): “up 20%, but they are comping down 30% and 40%. So it’s… just not showing improvement on a 2-year stack.” Gitlin’s answer was anecdotal — dealer inbound call volumes, pent-up demand, homeowners “getting a little bit more comfortable being uncomfortable with higher mortgage rates” — conceding the anecdotes support only “movement up mid-single digits.” Management’s own bridge concedes the rest: of a guided +20% second-half residential number, “10 of that will come from the absence of destocking.” Snyder carries the street-low $60 target (FACT, market data; not my view).

4. “Are hyperscalers double-ordering, and do the long-term agreements reprice?” Varun Govindaraj cited “chatter from some of your peers about customers pushing out delivery and potential delays” and asked about double-ordering in chillers. Gitlin denied both, saying customers are “pushing us to accelerate deliveries.” OPEN QUESTION: neither Carrier nor Trane discloses whether hyperscaler master agreements are fixed-price or repricing — the difference between a switching cost and a margin liability on a >$2B data-center backlog.

5. “Did Carrier overpay for Viessmann, and is the goodwill impaired?” Carried forward from June, now materially sharper: there is finally an external mark. Ariston disclosed that Riello changed hands at ~0.72× sales / 8.3× EBITDA, against the ~3.0× sales / ~17.1× unsynergized EBITDA Carrier paid for VCS — and Carrier booked a $46M impairment to clear it, i.e. sold below carrying value. The $7,586M of CSE goodwill is fully intact (the 1H decline was 99% FX; Riello carried only $3M), sits on the ~14% cushion PwC flagged as a Critical Audit Matter, and is untested in public since 2025-07-01. The annual test date is July 1, so the answer arrives in the Q3-FY2026 10-Q, ~late October 2026.

6. “Is the adjusted number itself clean?” A question the sell-side has not asked and should have. The 1H adjusted effective tax rate was 15.1% against 22.1%; the $99M Swiss valuation-allowance release and $18M state audit settlement were not excluded from adjusted EPS, worth $0.12–0.16/share. Tax-normalized, 1H adjusted EPS fell 17–19%, not the reported 9%, and the guide is worth ~$2.74–2.78, not $2.90.

INTERPRETATION: the smartest bear question is no longer “will the cycle turn?” — it turned — but “what is a dollar of Carrier’s earnings worth when the company converts revenue at 11% and cannot hold announced price?” The smartest bull question is whether the June-8 tariff cut plus LIFO fully explains the quarter, in which case a >$8B backlog and an easy Q4 comp do the rest.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Near a low on margin, but the volume trough is behind — and CHANGED: less of the depression is cyclical than June assumed. FACT: TTM (to 2026-06-30) adjusted operating margin is 13.42% against 15.1% in FY2025 and 15.8% in FY2024; TTM adjusted EPS of $2.45 is below FY2025’s $2.59; TTM GAAP operating margin is 7.79%. Every profitability line is beneath the FY2025 baseline the June report used. But FACT: Q2 organic sales were +3%, the first positive quarter in two years, with CSA residential +9% and light commercial +10%. INTERPRETATION: in June the depressed margin was attributed to volume deleverage, which is a cyclical explanation with a cyclical cure. Q2 ran the test — volume came back and all four segment margins still fell (CSA 27.0→24.4%, CSE 7.9→7.2%, CSAME 15.3→11.8%, CST 17.6→16.0%). So a material portion of the margin depression is mix and pricing, not the cycle. Earnings are cyclically depressed and structurally lower-quality than the June answer implied.

Driven by the external environment or internal actions? Both, and CHANGED: more internal than June credited. External: the tariff regime moved twice in one quarter (Section 232 expanded 2026-04-02/06, then cut on HVAC from 25% to 15% effective 2026-06-08, sunsetting 2027-12-31, with commercial unrelieved); copper hit an all-time COMEX high of $6.77/lb on 2026-08-07; the German 70% BEG subsidy survived to 2029 while Chinese OEMs grew 13–50% in Europe; China industry shipments fell 5.6% with Xiaomi pricing ~7% below market. Internal: Carrier chose to pre-commit to its channel that it would hand price back if tariffs fell, and did — with high-cost inventory still in COGS under LIFO. It chose share over price: Gitlin, “we want to at least maintain share. So that’s kind of a balancing act.” It chose to raise capex ~53% to ~$600M without raising the FCF guide. And the portfolio churn (one $14.2B acquisition, five divestitures, a four-segment recast) is entirely self-inflicted noise in the comparatives.

How stable are revenues? More visible than in June, no more recurring. FACT: backlog is now >$8B (~68% commercial, ~40% of that data centers ≈ $2.2B), up ~40% YoY and +20% sequentially — genuine new forward visibility, and the single best stability datapoint in the file. Against that: FACT: Q2 service sales were $717M (+12.7%) versus product +2.9%, but Carrier’s service gross margin is 24.4%, below its product gross margin of 27.6% — where Trane’s are 43% versus 32.6%. The management-defined “28% of sales is aftermarket” figure is unreconciled to any audited line, and the one auditable aftermarket-adjacent metric is dilutive. Organically, Carrier has produced roughly nothing across three years (FY2025 -1% organic; 1H-FY2026 +1%). INTERPRETATION: the recurring-revenue ballast is real (a ~110M-unit US installed base on a 15–20-year replacement life) but is smaller and lower-margin at Carrier than the narrative implies, and it is not what stabilized the top line this year — backlog is.

Outlook for products/services? Bifurcated and mix-adverse. Growing: commercial HVAC (orders +~65%), data-center cooling ($1.5B → $2.0B FY2026 guide, $2.5B exit rate), aftermarket, European heat pumps by volume. Flat or shrinking: US residential replacement (market ~7.0–7.5M units, “largely stable”), CSA commercial deliveries (-8% in Q2 on timing), global truck and trailer (down low teens, ACT data implying a 2027 turn), China RLC (-13/14%). INTERPRETATION — the important part: everything growing carries a lower margin than what it is growing away from. Goris put CSA commercial margins at mid-to-high teens against a 20–21% segment average, and guided Q3 CSA down to ~22% from 24.4% explicitly on commercial mix.

How big will this market be — growing, shrinking, domestic or international? Global HVAC equipment TAM ~$300–320B growing mid-single-digit; fragmented globally (Daikin ~15% top share), concentrated regionally. Data-center cooling ~$21B, with Carrier at ~9–10% share and underwriting a new plant on reaching 15–20%. ~52% of sales are international. CHANGED: the European heat-pump market premise Carrier bought on has been falsified — the 2023 deal materials said the segment was “expected to triple to $15 billion by 2027.” It did not: EHPA units grew +13% in 2025 (Germany +50%) and +17% in Q1-2026 — a healthy market, but a fraction of the underwritten trajectory, and one delivering volume without price.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More — the central CHANGED finding of the update. June 2026 delivered a near-controlled experiment: one exogenous input-cost shock applied to every major at once. FACT: four of the five HVAC majors lost margin. Carrier -190bp adjusted; Trane -203bp gross (its first decline in six years, on a record backlog and 123% book-to-bill); Lennox -222bp operating ex a one-time $30M IEEPA refund on organic growth of +3%, identical to Carrier’s, with its CEO “surprised by the price competition”; Daikin -0.8pt. Only Johnson Controls expanded (+243bp) — and JCI sold its residential business to Bosch. INTERPRETATION: the differentiator was volume leverage, not pricing. Trane took a larger gross-margin hit than Carrier yet held its operating decline to 99bp because +10.6% volume absorbed it. Nobody priced through the shock. In Greenwald’s terms, the 2022–25 margin expansion the June report read as dealer-channel captivity was a shared, transient, regulation-enabled episode (SEER2, the A2L conversion, the 2024 R-410A pre-buy, post-COVID pass-through) — the cooperative arm of a prisoner’s dilemma, not a barrier to entry.

How profitable is the business (ROIC, ROE)? Marginal, and deteriorated since June. FACT/COMPUTATION on average invested capital of $24,370M (TTM to 2026-06-30, 22% normalized tax):

Peers (FY2025 after-tax ROIC): Trane 25.1%, Lennox 29.1%, Johnson Controls 9.6%. INTERPRETATION: Greenwald’s diagnostic bands read 15–25%+ as advantages present and 6–8% as advantages absent. Carrier scores ~6–7%. It does not earn a ~9% WACC [ASSUMPTION] on any honest basis, and even a fully-delivered FY2026 guide reaches only ~8.6%. The June framing — “a franchise-grade asset inside a non-franchise capital structure” — is still the honest reconciliation, but it is harder to sustain now that the segment margins are compressing toward the consolidated reality rather than the reverse (total segment margin 15.3% vs 18.8% on the half).

How profitable is the industry — how many competitors, what barriers to entry? Five credible North American players; the profit pools genuinely differ by arena. Real barriers: transport refrigeration (Carrier Transicold vs Thermo King, a decades-stable duopoly, ~11% of sales) and NA residential two-step distribution. No barriers in the growth arena: management’s own share math for US commercial HVAC — from ~10% to “15% to 20%” — describes, on Greenwald’s own >5-point threshold, a barrier-free market. Seven well-capitalized competitors are adding capacity into data-center cooling simultaneously (Trane quadrupling applied capacity, nVent’s third liquid-cooling expansion, AAON capex +101%, Eaton’s $9.55B Boyd deal, Schneider >$700M, Vertiv at 2.5× capex). Marathon’s capital-cycle lens is unambiguous: margins are already compressing while demand is still accelerating — AAON cut full-year gross-margin guidance 200–300bp with ROIC falling 24.2%→9.3%, and Modine grew data-center sales +90% with a 960bp gross-margin decline. That is what the top of a capital cycle looks like from the inside.

Can the business be easily understood? The business, yes; the financial statements, no. It sells HVAC equipment, controls and service through direct, dealer/distributor and JV channels. The complications are all accounting: adjusted operating profit now sits 56% above GAAP (the widest post-spin wedge), add-backs grew 26% YoY to $606M in 1H, reported free cash flow is computed off total operating cash flow including discontinued operations contrary to the company’s own published definition, and three years of portfolio surgery plus a four-segment recast make year-over-year comparisons non-comparable without normalization.

Can it be undermined by foreign low-cost labor? Yes, and the threat is larger than the June answer allowed. The credible vectors: Daikin (Goodman/Amana) with a US low-cost position, whose North American unit grew +13% in local currency with segment margin up 2pts to 11% in the same quarter Carrier lost 260bp in CSA; Chinese OEMs growing 13–50% in Europe into a CSE segment earning 7.2%; and Xiaomi pricing ~7% below market in China. The defense is not technology — R&D is ~2.9% of sales (1H $148M vs $161M), parity-maintenance not IP leadership. The defense is the dealer channel, and June 2026 established that the channel delivers access, volume and share — not price.

Do brands matter? At the channel level, and less than June assumed. CHANGED, with a clean test case: Viessmann is the premium brand in European heating. It is now running “pricing promotions” (the phrase appears twice in the 10-Q, only for CSE), launching a cheaper Vitocal 200 value tier, and earning a 7.2% margin into a market growing 17%. A brand that must discount and trade down in a rising market is not commanding a premium. The Carrier/Bryant/Day & Night/Heil brand ladder still confers pricing power through the trained, stocked dealer — but the homeowner has essentially no loyalty, and the dealer’s loyalty was demonstrably purchasable this quarter.

What is the nature of competition? Scale, breadth, distribution and installed-base aftermarket. Applied/commercial is spec-and-bid with service lock-in; residential is distributor/contractor-intermediated and price-shopped; European residential heating is subsidy-mediated and fragmented; China is a price war. One double-edged datapoint: Carrier is a named defendant in consolidated HVAC price-fixing class actions (E.D. Mich., Berg v. Bosch), with amended consolidated complaints filed 2026-07-10, naming seven firms said to control >90% of the US market, seeking treble damages over a class period from 2020-01-01 — the same window this update concludes was a shared, regulation-enabled pricing episode. Both readings of that window are unflattering.

Customers’ switching costs? Real and narrow. High in applied/building controls (Automated Logic) and transport-refrigeration fleets — re-qualifying a Carrier Transicold container fleet on Thermo King is costly, which is why CST was the one segment with no pricing commentary anywhere in its disclosure in a quarter when every contested arena failed to price. The best new quantified captivity metric is ~55,000 channel-partner technicians monitoring systems in real time, +35% YoY — a technician whose diagnostics and service workflow runs on Carrier’s platform faces a genuine, rising switching cost. INTERPRETATION: this is a real and improving distribution asset, but it is an input metric. It has not yet appeared as attach rate, service margin or price realization — and the aftermarket growth it drives is currently gross-margin dilutive.

Moat verdict (INTERPRETATION) — CHANGED, DOWNGRADED to weak/narrow. What Carrier genuinely has: one textbook Greenwald franchise (transport refrigeration, ~11% of sales); a distribution/access advantage in the NA dealer channel that wins volume and share but demonstrably not price; and real procurement scale. What it does not have: pricing power that survives a cost shock, any defensible position in Europe or China, a margin-accretive aftermarket, or any barrier whatsoever in the segment generating all of its growth. A 7% after-tax ROIC against peers at 25–29% is not a moat that failed to show up in the numbers; it is the numbers reporting that the moat is largely industry-structural and shared, not company-specific and owned.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Three (INTERPRETATION unless noted). (1) The independent-dealer network and installed-base annuity — the actual economic advantage — is capitalized only indirectly through Viessmann’s acquired customer-relationship intangibles; the organic Carrier-brand channel carries no balance-sheet value. (2) FACT: up to ~$2.4B of AFFF/Kidde-Fenwal insurance-recovery rights that Carrier has chosen not to book — genuine conservatism. (3) NEW: FACT — Carrier booked no IEEPA tariff refund after the Supreme Court struck the IEEPA tariffs 6–3 on 2026-02-20, while Lennox recognized $30M and Daikin ~¥10.5B. That is an unrecognized contingent asset sitting outside the guide, and it also means every peer margin comparison this quarter flattered the peers on identical facts.

Off-balance-sheet liabilities? FACT: the AFFF/PFAS tail is the material one and it grew — defendant count moved from “>17,000” (FY2025 10-K) to “>18,000” at 2026-06-30. The structured settlement is $615M of cash over five years plus ~$115M of net KFI proceeds, largely insurance-backstopped; the court-approval process for the Direct Claims Settlements begins in Q3-2026 — a real near-term catalyst. For non-settling claimants Carrier still states it “is unable to assess the probability of liability or to reasonably estimate a range of possible loss.” Cash drag ~$100–120M/yr against ~$1.8B of FCF: manageable. NEW: the antitrust class actions carry no accrual at all, with treble damages sought over a six-year class period. Also standing: the residual UTC/RTX Tax Matters Agreement, operating leases, and the Carrier Enterprise JV with Watsco (through which Carrier cedes part of the most defensive distribution profit pool rather than owning it).

How conservative is the accounting? Mixed, leaning aggressive on the items that determine the multiple — and CHANGED for the worse. Aggressive: (i) the adjusted framework excludes ~$850M/yr of economically real acquired-intangible amortization that recurs through 2030 ($868M scheduled for 2026, $818M 2027, $742M 2028, $655M 2029, $599M 2030); (ii) restructuring is added back and more than doubled to $116M in 1H (from $55M), of which $89M is CSE; (iii) NEW and material — the adjusted ETR of 15.1% versus 22.1%, with the $99M Swiss valuation-allowance release and $18M state settlement not excluded from adjusted EPS, worth $0.12–0.16/share; (iv) the CSE goodwill is carried at a ~14% cushion untested in public since 2025-07-01; (v) reported free cash flow is computed off total operating cash flow including discontinued operations, contrary to Carrier’s own published definition — a disclosure-quality demerit that, in this instance, happened to understate performance. Conservative: no AFFF insurance asset booked; SBC is immaterial at $33M in 1H (~0.28% of sales), so adjusted earnings are not SBC-inflated; and the $46M Riello write-down was taken promptly rather than deferred. FACT: no Item 4.01/4.02 8-K, no restatement, no material weakness, and no change in accounting policy in the period; the IRANNOTICE filed 2026-07-28 is the routine ITRA Section 219 disclosure. One-time items to normalize out before any run-rate work: the $46M non-deductible Riello impairment, the elevated $116M restructuring, the $426M of 1H acquired-intangible amortization, the 9.4% GAAP ETR, and ~$0.05 of FY2026 EPS drag from the NORESCO exit and new-plant start-up costs.

How CapEx-hungry is the business? CHANGED — this is no longer the unambiguous strength it was in June. FACT: 1H capex was $211M vs $144M (+47%), and FY2026 capex guidance was raised to ~$600M, +53% versus FY2025’s $392M, taking capex/sales from 1.80% to ~2.61% — for a new US data-center-chiller plant (Texas or Alabama, targeted to run by end-Q1-2027, including vertical integration into compressors) plus a new India facility. FCF guidance was not raised alongside the profit raise, because, in Goris’s words, the higher earnings are offset by the ~$100M capex increase. INTERPRETATION: if data-center revenue really runs $2.0B (2026) → a $2.5B exit rate → higher in 2027–29, this is the start of a multi-year step-up toward normal industrial intensity of 2.5–3.0% of sales, not a one-off. The “asset-light compounder” description that supported a premium multiple is now materially qualified, and the increment is being spent to chase share in the lowest-margin part of the portfolio.

Tangible capital and leverage (FACT). Net debt $10,608M (up $330M in the half despite $795M of reported FCF, because of $748M of buybacks, $400M of common dividends, $65M of NCI dividends and $54M of acquisitions). Net debt / TTM GAAP EBITDA 3.53× (from 2.55×) and 3.14× on adjusted EBITDA — driven by a -27% TTM EBITDA decline, not by incremental borrowing; ~3.02× pro-forma for Riello proceeds, and ~2.6× forward if the guide lands. Debt shifted short: commercial paper $325M → $685M at 4.01%, and the $900M 2.493% notes due Feb-2027 reclassified current (short-term borrowings + current LTD $468M → $1,638M). The $2.5B revolver (Dec-2029) is undrawn and all covenants are met. Net interest expense is rising on a flat balance: 1H $195M vs $173M. Tangible book value is -$7,876M, or -$9.55/share (parent equity $13,147M less $15,267M goodwill and $5,756M intangibles) — marginally improved from -$9.66 only because amortization is running the intangible balance down, not because equity is building. INTERPRETATION: investment-grade and not distressed, but leverage is rising for the worst possible reason — a collapsing denominator — and there is no asset floor under the equity.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, and what is the philosophy? FACT and CHANGED in Carrier’s favour: on the correct continuing-operations basis, 1H-FY2026 FCF rose 22% to $742M from $608M, with working capital consuming $381M less cash than a year earlier; the headline “-20%” is an artifact of the discontinued-ops contamination noted above. TTM continuing FCF is $1,831M; FY2026 guidance is ~$2B, unchanged. Uses: 1H distributions of $1,210M against $795M of reported FCF, with the gap plugged by $361M of new commercial paper; since YE2024, $4.81B returned against $2.92B of FCF — a $1.89B shortfall that substantially accounts for the $2.22B rise in net debt. INTERPRETATION: the philosophy since the 2020 spin has been aggressive portfolio reshaping funded by disposals and debt, with distributions run at ~115–165% of free cash flow. That is a financing choice, not a return of surplus capital.

Significant acquisitions recently? FACT: Viessmann Climate Solutions, $14.2B, closed 2024-01-02 — the decision that governs the entire returns profile — plus 75F (building-management software, closed 2026-07-23, price undisclosed, no 8-K filed; Gitlin claims it “expands our TAM by about $20 billion”; pressed by Scott Davis on what was actually bought, he answered “Technology primarily” — ~91 engineers). Against the 2023 underwriting: €0.7B of EBITDA on €4B of sales, €200M of synergies with “the vast majority by year three” (i.e. 2026), a European heat-pump market “expected to triple to $15 billion by 2027,” and a “return to pre-transaction leverage within approximately two years.” FACT as of Q2-FY2026: CSE runs a 7.0% margin; its SG&A load is 21.1% of sales versus 21.3% a year earlier — essentially unchanged after $251M of cumulative restructuring, 62% of all segment restructuring; CSE gross margin is a healthy 29.7%, meaning the cost programme was aimed at the wrong line; leverage is 3.53× GAAP, above pre-transaction levels; and no quantified synergy scorecard appears in any SEC filing. The integration leader of nine years, Thomas Heim, was replaced on 2026-06-17 by external hire Thomas Donato (ex-Bosch Power Tools), with no 8-K filed, and Gitlin now says “we should have been more aggressive on cost” with a mid-teens margin target “over the next few years” and no dated milestone — an ~800bp repair from a 7.2% base. On the sell side: Riello closed 2026-07-01 at ~0.72× sales / 8.3× EBITDA with a $46M impairment (sold below carrying value), and NORESCO closed 2026-08-03 (price undisclosed by both parties; took ~$143M of project debt with it). Running tally since 2023: ~$14.2B acquired versus ~$10.42B divested, net ~-$3.8B. INTERPRETATION: this is the textbook Marathon top-of-cycle asset-growth red flag, and the direction of travel is now a retreat from the geography Carrier spent $14.2B to enter two years ago.

Buying back shares? Yes, and CHANGED — the execution has been genuinely good, correcting the June working assumption. FACT: 1H repurchases were 12.0M shares at an average of $62.33, against a 2026-08-14 close of $62.78 — essentially breakeven, not under water. The pattern within the half is deliberately opportunistic: 2.38M shares in April at $60.52, stepping down to 643k in June at $72.00. Program-to-date average is $59.24, ~6% in the money. Diluted weighted-average shares fell to 836.5M from 866.3M (-3.4%). ~$752M of buyback remains against the $1.5B FY plan; $4.6B remains under the $12.1B cumulative authorization. INTERPRETATION: the criticism is not the price paid — it is that the buyback is debt-funded while margins compress, EBITDA shrinks and capital intensity rises 53%.

Issuing large amounts of new shares to insiders? No (FACT). SBC is $33M in 1H, ~0.28% of sales — refreshingly small for an industrial, and treated as a real expense. The only large issuance was the 58.6M shares of Viessmann deal consideration, not a comp grant. The S-3ASR filed 2026-07-28 is a routine 715,279-share LTIP reoffer, not a capital raise. The live share-supply issue is not insider issuance but the 37.98M Viessmann family shares (~4.6%, ~$2.4B) that remain freely saleable, registered at Carrier’s expense under the February 2026 shelf, with 12.1M already sold at $62.01 in May 2026.

Compensation policy of directors/management? Formulaic, responsive in practice, and structurally tolerant of value-destructive growth — verified and hardened since June. FACT, by full-text sweep: the 2026 DEF 14A contains zero occurrences of “ROIC,” “return on invested capital,” “return on equity,” “economic profit,” or “EVA.” The proxy’s own Financial Performance Measures list is complete: Relative TSR, Adjusted EPS, Sales, Adjusted Operating Profit, Free Cash Flow. Annual bonus = three equally-weighted metrics (Sales / Adj. Op. Profit / FCF); the 2025 result was Sales 67.0%, Adjusted Operating Profit 50.0%, FCF 0.0%, giving a Company Performance Factor of 39.0% with no discretionary adjustment — the plan genuinely bit. LTI = 50% SARs / 50% PSUs, PSUs split 50/50 between 3-year Adjusted-EPS CAGR and Relative TSR; the 2023-vintage PSUs vested at 71.1%, Relative TSR below the 25th percentile. Governance: single class of stock, combined Chairman/CEO with a Lead Independent Director (John Greisch), board now 11 with 9 independent after Neil Barua (PTC CEO) joined the Compensation and Technology & Innovation Committees on 2026-07-24; Max Viessmann is the one non-independent outside director; CEO ownership guideline 6× salary. INTERPRETATION — the structural finding: all four company financial metrics are mechanically increased by a debt-funded, goodwill-heavy acquisition. Sales rises with purchased revenue; Adjusted Operating Profit and Adjusted EPS rise by construction because “adjusted” adds back the very amortization ($426M in 1H alone) and restructuring ($116M) the deal generates; FCF rises with acquired cash flow. Not one metric is reduced by the $14.2B of capital consumed. Relative TSR — 25% of LTI — is the only channel through which a value-destructive deal reaches a wallet, and it is a three-year market outcome, not a capital-discipline input. Not an accusation of bad faith: simply that management was paid to do exactly what it did.

Motivations of management (INTERPRETATION). Gitlin is incentivized on adjusted-EPS growth, absolute stock price and relative TSR — aligned with the recovery consensus underwrites. The strongest bullish tell in the file remains FACT: across 189 Form 4 filings over five years there is exactly one open-market purchase (code P) — Gitlin, 19,300 shares at $52.6203 on 2025-11-25, near the 52-week low, now ~19% in the money. But it decays: it was made nine months ago, ~16% below today’s price, on a different understanding of the margin structure. FACT: no insider has purchased since the July collapse. The most recent Form 4 of any kind is 2026-07-24, four days before the print; with a two-business-day filing deadline, that rules out any insider transaction through ~2026-08-13. Across a -13.6% two-day repricing and eleven sessions since, not one director or officer has bought a share, with the window open. Absence of a buy is not a sell signal — but it is the absence of the one signal that would have been genuinely bullish, and it stands against a corporate buyback that did keep running.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No (FACT). Carrier is a Delaware C-corporation with a single class of common stock (no dual-class, no super-voting), listed on the NYSE, ~824.3M shares outstanding. Dividends are reported on Form 1099. No ADR, no MLP, no K-1 — none of the tax-structure complications.

Dividend policy? Modest, steady, and — on a total-distribution basis — no longer self-funded (FACT). The quarterly rate has risen every year since initiation ($0.19 → $0.21 → $0.225 → $0.24, most recently declared 2026-06-04, unchanged), a ~$0.96 annual run-rate costing ~$800M/yr. At $62.78 the yield is 1.53%; adding the ~2.9% buyback gives a total shareholder yield of ~4.4–4.9%, against an FCF/EV yield of 2.92%. INTERPRETATION: the dividend alone is comfortably covered; total distributions are running at ~115% of free cash flow (and ~165% in the half), part-financed with commercial paper and disposal proceeds while capex steps up 53%. A yield financed from the balance sheet is a return of capital, not a return on it.

How profitable is the business? See Business Quality above. Headline for FY2026 to date: CSA 24.4% Q2 / 20.4% 1H (the franchise), CSE 7.2% / 7.0% (the $14.2B drag), CSAME 11.8% / 10.8%, CST 16.0% / 15.1%; total segment 18.0% Q2 / 15.3% 1H; consolidated adjusted operating margin 17.2% Q2 / 14.4% 1H against a TTM 13.42%; GAAP operating margin 13.0% Q2 / 9.3% 1H, TTM 7.79%. Honest ROIC ~6%, ROE 8.4%. Strong at one segment, mediocre consolidated — the arithmetic centre of the problem is ~$7.6B of CSE goodwill attached to a 7% margin.

Is net income diverging from cash from operations? Yes, favourably — with three caveats (FACT). TTM continuing FCF of $1,831M against TTM continuing net income to common of ~$1,174M is ~1.56× conversion, because GAAP net income absorbs ~$850M/yr of non-cash acquired-intangible amortization that never touches cash. That is the legitimate core of any premium-multiple argument. Caveats: (i) the conversion flatters precisely because of the amortization — which is economically real (it recurs through 2030 and reflects capital actually spent), so a high FCF/NI ratio here is a symptom of the acquisition, not evidence of business quality; (ii) reported operating cash flow is contaminated by discontinued-operations cash and must be read on a continuing basis; (iii) the capex step-up to ~$600M is now consuming the incremental earnings, which is why the FCF guide did not move when the profit guide did.

Valuation framing (embedded expectations only — NO price target, INTERPRETATION). At $62.78, market cap ~$51.8B and EV ~$62.7B (824.3M shares + $10,608M net debt + $325M NCI), Carrier trades at 18.5× TTM adjusted EBITDA, 15.9× forward if the guide is met, 21.6× forward adjusted EPS — and ~29.8× on amortization-honest FY2026E EPS of ~$2.10–2.13, or ~33.0× on trailing honest EPS of ~$1.90. Peers: Trane 24.1× EBITDA / ~32× amortization-honest; Lennox 14.2× / ~19.5×. Two conclusions matter. (a) CHANGED: Carrier is now ~7% cheaper than Trane on an amortization-honest basis, reversing the June claim — but on trailing honest earnings it is ~33.0× versus ~29–31× in June, i.e. more expensive at a 7% lower price, because the earnings base fell faster than the multiple. (b) The “21.6× forward” that makes it screen cheap is a stack of three optimistic adjustments — accept ~$850M/yr of real amortization as non-economic, accept ~$0.15 of undisclosed tax benefit as recurring, and believe a Q4 requiring +82% to +104% profit growth to a 15.5–17.3% margin against a post-spin Q4 best of 13.2%. A peer-multiple sum-of-the-parts reaches only $45.7–58.7B of EV against a market $62.7B; run in reverse it implies the market carries CSE at ~$11–22B (central ~$16B) — at or above the $14.2B paid — for a 7%-margin segment whose closest transaction comparable cleared at 8.3× EBITDA and required a write-down. Own-history percentiles (AZI valuation_index, 2026-08-14): P/S 74.9th, down from ~90th; composite 60.4, down from 72.3 — a genuine de-rate, still the expensive quartile of Carrier’s own post-spin range. Ignore the P/E percentile (91.3, corrupted by a depressed $1.53 TTM GAAP EPS denominator) and the P/B percentile (14.9, meaningless against -$7.9B of tangible equity). Market structure: institutional ownership 89.1%, short interest 2.66% of float (+14.8% MoM), BlackRock 7.6%, Capital Group 12.8%. Sell-side positioning (FACT, third-party data, not my view): no downgrades after the print, Bernstein raised to $78, Argus raised, Zacks to Rank #1 as estimates rose while the price fell; consensus Buy across 24 analysts, median $78, range $60–90. The de-rate was multiple compression, not a numbers cut.

Risks & Downside

What factors would cause the stock to decline? A tight, correlated cluster rather than independent risks — a pricing failure is precisely what would cause the Q4 miss, so the downside scenarios cluster rather than diversify:

On item 12, deliberately deflationary (INTERPRETATION): no complaint has been filed, no class period is alleged, and no specific misstatement is identified. These releases are advertising, issued mechanically after any sharp post-earnings drop in a large-cap. Informational content ~zero; it would become material only if an actual complaint alleged pre-announcement knowledge of the tariff/price giveback. Watch item only.

Risk of a catastrophic loss? Remote (ASSUMPTION, well-supported). Carrier is investment-grade, generates ~$1.8–2.0B of free cash flow, holds a #1/#2 position in most of its markets, has an undrawn $2.5B revolver and no near-term maturity wall beyond the $900M 2027 notes, and faces no solvency question. The realistic severe case is a compound one — Q4 miss plus a CSE impairment plus a data-center capex pause — which this report’s bear scenario sizes at roughly -17% to -31% from $62.78, i.e. multiple compression and years of sub-cost-of-capital returns, not impairment of the enterprise. One June statement is corrected here (FACT): a CSE goodwill write-down would leave tangible book value arithmetically unchanged, because goodwill is already excluded from tangible book. The damage from an impairment is informational — the auditor’s confirmation that the deal destroyed value — estimated in the report at ~$10–12/share if it causes the market to re-mark CSE toward the Riello comparable. The genuinely open-ended tails are the unreserved antitrust exposure and a failure of AFFF settlement approval.

Chance of a total loss? Negligible (INTERPRETATION). The impairment risk is non-cash; the litigation tails are early-stage or insurance-backstopped and cost ~$120M/yr against ~$1.8B of FCF; there is no leverage-driven or single-event path to permanent total loss of capital. The genuine risk in this name is price and return-on-capital — paying a compounder multiple for a business earning ~6% honestly — not ruin.

Recent News & Events

Has the business environment changed recently? Yes, materially, on five axes (FACT). (1) The tariff regime changed twice inside one quarter: Section 232 expanded 2026-04-02/06 (50% on steel/iron/copper/aluminium, 25% on derivatives, on full customs value), then cut on HVAC equipment from 25% to 15% effective 2026-06-08 under Proclamation 11032, through 2027-12-31 — with commercial HVAC left unrelieved; separately the Supreme Court struck the IEEPA tariffs 6–3 on 2026-02-20 and the CIT ordered refunds, for which Carrier booked nothing while peers booked $30M and ~¥10.5B. (2) The volume inflection arrived and the pricing power did not — organic +3%, all four segment margins down, adjusted EPS $0.86 vs $0.92, and the stock -13.6% over two sessions on a raised guide. (3) Data centers became the story: guide $1.5B → $2.0B with a $2.5B exit rate, orders +>300%, backlog >$8B. (4) Europe recovered in volume and not in profit — EHPA units +13% in 2025 (Germany +50%) and +17% in Q1-2026, the German 70% BEG subsidy retained to 2029, yet CSE at 7.2% running pricing promotions and launching a cheaper Vitocal 200 against Chinese OEMs growing 13–50%. (5) Input costs remain live: COMEX copper set an all-time high of $6.77/lb on 2026-08-07 with a refined-copper Section 232 determination pending. Also on the tape but low-signal: three plaintiff firms announced securities “investigations” between 2026-07-29 and 2026-08-13, with no complaint, no class period and no alleged misstatement.

Significant acquisitions? 75F, a building-management-software bolt-on, closed 2026-07-23 — price undisclosed, no 8-K, ~91 engineers, with Gitlin claiming a ~$20B TAM expansion and, when pressed, describing the purchase as “technology primarily.” On the disposal side: Riello closed 2026-07-01 (Ariston, ~€289M EV, ~0.72× sales / 8.3× EBITDA, with a $46M impairment) and NORESCO closed 2026-08-03 (Opterra / LS Power, price undisclosed, ~$143M of project debt transferred; the Q2 10-Q still described it as pending). FACT — a finding in the negative: no 8-K was filed for any of the Riello agreement or completion, the NORESCO agreement or completion, the 75F acquisition, or the CSE presidency change. Each was individually immaterial to a ~$63B-EV company; collectively they are the quarter’s story, and an investor tracking 8-Ks alone would have missed the entire European retreat.

Change in accounting policies? No (FACT). No change in accounting policy, no restatement, no material weakness, no Item 4.01/4.02 8-K. The comparability reset was the prior-year event — the May-2025 reorganization into four regional climate segments (CSA / CSE / CSAME / CST) with prior periods recast. What has changed is presentation quality rather than policy: the GAAP-to-adjusted wedge is now the widest post-spin (adjusted operating profit 56% above GAAP), add-backs grew 26% YoY to $606M in 1H, and the reported-FCF definition inconsistency described above persists.

Recent changes — new markets, facilities, management? New markets/products: data-center cooling at scale, now with a claimed ~9–10% share of a ~$21B market and an explicit ambition of 15–20% in US commercial HVAC; building-management software via 75F; the digital service footprint at ~55,000 monitored channel-partner technicians, +35% YoY; and, in Europe, a downward product extension (Vitocal 200 value tier). Facilities: a new US data-center-chiller plant (Texas or Alabama, decision imminent, targeted running by end-Q1-2027, with compressor vertical integration) and a new India facility — the drivers of the +53% capex guide. Management and board: Thomas Heim replaced as CSE president by Thomas Donato on 2026-06-17 (external, ex-Bosch Power Tools; no 8-K — the clearest available admission that the Viessmann integration has not worked); Neil Barua appointed independent director 2026-07-24 to the Compensation and Technology & Innovation Committees, taking the board to 11 with 9 independent. CFO Goris and CEO Gitlin remain in place; the combined Chairman/CEO structure is unchanged.

End of Appendix A. This questionnaire is supplemental to the main analysis in sections 1–15 and does not substitute for it; cross-references point to the corresponding sections above. No recommendation; no price target. Analyst targets, ratings and short-interest figures reported above are third-party market data recorded as facts about positioning, not my views.

APPENDIX B — Source Appendix

SOURCE APPENDIX — Carrier Global Corporation (NYSE: CARR)

Cross-checked against the completed analysis. Primary sources (SEC filings, company disclosures, statutory/agency documents) take precedence over secondary; every non-obvious claim above traces to an entry below. SEC EDGAR CIK 0001783180. This is an update report — a follow-up to the June 14, 2026 report: still-relevant primary sources are carried forward, and the emphasis here is on what is new — the Q2-FY2026 filings and call, the June 2026 Section 232 proclamation, the peer June-quarter prints, and the Riello/NORESCO/75F transactions.

EDGAR and XBRL data accessed 2026-08-15; price, factor and valuation-percentile data as of the 2026-08-14 close ($62.78); the full five-year SEC filing corpus (312 documents) was enumerated and reviewed as of 2026-08-15, with completeness reconciled against the EDGAR submissions history.

Corpus composition (five years to 2026-08-15): 5 × 10-K · 15 × 10-Q · 61 × 8-K (+1 8-K/A) · 189 × Form 4 · 12 × Form 3 · 5 × DEF 14A · 5 × DEFA14A · 3 × S-3ASR · 1 × S-3 · 2 × S-4 · 1 × S-8 · 5 × SD · 4 × ARS · 2 × 8-A12B · 1 × 4-A = 312 documents.

A. SEC Filings & Primary Company Sources

A.1 Periodic reports (10-K / 10-Q)

The remaining 10-Qs and 10-Ks in the five-year corpus (15 and 5 respectively) were swept for trend and one-time items but are not individually cited; the complete list is available from the EDGAR filing index for CIK 0001783180.

A.2 Earnings releases (Exhibit 99 to Form 8-K) — non-GAAP source of record

A.3 Material-event reports (Form 8-K)

A.4 Proxy statement and annual report to shareholders

Prior proxies (DEF 14A 2025-02-25, 2024-03-05 and earlier — five in the corpus) mirrored and swept for the multi-year incentive-metric trend.

A.5 Registration statements (shelf / resale)

Also mirrored, not individually cited: S-4 (2025-07-29), S-8 (2025-07-29), 8-A12B ×2, 4-A, SD ×5 (conflict minerals, most recent 2026-05-29).

A.6 Insider transactions and beneficial-ownership filings

A.7 EDGAR structured data and filing corpus

A.8 Company non-filing primary sources (press releases / IR)

B. Earnings Call Transcripts

Source of record: ROIC.ai transcript data (list_earnings_calls, get_earnings_call_transcript, get_latest_earnings_call). All management commentary is treated as hypothesis, not evidence, and is validated against the filings and the peer/industry data.

C. Industry, Regulatory & Market-Data Sources

Secondary/vendor unless the underlying instrument is a statute, proclamation, court decision or agency release. Vendor market-sizing carries wide dispersion and is treated as directional context, not primary. All accessed 2026-08-15 unless stated.

C.1 Trade policy — Section 232, IEEPA, and the June 2026 HVAC relief

C.2 Shipment, end-market and commodity data

C.3 European heat pumps, subsidy regime and China

C.4 Data-centre cooling — sizing and capital cycle

D. Peer & Comparable-Company Sources

Peer figures used in the Competitive Position margin table, the Valuation comp table and the Industry Dynamics capital-cycle evidence. Peer statement data pulled via ROIC.ai (get_income_statement, get_profitability_ratios, get_enterprise_value, quarterly, June-2026 quarter and prior-year comparatives) and cross-read against the companies’ own June-quarter filings and releases. Where the aggregator and a filing disagree, the filing governs.

E. Quantitative & Market Data Feeds

Authority bind (explicit). Everything in this section is third-party aggregated or modelled data — NOT primary. For a US filer, SEC EDGAR and the 10-K/10-Q are primary and win any conflict; these feeds accelerate and cross-check, they do not replace the filing. Any ratio that drives a verdict in the report was reproduced from the underlying statement line items. No analyst target, aggregator figure or factor estimate is treated as, or converted into, a price target.

F. Litigation Dockets

G. Data-Integrity, Corrections & Exclusions Note

This section records what was wrong, excluded, deflated or still unknowable, so that a reader can audit the difference between this report and its predecessor.

G.1 ROIC.ai enterprise value is period-end-priced — every EV in this report was re-struck

ROIC.ai’s get_enterprise_value values a company off its fiscal-period-end price. For Carrier that is the 2026-06-30 quarter-end close of $73.09, which produces a stored enterprise value of ~$71.9B — struck before the 2026-07-28/29 de-rate and ~16% above the correct figure. Using it would have overstated every EV multiple in the Valuation section by roughly 16%.

Every EV figure in the report is re-struck at the 2026-08-14 close of $62.78, giving market cap ~$51.8B (824.3M shares) and enterprise value ~$62.7B (net debt $10,608M from Carrier’s own reconciliation, plus $325M of non-controlling interests). The same re-striking was applied to Trane, Lennox, Watsco, AAON and Vertiv in the Valuation comp table. ROIC’s balance-sheet items, TTM EBITDA/sales and share-count derivations were used; its stored EV was not. The AZI price CSVs (E1) are the price source of record.

G.2 Four corrections to the June 14, 2026 report

A fifth item — the prior report’s reverse-DCF arithmetic ($1.9B growing at 6% for ten years then 2.5% discounts to ~$39B, not $62–66B) — is corrected inside the Valuation section and rests on the same inputs (E4, A7).

G.3 One claim EXCLUDED as uncorroborable

A claim encountered repeatedly while reviewing Trane’s June-2026 quarter — that “Carrier’s residential orders fell ~60%” — is recorded here because it is widely repeated and, on inspection, unsupported.

It is excluded from this report. It could not be corroborated against any Carrier primary source: it does not appear in the Q2-FY2026 earnings call transcript (B1), the Q2-FY2026 earnings release (A7), or the Q2-FY2026 Form 10-Q (A1). The only residential order figure Carrier disclosed in the quarter is +20%, and it is European. A ~60% decline in North American residential orders would also be flatly inconsistent with the quarter Carrier actually reported — residential sales +9%, total orders +~40%, and a raised full-year guide. Absent an independent primary source, the figure is not carried forward, and no conclusion in this report depends on it.

G.4 One claim DEFLATED — the “securities investigation” notices

Six plaintiffs’-firm releases between 2026-07-29 and 2026-08-13 (SueWallSt ×1, SBS Law ×2, Pomerantz LLP ×3 — F5) announce “investigations” into Carrier. These are attorney advertising, not litigation. The Pomerantz releases state only that the firm is investigating “whether Carrier and certain of its officers and/or directors have engaged in securities fraud or other unlawful business practices,” and cite as the trigger nothing more than the 2026-07-28 release showing a YoY decline in EPS and operating margin followed by the −8.9% session. No complaint has been filed. No class period is alleged. No specific misstatement is identified. No scienter theory is offered. Plaintiff firms issue these mechanically after any sharp post-earnings drop in a large-cap.

They are therefore recorded as a one-line factual entry and sit in the risk matrix at low likelihood / low impact — a watch item only, which would become material only if an actual complaint were filed alleging pre-announcement knowledge of the tariff/price give-back. This is entirely distinct from the genuine HVAC antitrust class actions in section F, which are filed, consolidated, amended, disclosed by Carrier itself in its 10-Q, and carried in the risk matrix at materially higher impact.

G.5 Reconciled discrepancies

G.6 Excluded or caveated third-party figures

• JCI’s ROIC-sourced EBITDA ($3,764M) sits only $153M above its EBIT ($3,611M) — an implausible D&A figure for a ~$25B-revenue building-controls business. The resulting 27.3× EV/EBITDA is unreliable and is excluded from the Valuation comp table rather than reported.
• Lennox’s revised FY2026 guidance could not be sourced (the ROIC news feed returned an empty set for LII). Its June quarter expanded reported margin, so the −21.0% session on 2026-07-29 was almost certainly an outlook cut, but the magnitude is an assumption; the derived LII forward EBITDA multiple is robust across the plausible range (14.2× flat to 15.0× at −10%) while the reason is not established.
• FactorsToday’s m3/m6 leaderboard windows do not reconcile to the AZI CSV (m3 de-annualizes to the CSV’s 126-day return; m6 to the 252-day total return). All trailing returns in the report are taken from the AZI CSV (E1), not the leaderboard.
• FactorsToday’s “All Factors” loadings model was last refreshed 2026-07-31 while the other three refreshed 2026-08-14, so the custom-basket read predates most of the post-print tape; those betas are treated as indicative only.
• Vendor market-sizing (C19–C21) carries wide dispersion and is used directionally, reconciled to Carrier’s own disclosed segment and data-center revenue.
• 13F-derived ownership items in the news feed (CalPERS, D.A. Davidson, E. Öhman J:or) are Q1-2026 vintage — they predate everything material to this update and were not used.

G.7 The CSE goodwill test — a dated unknowable, stated as such

Carrier’s annual goodwill impairment test date is July 1 (Q2-FY2026 10-Q, Note 4). The FY2026 test therefore fell inside Q3-FY2026, and its result is not yet public — it will first appear in the Q3-FY2026 Form 10-Q, expected ~late October 2026 (with fuller disclosure in the FY2026 10-K, ~February 2027, unless an interim trigger forces an 8-K).

The ~14% CSE fair-value cushion quoted throughout the report is the FY2025 disclosure (10-K, p.45, a PwC Critical Audit Matter) and is untested in public since 2025-07-01. The $7,586M CSE goodwill block is fully intact — the $234M 1H decline was 99% foreign-currency translation and Riello carried only $3M of CSE goodwill. Every statement in the report about the cushion is explicitly framed as pending, not as a known outcome, and this report notes a fourth possible outcome the scenario table does not model: a disclosure that the cushion narrowed without a charge.

G.8 Undisclosed items that limit precision (recorded, not guessed)

• NORESCO sale proceeds and 75F purchase price are undisclosed by all parties; no 8-K was filed for 75F. The FY2026 exit-adjusted revenue base therefore carries a real error bar — and the asymmetry runs against the guide: a revenue shortfall raises the required 2H margin from ~16.0% to ~16.8%.
• Carrier has never disclosed a dollar tariff impact. The only sizing anywhere is the Q1 call’s $400–450M gross 2026 input-cost exposure (~75% tariff-related). The report’s split of the ~185bp gross-margin decline between tariff, mix and non-tariff input cost is labeled an ASSUMPTION.
• Segment D&A is not disclosed (“not provided to the CODM”), so any CSE EBITDA — and therefore the delivered-versus-underwritten Viessmann multiple — rests on an assumed add-back. Directionally robust, not precise.
• The hyperscaler/colocation long-term agreements’ cancellation, repricing, indexation and take-or-pay terms are undisclosed at both Carrier and Trane. This is the highest-value undisclosed item in the quarter and is carried as an open question, not resolved.
• Carrier’s unrecognized IEEPA refund entitlement is unquantified and unbooked.
• No quantified Viessmann synergy scorecard appears in any SEC filing, three years into a deal whose €200M of synergies were promised “by year three.”
• No 2026 rating-agency action was located (the standing marks are S&P BBB+/Stable, 2025-03-05, and Moody’s Baa1/Positive).

G.9 Date integrity

• Primary sources are current. The load-bearing evidence is the Q2-FY2026 10-Q and earnings release (both 2026-07-28), the Q2-FY2026 call (2026-07-28), the FY2025 10-K (2026-02-05), the 2026 DEF 14A (2026-03-03) and the Form 3/4/5 corpus through 2026-07-24. No primary source relied upon is stale.
• EDGAR / XBRL data accessed 2026-08-15. The full five-year SEC filing corpus was enumerated as of 2026-08-15 — 312 documents, reconciled against the EDGAR submissions history, which is what licenses treating the single-open-market-purchase Form 4 result as a real finding rather than a retrieval gap.
• Price, factor and valuation-percentile data are as of the 2026-08-14 close. FactorsToday’s All Factors model is the one exception (data 2026-07-31, flagged in G.6).
• Industry data vintages: AHRI through June 2026; EHPA through Q1-2026 (2026-08-06 release) with the 2025 full-year report; Section 232 through Proclamation 11032 (2026-06-08); IEEPA through the 2026-02-20 decision and the CIT refund order; copper through 2026-08-07/14; ACT Research 2026 forecasts; housing data through July 2026.
• Peer data is the June-2026 quarter for every comparable (Trane, Lennox, Johnson Controls FQ3, Daikin Q1-FY2027, AAON, Modine, Vertiv), i.e. exactly coincident with Carrier’s Q2 — which is what makes the natural-experiment comparison valid.
• Post-period events are flagged where they fall outside the 10-Q: the NORESCO close (2026-08-03) and the 75F acquisition (2026-07-23) both post-date or sit outside the filing’s disclosure, and the Q3-FY2026 10-Q is identified as the filing that will carry them, the goodwill-test result and the Q4 margin evidence together.

No claim appears above without a corresponding source in this appendix; where a claim could not be sourced, it was excluded (G.3), deflated (G.4) or explicitly labeled as an assumption or open question (G.8).