Air Products and Chemicals APD

Reference table

1. The two years of write-offs, and what the filing actually says

Consolidated income statement, $M, filing-verified:

Read the two bold rows against each other and the whole company is explained. The GAAP line has been destroyed in each of the last two years by a charge of almost exactly the same size — $2,952.0 million in the nine months to June 2025 and $2,929.4 million in the nine months to June 2026. The adjusted line, which the company publishes and reconciles, grew 14.1% in the quarter and 14.8% across nine months.

The 2026 charge, from the 8-K filed 2026-06-30 under Item 2.06 — Material Impairments, quoted in full because it is the most consequential disclosure in this file:

> "On June 26, 2026, Air Products and Chemicals, Inc. ... determined that, as part of a review initiated by its Board of Directors and Chief Executive Officer, it would exit projects to develop a clean energy complex to produce low carbon hydrogen and ammonia in Louisiana (the 'Louisiana Clean Energy Complex'), a facility to produce green hydrogen in Arizona (the 'Casa Grande Project') and other smaller scale projects supporting clean energy distribution. As a result of these decisions, the Company expects to record a pre-tax charge of up to $2.9 billion, or $2.2 billion on an after-tax basis, in its fiscal 2026 third quarter, primarily to write down assets and terminate contractual commitments. Cash expenditures related to these charges are currently estimated not to exceed $925 million based on contractual terms and commitments... The Company previously disclosed that it would not make a final investment decision with respect to the Louisiana Clean Energy Complex unless it determined that it would be able to execute a de-risking strategy that included signing firm offtake agreements for hydrogen and nitrogen supply and achieving construction and capital costs in line with its return expectations... After a detailed review, the Company determined that the expected financial returns from the project would not meet its return criteria."

Three things to take from that paragraph.

The reason given is return on capital, stated without euphemism. "The expected financial returns from the project would not meet its return criteria." That is a company declining to spend good money after bad, and it is the correct decision reported in the correct language.

The gating condition it failed was demand. The de-risking strategy required "signing firm offtake agreements for hydrogen and nitrogen supply." The projects were abandoned because customers did not sign. That is a fact about the clean-hydrogen market, not only about Air Products, and it is the most useful sector datum in this dive.

$925 million of cash has not yet been spent. The charge is largely non-cash asset write-downs, but "cash expenditures related to these charges are currently estimated not to exceed $925 million", with the company anticipating "lower cash spend once negotiations and ultimate settlements are finalized with third parties." That is a real, future, unquantified-within-a-range obligation and it is one of the kill criteria in Section 8.

The governance history that produced this

The income statement carries a line the vendor payload does not explain: "Shareholder activism-related costs" of $86.3 million in the nine months to June 2025 and $25.0 million in the June 2025 quarter alone. The 10-Q's related-party note supplies the context: "During the third quarter of fiscal year 2025, we reimbursed $24.7 to Mantle Ridge LP and certain of its affiliated entities... for costs they incurred in connection with the proxy contest that concluded in January 2025."

A proxy contest concluded in January 2025, cost the company $86.3 million, and produced the current chief executive, Eduardo F. Menezes. The 2026 project exits are described in the 8-K as arising from "a review initiated by its Board of Directors and Chief Executive Officer." The causal chain — activist campaign, board change, strategic review, project exits, capital-expenditure reduction, free-cash-flow inflection — is visible across these documents and is the single most important context for reading the numbers in Section 2.

2. The cash-flow inflection, which is the actual news

Consolidated statement of cash flows, nine months ended 30 June, $M, filing-verified:

A $3,464 million improvement in nine-month free cash flow, from a $3.5 billion outflow to essentially breakeven. It comes from both directions at once: operating cash flow up 65.8% and capital expenditure down 39.1%.

The operating-cash-flow improvement is roughly two-thirds working capital. The nine-month working-capital drag fell from $1,069.5 million to $210.8 million — an $858.7 million swing, of which the largest single item is "other working capital" moving from a $624.6 million use to a $56.6 million source. That is a genuine improvement in discipline and it will not repeat at the same magnitude. The remaining improvement is operating.

The capital-expenditure reduction is the durable half and it is the direct consequence of the project exits. Long-term debt proceeds fell from $3,978.2 million to $644.0 million — the company stopped borrowing to build — and of the $644.0 million drawn, the 10-Q states "$543 drawn from non-recourse project financing available to the NGHC joint venture." So Air Products' own recourse borrowing across nine months was approximately $101 million.

Total debt confirms it. From the 10-Q's financing section, verbatim: "Total debt was $17.7 billion as of both 30 June 2026 and 30 September 2025." Flat, to the stated precision, across three quarters, on a balance sheet that had grown debt from $11.0 billion (FY2023) to $15.0 billion (FY2024) to $18.4 billion (FY2025) in the two years before.

The annual context, from cf_a, $M — and the trajectory is the point:

Trailing free cash flow is approximately MINUS $301 million, a yield of MINUS 0.46%. The vendor reports freeCashFlowYieldTTM of PLUS 3.135%, and Section 6 sets out exactly why that is wrong and by how much.

What this means for the dividend, which is the practical question. The dividend is $7.20 per share on 222.7 million shares — approximately $1.6 billion a year — and the prior-year nine months carried "dividend payments to shareholders of $1.2 billion." Air Products has been paying that dividend out of borrowings for three years. On the current trajectory it will be paying it out of cash flow within a year, and that is the single most consequential change in this file.

3. What the business is, and where the revenue comes from

Revenue by product line, from seg_prod — the block reconciles exactly and is used ($M):

The FY2025 rows sum to $12,037.3 million, which is exactly consolidated revenue. Clean.

On-site at 51.3% is the business. An on-site plant is built adjacent to a customer's facility — a refinery, a steel mill, a chemical plant — under a long-term take-or-pay supply agreement, typically fifteen to twenty years, with energy costs passed through. The customer cannot switch supplier without rebuilding, and the revenue does not depend on the customer's own volumes. That is the most durable revenue in this batch and it is why the exponential score of 3 is a compliment about quality rather than a criticism.

Merchant at 44.3% is liquid and packaged gas delivered by truck and cylinder, which is a route-density business with local scale economics — good, and cyclical.

Sale of Equipment fell 41.5% over two years, from $889.0 million to $520.0 million, which no vendor field explains and which is consistent with the wind-down of the large project pipeline.

Geography — seg_geo is clean for the recent years:

The four segments sum to $11,517.3 million against $12,037.3 million of consolidated revenue — a 4.3% gap that corresponds almost exactly to the $520.0 million Sale of Equipment line, which is reported separately from the geographic segments. The Americas is correctly identified as the largest region. Note that the FY2022 row uses a completely different structure (China / United States / Other Foreign Operations) and FY2019 and earlier mix segment and country labels, so only FY2023-25 are comparable.

4. Valuation — priced in or room?

At $294.67 (market cap $65.62B, 222,685,530 shares). Air Products' fiscal year ends 30 September; the estimate rows are labelled by the fiscal year they cover.

Estimate coverage is good on the anchor years and collapses beyond. FY2027 rests on 12 analysts for EPS and 14 for revenue, with an EPS range of $14.150 to $14.605 — a 3.2% spread, among the tightest in this batch. FY2028 rests on 5 analysts for EPS. FY2029 and FY2030 rest on ONE analyst each for EPS and are internally incoherent — FY2030 revenue of $15,293 million is 7.0% BELOW FY2029's $16,452 million and FY2030 EPS of $16.67 is 11.1% below FY2029's $18.75. A consensus that has revenue and earnings falling in the final year with a single contributor is not a consensus; both rows are excluded from every conclusion in this dive.

The trailing P/E of MINUS 1,417x is arithmetically correct and analytically void, because trailing net income is approximately negative $47 million after two write-off cycles. priceToEarningsRatioTTM, returnOnEquityTTM (−0.32%), returnOnAssetsTTM (−0.12%), earningsYieldTTM (−0.07%), interestCoverageRatioTTM (−2.68x), incomeQualityTTM (−3,809), dividendPayoutRatioTTM (−33.79) and effectiveTaxRateTTM (104.9%) are ALL rejected as lenses on this name, and every one of them is a mechanical artefact of the same $5.9 billion of charges. grahamNumberTTM reads null, which is the payload correctly declining to compute a square root of a negative number.

est.ebitdaAvg and est.ebitAvg are REJECTED for a fixed-ratio fabrication signature. From FY2023 through FY2030, ebitdaAvg is exactly 34.303% of revenueAvg and ebitAvg exactly 22.483%, in every year without exception. Air Products' realised EBITDA margin has ranged from 11.1% (FY2025, charge-depressed) to 53.6% (FY2024, gain-inflated) across the same window. All forward valuation uses epsAvg.

Enterprise value — the noncontrolling interests are NOT fully included. enterpriseValueTTM of $82,795.5M less market capitalisation of $65,618.7M implies $17,176.8M. The 10-Q states total debt of $17.7 billion at 2026-06-30; net of the FY2025 cash balance of $1,856.0M that is approximately $15,844M, to which the $2,324.9 million of noncontrolling interests should be added, giving approximately $18,169M. The vendor's figure is roughly $992 million short — about 5.8% of net debt and 1.2% of enterprise value. The noncontrolling interests here are not a rounding item: they are principally the NEOM Green Hydrogen Company joint venture, which is consolidated but funded substantially by non-recourse project financing and by partner equity, so a reader who ignores them is understating the claims on this business. evToEBITDATTM of 63.81x and netDebtToEBITDATTM of 13.24x are both rejected outright — they divide by a trailing EBITDA that the write-offs have crushed and they describe nothing.

Peer context — the set is half wrong. The vendor peers are Barrick Mining, Corteva, Ecolab, Freeport-McMoRan, Martin Marietta, PPG, Sherwin-Williams, Vale, Vulcan Materials and Wheaton Precious Metals. Four of the ten are mining companies (B, FCX, VALE, WPM) and two are aggregates producers (MLM, VMC), none of which shares any economic characteristic with contracted industrial-gas supply. Linde and Air Liquide — the only two genuine global comparables, and the companies against which every judgement about this business is made — are both absent. No peer multiple comparison is drawn, though the industrial-gas group has historically supported mid-twenties to low-thirties earnings multiples, which is the context for calling 20.4x a discount.

4a. What today's price assumes (the inversion)

At $294.67 — 22.0x FY2026 consensus, 20.4x FY2027, 4.73x book — the price embeds:

• Adjusted EPS reaches $13.385 in FY2026 and $14.423 in FY2027. (Consensus; 11 and 12 analysts.) The first three quarters delivered $3.16 + $3.20 + $3.47 = $9.83, so FY2026 needs $3.555 in the fourth quarter against a consensus of $3.60. Essentially in hand.
• The write-offs are finished. (Our derivation; the charges are the company's.) This is the most fragile assumption in the price. Two consecutive fiscal years at approximately $2.9 billion each, the second described as arising from a board-and-chief-executive review. The NEOM Green Hydrogen Company joint venture remains consolidated and is the largest surviving clean-energy commitment.
• Capital expenditure keeps falling toward a maintenance level. (Our number, from the filed nine-month figures.) From $7,022.6 million (FY2025) to an annualised $4,472 million on the nine-month run rate. Against depreciation and amortisation of roughly $1,508 million annualised, capex is still 3.0x D&A, so there is further room to fall — and every dollar of it converts directly into free cash flow.
• The dividend is safe and is funded from operations within a year. (Our derivation.) $7.20 per share is approximately $1.6 billion a year against trailing free cash flow of MINUS $301 million. On the current trajectory that gap closes; it has not closed yet.
• The market keeps paying roughly 20-23x forward earnings. (Our number.) At 18x FY2027E the stock is $260; at 25x it is $361. The industrial-gas group has historically commanded more; the discount is the market's price for two years of destroyed capital, and closing it is the bull case.

4b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: adjusted EPS growth (+7.8%, FY2026E $13.385 to FY2027E $14.423) + multiple drift (assumed modest EXPANSION, 22.0x on the forward year to about 22.5x on the then-forward year) + dividend yield (+2.44%) ≈ +10% to +12%.

Roughly two-thirds of the base case is earnings growth plus dividend and about one-third is a modest re-rating. That is a reasonable decomposition and it is not an exciting one — 7.8% consensus earnings growth on 5-6% revenue growth from a company with a 2.44% yield is a low-double-digit total return, which is what the price says.

The interesting asymmetry is not in the earnings, it is in the free cash flow. If capital expenditure continues toward a maintenance level, trailing free cash flow goes from MINUS $301 million to plausibly $2 billion or more within eighteen months, at which point the free-cash-flow yield goes from negative to roughly 3% and the dividend is comfortably covered for the first time since fiscal 2022. That is a balance-sheet story rather than an earnings story, and it is the reason the medium-horizon stance is a tailwind while the verdict is a Hold.

If the multiple compressed to 18x FY2027E the price would be $260 (−11.8%). At 25x, $361 (+22.4%).

4c. Variant perception (where we differ, what would surprise)

• We are BELOW the street and the gap is modest. Our base of $325 is 6.2% below the consensus target of $346.44 and 7.1% below the median of $350. The street's LOWEST published target, $320, is 8.6% above spot, and 22 of 42 analysts rate the shares Buy with none at sell. We have no variant perception on direction and a slightly more conservative view on magnitude.
• We think the free-cash-flow inflection is materially under-weighted relative to the write-off headline. Nine-month free cash flow moved $3,464 million, from MINUS $3,509 million to MINUS $45 million, while total debt stayed flat at $17.7 billion. Watchable number: additions to plant and equipment in the fiscal 2026 10-K, against $7,022.6 million in FY2025 and $3,354.5 million in nine months. A full-year figure below $4.5 billion would confirm the reset; below $3.5 billion would make free cash flow decisively positive in FY2027.
• We think the write-off risk is not fully retired and we have not priced it as retired. Two consecutive years at approximately $2.9 billion. The NEOM joint venture survives. Watchable event: any further Item 2.06 filing. The bear case at $235 assumes one more cycle.
• We have NO expert overlay and the absence is conspicuous rather than neutral. Zero knowledge-base claims out of 52,021, on a name whose entire recent history is a clean-hydrogen retreat — and the free-text sweep on "hydrogen" returned nothing naming this company. The store discusses the theme and not the operator, which is a real gap and is reported as one.
• Positive surprise that would force a re-rate: fiscal 2026 full-year capital expenditure below $4.5 billion with no further impairment; the fiscal 2027 outlook implying positive free cash flow; a disclosed resolution of the NEOM position; or a buyback authorisation, which the company has not run at all — commonStockRepurchased reads $0 in each of FY2025, FY2024 and FY2023.
• Negative surprise that would break the thesis: a third impairment cycle of any size; cash costs from the June exits exceeding the $925 million estimate; total debt rising above $19 billion; adjusted pre-tax income growth falling below 5%; or a dividend that has to be funded by borrowing for a fourth consecutive year.

Synthos fair values

All three anchors are multiples of the FY2027 consensus adjusted EPS distribution (mean $14.423, low $14.150, high $14.605, 12 analysts), cross-checked against FY2026 and FY2028. The analyst EPS range is only 3.2% wide, so the width of the range below comes almost entirely from the multiple.

• Bear ~$235 — 16.6x the FY2027 consensus LOW of $14.150, and 17.6x FY2026E. Cross-check: 1.8% above the 52-week low of $230.76; 3.15x book; a 3.06% dividend yield at that price. The scenario: a third impairment cycle appears, cash costs from the June exits exceed $925 million, capital expenditure does not fall further, free cash flow stays negative into FY2027, and the market prices a company that has destroyed $5.9 billion at a deep discount to its industry. −20.2%.
• Base ~$325 — 22.5x the FY2027 consensus MEAN of $14.423, and 24.3x FY2026E, 20.7x FY2028E. Cross-check: 4.36x book; 3.4% above the 52-week high of $314.19; 6.2% below the street consensus of $346.44. Sensitivity, stated openly: 20.4x FY2027E gives $294.67 — spot exactly — and 25x gives $361. The base is a two-turn re-rating on an estimate row twelve analysts agree on to within 3.2%. The scenario: no further write-offs, capital expenditure keeps falling, free cash flow turns positive during FY2027, adjusted earnings compound at the consensus 8%, and the multiple recovers part of its discount to the industrial-gas group. +10.3%.
• Bull ~$380 — 26.0x the FY2027 consensus HIGH of $14.605, and 24.3x FY2028E. Cross-check: 5.10x book; 21.0% above the 52-week high; 1.9% above the highest published street target of $373. The scenario: capital expenditure falls to near maintenance, free cash flow reaches $2 billion or more, a buyback is initiated for the first time in this file's history, the NEOM position is resolved, and Air Products re-rates back toward the multiple Linde and Air Liquide command. +29.0%.

Base is 10.3% above spot; asymmetry roughly 1.44:1 (20.2% down, 29.0% up), plus a 2.44% dividend. A base near 10% with a payoff ratio below 1.5:1 does not clear a Buy bar, and it is a perfectly adequate description of a good business that is two-thirds of the way through repairing itself and is priced approximately correctly for that.

5. Knowledge base — 23 raw hits, ZERO claims, and the absence is conspicuous

Raw hits: 23. Entity matches after a case-sensitive re-run: ZERO. Text matches: ZERO. Name-level claims on Air Products: ZERO. Discarded: 23.

The primary sweep ran the entity terms APD, Air Products and Air Products and Chemicals, plus free text on industrial gas and hydrogen, across all 52,021 distilled claims. It returned 23 hits across 11 channels — jacob_shapiro (5), we_study_billionaires (4), macrovoices (4), business_breakdowns (2), doomberg (2) and one each from six others. Not one names this company, and the case-sensitive entity re-run returns nothing at all.

The Synthos knowledge base contains ZERO claims on Air Products and Chemicals, ZERO on the Louisiana Clean Energy Complex, ZERO on the Casa Grande project and ZERO on the NEOM Green Hydrogen Company joint venture.

Breadth 0, claim count 0, net conviction empty. This is the fifteenth void this programme has found and reported honestly.

And unlike the other empty lanes in this batch, this one is worth a paragraph rather than a sentence, because the absence has a shape. The free-text term "hydrogen" was included in the sweep precisely because clean hydrogen has been one of the most-discussed industrial themes of the last three years, and several of the eleven channels that surfaced — doomberg, macrovoices, jacob_shapiro — discuss energy transition regularly. The store therefore contains sector-level commentary on the theme and no claim whatsoever about the company that had committed the most capital to it. Air Products has now written off approximately $5.9 billion pursuing that theme across two fiscal years, and the 8-K states that the projects failed on the inability to sign "firm offtake agreements for hydrogen and nitrogen supply" — which is, in one sentence, the most concrete evidence about clean-hydrogen demand that this programme has encountered, and it arrived from a filing rather than from an expert.

On a name whose entire recent history is a strategic reversal, having no independent voice is a genuine limitation on the analysis and not a neutral fact. It is stated as one, and no concentration test, attribution note or speaker analysis is possible on zero claims.

6. Data integrity — what we rejected and why

Six findings. APD's payload is a case study in what two years of large impairments do to a derived-metrics block: nine separate trailing ratios are arithmetically correct and analytically void, the capital-expenditure field is 48% understated with the wrong sign on free cash flow, and the vendor's own composite rating scores 1 out of 5 on every dimension for reasons that have nothing to do with the operating business.

1. capitalExpenditure in the trailing block is approximately 48% understated, and the sign of free cash flow is WRONG. REJECTED. capexPerShareTTM of $11.281 on 222.686 million shares implies trailing capital expenditure of approximately $2,512 million. The filed figures are $3,354.5 million for the nine months to 2026-06-30 alone and $7,022.6 million for fiscal 2025, giving a rebuilt trailing figure of $4,872.2 million ($7,022.6M − $5,504.9M + $3,354.5M). The vendor's implied trailing operating cash flow of $4,568 million matches a rebuild of $4,570.8 million almost exactly, so the error is confined to capex. Consequently freeCashFlowYieldTTM of PLUS 3.135% is REJECTED — the correct figure is approximately MINUS $301 million, a yield of MINUS 0.46%, a swing of roughly $2.36 billion and a change of sign. Also rejected: freeCashFlowPerShareTTM ($9.234), priceToFreeCashFlowRatioTTM (31.89x), evToFreeCashFlowTTM (40.24x), capexToOperatingCashFlowTTM (0.550), capexToRevenueTTM (0.199), capexToDepreciationTTM (1.628) and freeCashFlowToEquityTTM ($2,184.9M). Note that the vendor's figure is close to what Air Products' OWN non-GAAP "capital expenditures" measure would produce — the 10-Q defines that measure as excluding "spending for additions to plant and equipment by our consolidated joint venture, NEOM Green Hydrogen Company, to the extent such spending is funded by sources other than Air Products' cash" — so the vendor may be tracking a company-defined figure rather than the cash-flow-statement line. The house rule is filed capex, and this dive uses it, while recording that the discrepancy has a plausible non-random explanation.

2. NINE trailing ratios are arithmetically correct and analytically void — REJECTED as a class. Two years of roughly $2.9 billion charges have driven trailing net income to approximately MINUS $47 million, and every ratio with earnings in the numerator or denominator has become nonsense: priceToEarningsRatioTTM MINUS 1,416.68x, priceToEarningsDilutedRatioTTM the same, returnOnEquityTTM −0.315%, returnOnAssetsTTM −0.117%, returnOnCapitalEmployedTTM −1.653%, earningsYieldTTM −0.072%, interestCoverageRatioTTM MINUS 2.68x, incomeQualityTTM MINUS 3,809, dividendPayoutRatioTTM MINUS 33.79 and effectiveTaxRateTTM 104.87%. grahamNumberTTM reads null, which is the payload correctly refusing to take the square root of a negative product. netDebtToEBITDATTM of 13.24x and evToEBITDATTM of 63.81x divide by a charge-crushed trailing EBITDA and describe nothing. None of these is used anywhere in this dive; every profitability figure comes from the company's adjusted reconciliation in the 10-Q.

3. Noncontrolling interests of $2,324.9 million are NOT fully included in enterprise value. The implied net debt of $17,176.8M sits roughly $992 million below a rebuild of filed total debt ($17,700M) less cash ($1,856.0M) plus noncontrolling interests ($2,324.9M) = $18,169M. The NCI here is principally the NEOM Green Hydrogen Company joint venture, which is consolidated but funded substantially by non-recourse project financing and partner equity — so it is a material claim on the consolidated business, not a rounding item. This is the standard omission, present and quantified at 1.2% of enterprise value.

4. est.ebitdaAvg and est.ebitAvg carry a fixed-ratio fabrication signature — REJECTED. From FY2023 through FY2030, ebitdaAvg is exactly 34.303% of revenueAvg and ebitAvg exactly 22.483%, in every year. The company's realised EBITDA margin over the same window ranges from 11.1% to 53.6%, so the fabricated ratios track nothing. All forward valuation uses epsAvg.

5. The FY2029 and FY2030 estimate rows rest on ONE analyst each and are internally incoherent — EXCLUDED. FY2030 revenueAvg of $15,293 million is 7.0% BELOW FY2029's $16,452 million, and FY2030 epsAvg of $16.67 is 11.1% below FY2029's $18.75. A forward "consensus" in which the terminal year shows revenue and earnings falling, contributed by a single analyst, is not a consensus. Both rows are excluded from every conclusion in this dive.

6. The vendor composite rating of D+ / 1 is mechanically derived from the rejected metrics and carries no weight. Every one of the six sub-scores reads 1 out of 5 — discountedCashFlowScore, returnOnEquityScore, returnOnAssetsScore, debtToEquityScore, priceToEarningsScore and priceToBookScore. Four of those six are computed on trailing earnings that two write-off cycles have made meaningless, and the cash-flow sub-score is built on the capital-expenditure figure rejected in finding 1. A uniform 1-out-of-5 rating on a company whose adjusted pre-tax income grew 14.8% and whose free cash flow improved $3.46 billion is a demonstration of what happens when a rating engine meets a non-recurring charge, and it is recorded as such.

Verified CLEAN — recorded because clean checks are findings:

• Share count — essentially exact. The 10-Q cover states 222,685,530 shares outstanding at 2026-06-30 against 222,679,000 implied by market capitalisation ÷ price — a 0.003% match. weightedAverageShsOutDil has moved only from 222.3 million to 222.9 million across six fiscal years, and commonStockRepurchased is $0 in every year of cf_a — Air Products has neither bought back nor issued meaningful equity.
• seg_prod reconciles EXACTLY. On-site $6,180.4M + Merchant $5,336.9M + Sale of Equipment $520.0M = $12,037.3M, exactly FY2025 consolidated revenue.
• seg_geo is coherent for FY2023-25 and correctly identifies the Americas as the largest region, with the four segments summing to $11,517.3M against $12,037.3M — the 4.3% gap corresponding to the separately reported Sale of Equipment line. Earlier years use incompatible structures and are not used.
• tech.max_dd_from_peak of −12.838% DIVERGES from pct_from_hi of −6.213%, which per the data contract is correct and informative: the six-year peak lies above the fifty-two-week high, implying a multi-year maximum near $338. This is one of only two names in this batch where the two fields separate, and it is used as intended — as the current distance from the multi-year high, never as a maximum drawdown.
• quote versus tech. tech.last of $294.67 equals quote.price exactly. quote.yearHigh/yearLow ($314.87/$229.11) against tech.hi52/lo52 ($314.19/$230.76) — 0.22% and 0.72%. tech is used throughout.
• dividendPerShareTTM of $7.20 matches profile.lastDividend of 7.2 and reconciles to dividendYieldTTM of 2.443%. Clean.

Non-equity tripwire — checked and passed. APD is common stock, par value $1 per share, NYSE-listed, with 222,685,530 shares outstanding at 2026-06-30. Price of $294.67 is not par-like; beta is 0.736; volume was 0.70M shares (~$207M of turnover, the lightest in this batch); the 52-week band of $230.76 to $314.19 is a 36% range; the dividend is a regular quarterly common dividend. Note the $2,324.9 million of noncontrolling interests and the related-party shareholder loans of $215.7 million from joint-venture partner Lu'An Clean Energy Company, neither of which is this security. This is common equity.

7. Technicals and insiders

• Price $294.67. −6.2% from the 52-week high of $314.19; +27.7% above the 52-week low of $230.76. Position within the annual range: 76th percentile.
• AND −12.8% from the six-year peak, implying a multi-year maximum near $338 — so the stock is closer to its recent high than to its all-time one.
• Both moving averages are below the price and the gap is modest. +1.8% above a 50-day average of $289.60; +6.6% above a 200-day average of $276.56.
• RSI 51.4 — neutral, the closest to 50 in this batch. MACD +0.79 — mildly positive.
• Relative performance: 3-month −1.2% against SPY +7.6%; 6-month +8.7% against SPY +11.1%; 12-month +4.6% against SPY +24.3% and QQQ +30.8%. Air Products has underperformed the index over every window, most severely over twelve months — which is what a company taking two consecutive multi-billion write-offs should do.
• Sentiment: 22 buy, 20 hold, 0 sell across 42 analysts; consensus target $346.44 (+17.6%), median $350, high $373 (+26.6%), low $320 — 8.6% ABOVE spot. No covering analyst has a target below the current price. The usual caution about unanimity applies with reduced force here, since the shares have already underperformed the index by twenty points over a year.

Today's move: APD closed 2026-08-04 at $294.67, up $1.73 or 0.59%, from $292.94, on 0.70M shares — the lightest turnover in this batch at approximately $207 million. The most recent filings are the 8-K and 10-Q of 2026-07-30, five days earlier.

Insiders — eight filings and no signal whatsoever. The most recent, filed 2026-07-02 for a 2026-06-30 transaction, is an A-Award of 6.4626 phantom stock units to director Howard I. Ungerleider, taking his holding to 975.3141 units — a dividend-equivalent accrual on a deferred-compensation balance, not a grant and not a decision. The file contains no open-market purchase and no open-market sale by any person. On a name where a proxy contest in January 2025 replaced the strategy and where two write-off cycles have followed, the complete absence of discretionary insider activity is uninformative and is reported as uninformative. Note that Howard Ungerleider is a former chief financial officer of a major chemicals company and his presence on the board is consistent with the post-activism reconstitution, but nothing in this archive documents that.

8. Verdict, kill-criteria and flip conditions

Hold.

The case for the business. Air Products owns one of the best revenue models in industrials: $6,180.4 million of FY2025 revenue, 51.3% of the total, from on-site plants under long-term take-or-pay contracts that a customer cannot exit without rebuilding its own facility. The underlying performance is good and improving: sales +4.6% in the quarter and +6.4% across nine months; adjusted income from continuing operations before taxes +14.1% and +14.8%; equity affiliates' income +22.4%; interest expense down. Adjusted earnings have beaten consensus in each of the last five quarters. And the capital picture has turned decisively: capital expenditure down 39.1% to $3,354.5 million across nine months, operating cash flow up 65.8% to $3,309.6 million, free cash flow from MINUS $3,509 million to MINUS $45 million, and total debt flat at $17.7 billion. Beta is 0.736 and the dividend yields 2.44%.

The case against the price, in four parts.

First, the arithmetic is ordinary. Base $325, +10.3%, asymmetry 1.44:1. Adding the dividend gives roughly a 12.7% twelve-month expected return against 20.2% of downside. At 20.4x FY2027 consensus on 7.8% expected earnings growth, the price is approximately right.

Second, the write-offs may not be finished. $2,952.0 million in the nine months to June 2025 and $2,929.4 million in the nine months to June 2026 — approximately $5.9 billion in twenty-four months. The NEOM Green Hydrogen Company joint venture remains consolidated and is the largest surviving clean-energy commitment, and $925 million of cash costs from the June exits has not yet been spent.

Third, free cash flow is still negative. Trailing MINUS $301 million against a dividend costing approximately $1.6 billion a year. The trajectory is excellent and the destination has not been reached — Air Products has funded its dividend with borrowings for three consecutive fiscal years.

Fourth, the knowledge base is empty and the absence is conspicuous. Zero claims out of 52,021 on a company whose entire recent history is a clean-hydrogen retreat, from a store that discusses the hydrogen theme at the sector level. There is no independent voice on this name at all.

And the honest counterweight, which is why this is a Hold rather than an Avoid. The reset is working, it is visible in filed cash-flow statements rather than in management language, and if capital expenditure continues toward a maintenance level then free cash flow turns positive during fiscal 2027 and the dividend is covered from operations for the first time since fiscal 2022. That is a balance-sheet inflection worth owning at a lower price, and the upgrade conditions below say exactly where.

Pre-registered KILL criteria — what would take this to Avoid:

• Any further Item 2.06 material-impairment filing, of any size, which would make three consecutive years of write-offs and would end the argument that management has finished re-underwriting the portfolio.
• Cash costs from the June 2026 exits exceeding the stated $925 million estimate.
• Total debt rising above $19 billion, against $17.7 billion flat for three quarters.
• Full-year fiscal 2026 capital expenditure above $5.5 billion, which would mean the nine-month reduction was timing rather than a reset.
• Adjusted income from continuing operations before taxes growing below 5%, against +14.8% across nine months.
• A dividend funded by net new borrowing for a fourth consecutive fiscal year.

Pre-registered UPGRADE conditions — what would take this to Buy — Tactical:

• Full-year fiscal 2026 capital expenditure below $4.5 billion with no further impairment, which would put free cash flow on a clear path to positive during fiscal 2027. This is the cleanest trigger and it resolves on the 2026-11-05 print.
• A fiscal 2027 outlook implying positive free cash flow — the first since fiscal 2022.
• A price below approximately $265, which is 18.4x FY2027E with a 2.72% dividend yield and would put the base case above 22%.
• A disclosed resolution of the NEOM Green Hydrogen Company position — sale, restructuring or a firm offtake agreement — which would retire the largest surviving clean-energy exposure.
• A share-repurchase authorisation. commonStockRepurchased reads $0 in every year of the payload, so any buyback would be a genuine change in capital-allocation posture.
• Any independent knowledge-base claim naming this company. At present there are zero.

Where APD fits in the Synthos Framework Portfolio. No position today, with a price trigger at approximately $265 and a calendar trigger at the 2026-11-05 fiscal-year print. The materials sleeve would take this name at a target weight of 2% on either the capital-expenditure confirmation or the price. Sizing note: this is a high-quality annuity business mid-repair, so the correct discipline is patience rather than a smaller position — the free-cash-flow inflection is a 2027 event and there is no reason to pay for it in 2026. Logged as a tracked Synthos call (Hold) as of 2026-08-04 at $294.67, with the fair-value anchors, kill criteria and upgrade conditions all gradeable.

Single biggest risk: that the write-offs are not finished. Air Products has recorded "business and asset actions" of $2,952.0 million in the nine months to June 2025 and $2,929.4 million in the nine months to June 2026 — approximately $5.9 billion of shareholder capital declared not worth what was spent on it, in twenty-four months. The second tranche was decided on 2026-06-26 by a board and chief executive installed after a proxy contest that itself cost $86.3 million, and the reason given was the plainest possible one: "the expected financial returns from the project would not meet its return criteria," after the company could not sign "firm offtake agreements for hydrogen and nitrogen supply." What remains is the NEOM Green Hydrogen Company joint venture in Saudi Arabia — consolidated, funded substantially by non-recourse project financing and partner equity, sitting behind $2,324.9 million of noncontrolling interests — plus $925 million of estimated cash costs from the June exits that have not yet been paid. The underlying industrial-gas business is genuinely excellent and genuinely improving, the capital-expenditure reset is visible in filed cash-flow statements rather than in management language, and free cash flow has moved $3.46 billion in the right direction in a single year. But a company that has taken two nine-figure-per-quarter write-offs in consecutive years does not get the benefit of the doubt on the third at 20.4 times forward earnings, and the knowledge base offers no independent view to weigh against that judgement — because it contains none.

Provenance & disclosures

• Traceability: ZERO name-level knowledge-base claims name Air Products and Chemicals out of 52,021 distilled claims (raw hits 23, entity matches after a case-sensitive re-run: 0, text matches 0, discarded 23; breadth 0, claim count 0, net conviction empty). The sweep ran the entity terms APD, Air Products and Air Products and Chemicals, plus free text on industrial gas and hydrogen, and returned 23 hits across 11 channels led by jacob_shapiro (5), we_study_billionaires (4), macrovoices (4), business_breakdowns (2) and doomberg (2) — none naming this company. The store also contains zero claims on the Louisiana Clean Energy Complex, zero on the Casa Grande project and zero on the NEOM Green Hydrogen Company joint venture. The absence has a shape and is reported rather than passed over: the free-text term "hydrogen" was included precisely because clean hydrogen has been a heavily discussed theme, and the store contains sector-level commentary on the theme and no claim whatsoever about the company that committed the most capital to it — a company that has now written off approximately $5.9 billion pursuing it across two fiscal years, with the 8-K stating that the projects failed on the inability to sign "firm offtake agreements for hydrogen and nitrogen supply." That is the most concrete evidence about clean-hydrogen demand this programme has encountered and it arrived from a filing rather than from an expert. This is the fifteenth void found and reported honestly. No concentration test, attribution note or speaker analysis is possible on zero claims and none is manufactured.
• Data as-of: fundamentals — income statement, adjusted pre-tax reconciliation, cash-flow statement, capital expenditure, total debt, share count and related-party disclosures through 2026-06-30, all filing-verified from the 10-Q filed 2026-07-30; annual figures for fiscal 2020 through fiscal 2025 from the vendor payload · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873599 = 2026-08-04T19:59:59Z ($294.67, +0.59%; 50-DMA $289.60; 200-DMA $276.56; RSI 51.4; MACD +0.79) · knowledge-base claims 2026-08-04. Air Products' fiscal year ends 30 September; "FY2025" means the year ended 2025-09-30 and the estimate rows are labelled by the fiscal year they cover. All figures come from the Synthos vendor data file for APD or from the SEC filings in the APD archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2025-11-20 (fiscal year ended 2025-09-30); 10-Q filed 2026-04-30 (March 2026 quarter); 10-Q filed 2026-07-30 (June 2026 quarter — source of the income statement, the adjusted income from continuing operations before taxes reconciliation, the nine-month cash-flow statement, the capital-expenditure figures, the "$17.7 billion total debt at both dates" disclosure, the non-GAAP capital-expenditures definition excluding NGHC non-recourse spending, the Mantle Ridge related-party reimbursement and the share count); 8-K filed 2026-04-30 (Item 2.02, second-quarter results — the Exhibit 99.1 press release is furnished but not in this archive); 8-K filed 2026-06-30 (Item 2.06 — MATERIAL IMPAIRMENTS: the decision of 2026-06-26 to exit the Louisiana Clean Energy Complex, the Casa Grande green hydrogen project and other clean-energy distribution projects, at a pre-tax charge of up to $2.9 billion / $2.2 billion after tax, with cash costs estimated not to exceed $925 million); 8-K filed 2026-07-30 (Item 2.02, third-quarter results — exhibit likewise not in this archive). All carry preserved [TABLE] statement data — 94 tables in the June 10-Q.
• Where the filings contradicted or corrected the vendor (detailed in Section 6): capitalExpenditure in the trailing block, implying approximately $2,512 million against filed figures of $3,354.5 million for nine months and $7,022.6 million for fiscal 2025, giving a rebuilt trailing figure of $4,872.2 million — a 48% understatement that changes the SIGN of free cash flow, from the vendor's PLUS 3.135% yield to a correct MINUS 0.46%, a swing of roughly $2.36 billion (recorded with the note that the vendor's figure is close to what the company's OWN non-GAAP capital-expenditures measure would produce, since that measure excludes NGHC spending funded by non-recourse project financing and partner equity — the house rule is filed capex and this dive uses it); noncontrolling interests of $2,324.9 million not fully included in enterprise value, understating it by roughly $992 million, and material here because the NCI is principally the consolidated NEOM joint venture; nine trailing ratios rendered void by two write-off cycles — priceToEarningsRatioTTM MINUS 1,416.68x, returnOnEquityTTM −0.315%, interestCoverageRatioTTM MINUS 2.68x, incomeQualityTTM MINUS 3,809, dividendPayoutRatioTTM MINUS 33.79, effectiveTaxRateTTM 104.87%, netDebtToEBITDATTM 13.24x, evToEBITDATTM 63.81x and grahamNumberTTM null — all rejected as a class, with every profitability figure in this dive taken from the company's adjusted reconciliation instead; est.ebitdaAvg/ebitAvg, fixed at exactly 34.303% and 22.483% of revenueAvg in every year from FY2023 to FY2030 against a realised EBITDA margin ranging from 11.1% to 53.6%; and the FY2029 and FY2030 estimate rows, each resting on ONE analyst and internally incoherent, with FY2030 revenue 7.0% below FY2029 and FY2030 EPS 11.1% below — both excluded. The entire $2.9 billion project-exit decision of 2026-06-26 is likewise absent from every vendor field, bal_a ending 2025-09-30 and cf_a at FY2025. Where vendor and filing AGREED — recorded, because clean checks are findings: the share count matches to 0.003% (10-Q cover 222,685,530 at 2026-06-30 against 222,679,000 implied), with commonStockRepurchased at $0 in every year; seg_prod reconciles EXACTLY to $12,037.3M of FY2025 revenue; seg_geo is coherent for FY2023-25 and correctly identifies the Americas as largest; dividendPerShareTTM of $7.20 reconciles to the 2.443% yield; and tech.max_dd_from_peak of −12.838% DIVERGES from pct_from_hi of −6.213% correctly, the six-year peak lying above the fifty-two-week high at approximately $338 — one of only two names in this batch where the two fields separate.
• Basis note — read before using any profitability figure: Air Products' GAAP income statement has been dominated for two consecutive years by "business and asset actions" of $2,952.0 million (nine months to June 2025) and $2,929.4 million (nine months to June 2026), producing a June-quarter operating LOSS of $2,097.1 million and a trailing net income near MINUS $47 million. Every profitability figure in this dive comes from the company's own reconciled "adjusted income from continuing operations before taxes" — $972.9 million in the quarter against $852.5 million, and $2,743.5 million across nine months against $2,389.6 million — at adjusted effective tax rates of 18.6% and 18.4%. Consensus EPS estimates are on the same adjusted basis, which is why the forward multiples in Section 4 are usable while the trailing one is not. The prior-year comparison additionally carries $86.3 million of shareholder-activism costs relating to a proxy contest concluded in January 2025, of which $24.7 million was reimbursed to Mantle Ridge LP.
• Estimate coverage: 11 analysts on FY2026 adjusted EPS; 12 on FY2027 — the anchor for all three fair values, with a range of $14.150 to $14.605, a 3.2% spread among the tightest in this batch; 5 on FY2028; ONE analyst each on FY2029 and FY2030, whose rows are additionally internally incoherent and are excluded from every conclusion. Revenue estimates rest on 14 analysts across FY2026-28.
• Peer note: the vendor peer set — Barrick Mining, Corteva, Ecolab, Freeport-McMoRan, Martin Marietta, PPG, Sherwin-Williams, Vale, Vulcan Materials, Wheaton Precious Metals — is substantially unusable: four are mining companies and two are aggregates producers, none sharing any economic characteristic with contracted industrial-gas supply. Linde and Air Liquide — the only two genuine global comparables — are both absent. No peer-multiple comparison is drawn, though the industrial-gas group has historically supported mid-twenties to low-thirties earnings multiples, which is the basis for describing 20.4x FY2027 as a discount.
• Fair-value caveat: the $235 / $325 / $380 anchors are multiples of the FY2027 consensus adjusted EPS distribution — 16.6x the low of $14.150, 22.5x the mean of $14.423, and 26.0x the high of $14.605 — cross-checked against FY2026, FY2028 and book value per share of $74.49 (3.15x / 4.36x / 5.10x). Stated arithmetic, not a discounted cash flow. Because the analyst EPS range is only 3.2% wide, essentially the entire width of the range comes from the multiple. The base is sensitivity-disclosed: 20.4x FY2027E gives $294.67, which is spot exactly; 25x gives $361. Section 4b notes that roughly two-thirds of the base case is earnings growth plus dividend, and that the more interesting asymmetry is in free cash flow rather than earnings — a balance-sheet inflection rather than an earnings one, which is why the medium-horizon stance is a tailwind while the verdict is a Hold. The base of $325 is 6.2% BELOW the street consensus of $346.44, whose lowest published constituent of $320 is already 8.6% above the current price.
• Timing: third-quarter fiscal 2026 results were released 2026-07-30, five days before this dive, and BEAT the consensus adjusted EPS estimate by 3.9% ($3.47 against $3.34) — the fifth consecutive beat, after 3.3%, 0.3%, 3.9% and 4.6% — while simultaneously reporting a GAAP operating LOSS of $2,097.1 million on the project exits announced 2026-06-26 and filed 2026-06-30. The next print is 2026-11-05, 93 days away, on consensus adjusted EPS of $3.60 and revenue of $3,313M, and will complete fiscal 2026. The most recent insider filing is dated 2026-07-02 and is a 6.4626-unit phantom-stock dividend-equivalent accrual to a director; the file contains no open-market transaction by any person. 2026-08-04 carried no company filing.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.