Linde LIN

Reference table

1. What the business is, and how the disclosure works

Linde plc is the world's largest industrial gases company, formed by the 2018 merger of Praxair and Linde AG. Chief executive Sanjiv Lamba; 65,177 employees at 2025-12-31 per the 10-K, "a decrease of 112 employees from 2024, primarily driven by the impact of ongoing cost reduction programs partially offset by acquisitions." Listed on NASDAQ, incorporated in the United Kingdom, reporting in US dollars — profile.currency and inc_a[0].reportedCurrency are both USD, so there is no currency-mixing trap here and no multiple in this dive needed adjusting.

Four reportable segments, and the vendor's seg_prod block is actually the segment table (unlike KLAC's, where product categories were mislabelled as segments). Linde manages industrial gases geographically and Engineering globally:

The tie is imperfect and we report it rather than smooth it. The four vendor lines fall $866M short of reported 2025 revenue. That gap is Linde's "Other" category — the 10-K's segment note carries an Other column covering global helium wholesale and surface technologies, and the 10-Q separately discloses intersegment sales from Helium of $123M in the June quarter. The vendor does not supply the Other line, so any share-of-revenue computed from seg_prod alone is 2.5% too high on each segment. We have restated the shares against total revenue above. This is a mild defect, not a fabrication: nothing double-counts and the residual is stable and shrinking.

Geography (seg_geo) ties exactly. The eight 2025 lines sum to $33.986B against reported revenue of $33.986B — an exact match, which is worth recording after the segment shortfall.

The Germany line is the most under-discussed number in this file. Revenue in the home market of the acquired half of this company has fallen 35% in absolute dollars since 2020, from $3.740B to $2.432B, and from 13.7% of the company to 7.2%. Neither the vendor's derived metrics nor the analyst commentary in this file addresses it. It is the arithmetic of European deindustrialisation showing up in an industrial-gas revenue line, and it happened while the United States grew 43.7%. The geographic mix shift is a first-order fact about this business and it is a favourable one — but it also means the "global diversification" framing is weaker than it was.

The three distribution methods, from the 10-Q's own revenue-recognition note, are the moat. On-site: "Linde constructs plants on or adjacent to these customers' sites and supplies the product directly to customers by pipeline... On-site product supply contracts generally are total requirement contracts with terms typically ranging from 10-20 years and contain minimum purchase requirements and price escalation provisions." Merchant: "Due to the relatively high distribution cost, merchant oxygen and nitrogen generally have a relatively small distribution radius from the plants at which they are produced" — three-to-seven-year supply agreements. Packaged: cylinders, one-to-three-year contracts.

And the number that quantifies it: "As of June 30, 2026, the company estimates the consideration related to future minimum purchase requirements and plant sales was approximately $64 billion... The actual duration of long-term supply contracts ranges up to thirty years. The company estimates that approximately half of the revenue related to minimum purchase requirements will be earned in the next six years and the remaining thereafter."

$64 billion is 1.9 years of revenue, contracted, with minimum-purchase obligations and price escalators attached. There is no field in the vendor payload that contains this or anything like it.

2. Growth, margin, and the conversion engine

Reported and adjusted operating profit for 2024 and 2025 are the 10-K's figures ($8,635M and $8,923M reported; $9,720M and $10,137M adjusted). The vendor's operatingIncome of $8.595B and $8.935B differs by 0.5% and 0.1% respectively — immaterial, and we note the agreement rather than treating it as a defect. Adjusted EPS is derived from the earn_cal quarterly actuals, which sum to $16.45 for 2025 against the est row's $16.428.

Read the operating-profit column against the revenue column and the whole company is visible. Revenue grew 24.8% across five years. Reported operating profit grew 129.4%. Operating margin went from 14.3% to 26.3%. That is the merger synergy programme, the cost-reduction programme, and five years of price attainment landing on a contracted revenue base — and it is why a 4.5% top-line compounder produced a stock that the one knowledge-base claim in this file describes as compounding 12% a year from 1993 to 2024.

Quarterly, using the filings' reported operating profit where the vendor is wrong (see Section 8):

(v) = vendor inc_q, used where no filing figure is in the archive. The 2026 quarters use the 10-Q's stated figures because the vendor's Q1 2026 operating income is wrong by $825M — see Section 8, finding 1.

Five consecutive adjusted-EPS beats, all small (+0.2% to +1.5%). No quarter missed. The second quarter's 9% revenue growth decomposes, per the 10-Q, as "price attainment 2%... Volumes increased sales by 2% primarily driven by the electronics, manufacturing, and chemicals and energy end markets. Currency translation increased sales by 2%... Acquisitions increased sales by 1%. Cost pass-through... increased sales by 1%. Engineering sales increased by 1%." Only four of the nine points are organic price and volume. Currency and cost pass-through are not earnings.

And the margin lever is now working against the company at the gross line. The 10-Q: "Cost of sales, exclusive of depreciation and amortization, was 52.3% and 51.9% of sales for the quarter and six months ended June 30, 2026, respectively, versus 50.7% and 51.0% for the respective 2025 periods." That is a 160 basis point deterioration in the quarter and 90 basis points in the half, attributed to currency translation and cost inflation partially offset by productivity. Reported operating margin still held at 27.5% against 27.7%, because selling and administrative leverage absorbed it — but the raw cost line has turned. For a company whose entire case is 5% revenue converting to 10% earnings, that is the number to watch, and it is the direct opposite of what the vendor's gross-margin series says (Section 8, finding 2).

Research and development is $147M — 0.43% of revenue. This is not a technology company and does not pretend to be.

3. Balance sheet, and the staleness problem

The vendor's most recent balance sheet is 2025-12-31. The most recent filed balance sheet is 2026-06-30, seven months later, and the 10-Q discloses financing activity inside that gap that the vendor cannot see.

Three observations, each checked against the filings.

First, the netDebt field is CORRECT here, and that is worth saying after the KLAC and META cases. Linde carries $0 of short-term investments on all three balance sheets, so the standard omission cannot occur. netDebt of $21.933B = totalDebt $26.989B − cash $5.056B. Verified clean. The vendor's enterpriseValueTTM of $247.309B reconciles to market cap $224.194B + net debt $21.933B + minority interest $1.496B = $247.623B, a residual of 0.13%. We accept the vendor EV.

Second, capitalLeaseObligations drops to $0 in the 2025 balance sheet after $986M and $942M in the prior two years. This is the identical defect class seen on KLAC. Linde has not extinguished its leases in a year in which it acquired businesses and built plants. The effect is that the year-on-year totalDebt comparison from $22.609B to $26.989B is not on a consistent basis — the increase is understated by roughly $1.0B, not overstated. We use the figure as a floor.

Third, the balance sheet is seven months stale and the 10-Q tells us what happened. "Cash provided by debt was $1,645 million for the six months ended June 30, 2026... During the six months ended June 30, 2026, Linde issued €1.6 billion of Euro-dominated notes and repaid $725 million of U.S. dollar-denominated notes." The May 2026 issuance was three tranches: €600M floating due 2028, €500M of 3.200% notes due 2030, €500M of 3.800% notes due 2036. The 10-Q's fair-value note gives the long-term-debt carrying value directly: $23,152M at 2026-06-30 against $22,479M at 2025-12-31.

Note that the filing's $22,479M long-term-debt carrying value at 2025-12-31 does not equal the vendor's longTermDebt of $20,683M. The difference of $1,796M is almost certainly the current portion of long-term debt, which the filing's fair-value disclosure includes in the "long-term debt portfolio" and the vendor's balance sheet classifies inside short-term debt. We do not force a reconciliation and we do not restate total debt on it, because total debt as vendor-computed already contains both pieces. What we do take from the filing is the direction and the magnitude: debt rose roughly $1.6-1.7B in the first half of 2026, so a working net-debt figure at 2026-06-30 is meaningfully above $21.933B, and every leverage ratio in this dive computed on the 2025 balance sheet is conservative in the wrong direction — i.e. it flatters.

The leverage picture on the vendor's own basis:

Goodwill and intangibles are $39.798B — 45.8% of total assets (the vendor's intangiblesToTotalAssetsTTM says 44.7%, computed on a slightly different asset base). All of it is Praxair/Linde AG merger accounting. This is the single reason return on invested capital is 9.93% while return on equity is 18.87% and return on tangible assets is 14.91%. We use ROIC as the honest measure. Sub-10% ROIC on a business earning a 26% operating margin is entirely a denominator story, and readers should understand that the operating business earns far more than 9.93% on the capital it actually deploys — but the capital shareholders paid for includes the merger premium, and ROIC is the measure that remembers that.

Liquidity is not a question. "The company maintains a $5.0 billion and a $1.5 billion unsecured and undrawn revolving credit agreement with no associated financial covenants. No borrowings were outstanding under the credit agreements as of June 30, 2026."

4. Cash flow and capital returns

Free cash flow is the weak line in this file and the first half of 2026 makes it explicit. Operating cash flow rose 3% year on year to $4,511M — "driven primarily by higher net income" — but capital expenditure rose $253M to $2,780M, "primarily due to investments in new plant and production equipment for backlog growth requirements." Free cash flow therefore FELL 6.2%, from $1,845M to $1,731M, in a half in which sales grew 9% and adjusted EPS grew 10%. Capex is now 1.44x depreciation and 15.5% of revenue on a trailing basis, against 0.75x and 9.5% in 2022.

This is not deterioration; it is the growth model. The $8.1B sale-of-gas backlog of large plants under construction is being paid for now and generates contracted revenue for the following decade. But it means the 44.1x price-to-free-cash-flow multiple and 2.22% free-cash-flow yield are the honest cash valuation, and both are worse than the 27.1x earnings multiple suggests. A reader anchoring on earnings will overestimate the cash this business returns.

Buyback, and it has halved. Net share purchases were $1,656M in the first half of 2026 against $2,207M in the first half of 2025 — a 25% reduction. Under the October 2023 $15.0B programme, $9.4B has been repurchased and $5.6B remains authorised at 2026-06-30. Annualising the first half gives roughly $3.3B, or 1.48% of market capitalisation, down from 2.05% in 2025.

Dividend, and a vendor labelling trap identical to KLAC's. The vendor reports lastDividend: 6.2 and dividendPerShareTTM: 6.2, yielding 1.28%. That $6.20 is the trailing-twelve-month total, not the last declared quarterly dividend. The 10-Q states the actual rate: "a 7% increase in quarterly dividends per share from $1.50 per share to $1.60 per share." Two quarters at $1.50 plus two at $1.60 is exactly $6.20. The forward run-rate is $6.40, a 1.32% yield, not 1.28%. Payout ratio on adjusted EPS is 37.1%; the vendor's dividendPayoutRatioTTM of 39.5% is computed on GAAP and is also correct on its own basis.

Total shareholder yield is therefore about 2.8% — 1.32% dividend plus roughly 1.48% net buyback. That is the real cash return, and it is a material part of a mid-teens total-return case.

5. Valuation — priced in or room?

At $484.64 (market cap $224.194B, enterprise value $247.309B):

Two estimate-block problems, both stated and both handled.

The FY2029 and FY2030 rows are non-monotonic and thin. Consensus adjusted EPS is $25.487 in FY2029 and $25.05 in FY2030 — a decline, on three analysts each. Revenue over the same two years rises from $43.530B to $44.140B. There is no coherent business story in which revenue rises 1.4% and EPS falls 1.7% for a company with rising margins and a shrinking share count. Both rows are excluded from every conclusion in this dive.

est.ebitdaAvg and est.ebitAvg carry a fabrication signature and are rejected outright. Across all eight years in the file, ebitdaAvg / revenueAvg = 47.43% and ebitAvg / revenueAvg = 39.98%, to four significant figures, every single year from 2023 to 2030. Actual 2024 EBITDA was $12,585M on $33,005M of revenue — 38.1%, not the row's $15.812B and 47.9%. Actual 2025 reported operating margin was 26.3%, not 39.98%. These are fixed-margin fills, not analyst estimates, and they overstate EBITDA by roughly 25% and EBIT by roughly 50%. Nothing in this dive touches them; all forward valuation runs on epsAvg and revenueAvg, per the data contract.

Peer context. The vendor peer set is largely unusable — Sherwin-Williams, Ecolab, Southern Copper, H.B. Fuller and Sensient are not industrial-gas comparables. The one genuine peer is Air Products and Chemicals (APD, $65.6B market cap at $294.71). Linde is 3.4x its size. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn, and the 10-K's competition discussion does not name competitors in a way that produces a usable set.

5a. What today's price assumes (the inversion)

At $484.64 — 27.1x FY2026 consensus adjusted EPS and 22.5x FY2028 — the price embeds the following falsifiable claims:

• Revenue reaches $36.156B in 2026, $37.918B in 2027 and $40.038B in 2028 — 5.6% compound. (Consensus-derived; 8, 18 and 20 analysts.) The near check is the 2026-10-30 print at $9.106B, which would be +5.7% year on year against the +9.3% just delivered — i.e. consensus itself expects the growth rate to halve within one quarter, because currency and acquisitions do not repeat.
• Adjusted EPS reaches $21.555 by 2028 — 9.0% compound off $16.43 in 2025 — which requires the conversion engine to keep running at exactly its historical rate. (Consensus-derived; 9 analysts on the FY2028 line.) This is the most fragile assumption in the price, because the raw cost line has turned: cost of sales excluding depreciation was 52.3% of sales in the June quarter against 50.7% a year earlier, per the 10-Q. Half the historical conversion came from margin, and margin at the cost-of-sales line is now moving the wrong way.
• The market keeps paying roughly 22-25x on the forward year at every date between now and 2028. (Our number.) At 20x FY2028E the stock is $431; at 26x it is $560. The multiple, not the earnings, is where the outcome lives — and it is the one thing that has actually moved over the last twelve months, during which earnings beat five times and the stock returned 5.5%.
• Capital expenditure of roughly 15% of revenue earns an adequate return. (Our framing; the $8.1B backlog and the 1.44x capex-to-depreciation ratio are filing figures.) Free cash flow fell 6.2% year on year in the first half while earnings rose 10%. The price is being set on earnings; the cash is being spent on plants that will not produce revenue for years.
• Buyback continues at roughly $3.3B a year. (Our annualisation of the H1 figure.) $5.6B remains authorised — about twenty months at the current pace. Roughly 1.5 points of the ~10% EPS growth is share-count arithmetic, and it is the leg with a visible expiry date.

5b. The return bridge (why the multiple moves)

Expected return over a two-year horizon decomposes as: adjusted EPS growth (+20.5%, from FY2026E $17.895 to FY2028E $21.555) + multiple drift (COMPRESSION, from 27.1x on the forward year to 25x, −7.7%) + shareholder yield (+2.8% a year, ~5.7% compounded) ≈ +18% to +19% total, or roughly +9% annualised.

Our base case assumes mild multiple compression and we should say why, because it is not a rolldown artefact. Today the market pays 27.1x on the next calendar year. Two years out, if it pays 25x on the then-next year, that is a genuine de-rating of about 8%. We assume it for one reason: the market has been de-rating this stock steadily for twelve months in the face of five consecutive beats, and nothing in the fundamentals has appeared to stop it. The twelve-month total return of +5.5% against +24.3% for the index, on earnings that grew about 10%, means the multiple contracted roughly 15% in a year. Assuming that process fully halts and reverses is a stronger assumption than assuming it decays.

Almost all of the expected return is earnings growth plus dividends, and the multiple is a small headwind. That is a healthier structure than a name needing re-rating. If the multiple simply HELD at 27.1x on FY2028E the price would be $584 (+20.5%); if it compressed to 20x the price would be $431 (−11.1%). The spread between those is the entire uncertainty and it is not resolvable from the data in this file.

The bull case at $652 requires no multiple expansion — it holds today's 27.1x forward multiple against the FY2028 consensus high of $23.292, plus a point of drift. The bear case at $418 is a de-rating to 21x against the FY2028 consensus low of $19.901, which is roughly where the stock traded at its 52-week low of $389.38 plus the two years of earnings growth since.

5c. Variant perception (where we differ, what would surprise)

• We differ from the street on almost nothing about the business, and we say so. 25 buy, 3 hold, 0 sell, consensus $557.67. Our base of $539 is 3.3% BELOW consensus. The knowledge base contributes one claim. When our arithmetic lands under the crowd's on a name with no independent conviction lane, we have no edge, and no-edge names default toward Watch.
• We differ on the direction of gross margin, and this is the one place the vendor and the filing point opposite ways. The vendor's grossProfitMarginTTM is 45.64% and its annual series shows gross margin rising from 36.6% (2024) to 43.3% (2025) — 672 basis points of expansion. The 10-K says cost of sales excluding depreciation was 51.2% of sales in 2025 against 51.9% in 2024 — 70 basis points of improvement, one-tenth of what the vendor implies — and the 10-Q says the ratio deteriorated 160 basis points in the June quarter. The filing wins. Anyone screening this name on vendor gross margin is looking at a manufactured trend (mechanism in Section 8). Watchable number: cost of sales excluding depreciation as a percentage of sales in the 2026-10-30 report. A third consecutive quarter above 52% attacks the conversion engine directly.
• We differ on how much of the ~10% EPS growth is durable. Roughly 1.5 points is buyback with $5.6B of authority left, and a meaningful slice of the rest is price attainment running at 2% against cost inflation the company itself says is adverse. Strip buyback and currency and the organic earnings engine is closer to 6-7% than 10% — which at 27.1x is a materially different security.
• Positive surprise that would force a re-rate: an October print with cost of sales excluding depreciation back below 51% of sales while volumes stay at +2% or better, which would show the margin turn was currency and not structural; or a step-up in the electronics volume contribution, which is the one end market with a genuine slope behind it; or a new buyback authorisation ahead of the current programme's exhaustion.
• Negative surprise that would break the thesis: volume growth going negative — it is +2% and two of the nine points of headline growth are currency, which reverses; free cash flow declining for a full year while capex stays at 15% of revenue; or a further leg down in Germany and EMEA, where revenue has already fallen 35% since 2020 in the largest European country. None of these hits a stock with a valuation cushion: 27.1x forward and 44.1x free cash flow is not a floor.

Synthos fair values

All three anchors are multiples of the FY2028 consensus adjusted-EPS distribution (mean $21.555, low $19.901, high $23.292, 9 analysts), cross-checked against FY2027E of $19.616.

• Bear ~$418 — 21x the FY2028 consensus LOW of $19.901. Cross-check: 21.3x FY2027E; 7.5% below the 52-week low of $389.38 grown two years at the consensus rate. The scenario: cost inflation persists, currency reverses, volume growth stalls at zero, the buyback slows as authority runs down, and the market pays a utility multiple for a utility-like business. −13.7%.
• Base ~$539 — 25x the FY2028 consensus MEAN of $21.555. Cross-check: 27.5x FY2027E; 30.1x FY2026E. Sensitivity, stated openly: 23x gives $496 — 2.3% above spot — and 27x gives $582, above the street. The whole answer lives in a 23-27x band and we claim no more precision than that. The scenario: revenue compounds 5.6%, the conversion engine delivers 9% EPS growth, the multiple drifts down modestly as the market continues to reprice slow compounders. +11.2%, plus about 5.7% of compounded shareholder yield over the period.
• Bull ~$652 — 28x the FY2028 consensus HIGH of $23.292. Cross-check: 33.2x FY2027E. This bull case requires a modest re-rating AND an earnings beat, which is why it is the least likely of the three: electronics volumes accelerate, the $8.1B backlog commissions early, Europe stabilises, and the market decides that ~$64B of contracted revenue with thirty-year tails deserves a scarcity premium in a volatile market. +34.5%.

Base is 11.2% above spot; asymmetry roughly 2.5:1 to the upside (13.7% down, 34.5% up). That ratio is genuinely attractive and it is the best argument for this name. What holds it at Watch rather than Buy is that the base case sits below the street, the conviction lane is one claim, the bear case is only 13.7% away, and the multiple has been compressing for a year without an identifiable stopping point.

6. Knowledge base — one claim

Raw entity hits: 1. Entity matches used: 1. Text matches used as colour: 1. Discarded for collision: an entire bare-ticker sweep.

The primary search covered Linde, LIN, Linde plc, industrial gas, industrial gases and Air Products across all 51,928 distilled claims. The bare ticker LIN was discarded in full — as a three-letter string it collides with ordinary word fragments in thousands of claims and produced no genuine hit that the name search did not already return. This is the homograph discipline applied, and it matters here more than for most names.

> 2026-03-05 · bullish · conviction 68 · horizon: thesis · entities: Linde, LIN, Lindy PLC · channel: we_study_billionaires · no named speaker

> "Linde, world's largest industrial gas company, is a boring stable compounder with durable moat; compounded 12%/yr 1993-2024 vs S&P's 8%."

One adjacent claim, used as supporting colour only, does not name the company but names the category:

> 2023-10-12 · bullish · conviction 80 · horizon: principle · channel: we_study_billionaires · no named speaker

> "Prefer stable, predictable industries (elevators/escalators, spirits, industrial gases) where sustainable growth rate is estimable; skip exciting-but-unknowable TAM stories — the too-hard pile."

What this lane is, stated without inflation. It is one claim, from one channel, with no named speaker, which under the 4-lane attribution policy is the weakest sourcing we accept — a channel tells you where something was said, not who said it or what their record is. Note also that the entity list contains "Lindy PLC", which is a distillation error for "Linde plc"; the claim is unambiguously about the right company, but the entity tagging on this record is imperfect and we flag it.

The claim's substance is a trailing-return observation — 12% a year from 1993 to 2024 against the S&P's 8% — which is precisely the kind of backward-looking compounder framing the house selection philosophy is sceptical of. It is not a forward view. It contains no number about 2026, no view on the multiple, and no falsifiable claim. There is no bearish claim on Linde anywhere in the knowledge base, and the absence of a bear case in a lane of one is not evidence of anything.

No claim in this lane carries a management speaker_role, so the half-weight management discount does not apply. No promoter-class voice appears.

Conclusion, stated as a finding rather than dressed up: the Synthos knowledge base has no meaningful view on Linde plc. Conviction Low, breadth 1, and the thinness is a direct input to the Watch verdict. This is a name where the quantitative work has to stand entirely on its own.

7. Data integrity — what we rejected from the vendor file and why

Nine findings. Two of them are large. Note that LIN's file is clean where it matters most for enterprise value and is defective where it matters most for margin analysis — the reverse of the usual pattern.

1. inc_q operating income for Q1 2026 is overstated by $825M — REJECTED and replaced with the filing. The vendor reports $3.264B of operating income for the quarter ended 2026-03-31 on revenue of $8.781B, a 37.2% operating margin. The 10-Q filed 2026-05-01 states: "Reported operating profit for the first quarter of 2026 was $2,439 million, or 27.8% of sales." The vendor figure is 33.8% too high. It is also internally impossible: an operating margin of 37.2% with a 23.5% effective tax rate and $8.781B of revenue implies net income near $2.4B, but the vendor's own net income for the quarter is $1.857B and its EPS of $3.98 matches the filing exactly. The corruption is confined to the operating-income line; revenue, net income, EPS and share count all tie to the filing. The vendor reports $3.264B; the 10-Q shows $2,439M; we use $2,439M, and every trailing operating-profit and EBITDA figure in this dive is computed on the filing's number.

2. The gross-margin series is computed on alternating bases and manufactures 672 basis points of margin expansion — REJECTED in full. The vendor's annual gross margin goes from 36.61% in 2024 to 43.33% in 2025. The 10-K says: "Cost of sales, exclusive of depreciation and amortization, was 51.2% and 51.9% of sales, in 2025 and 2024, respectively" — i.e. 48.8% versus 48.1%, an improvement of 70 basis points, one-tenth of the vendor's implied step. The mechanism is visible in the quarterly series: 2025's four quarters print 37.5%, 49.3%, 38.0%, 48.1% — alternating between a basis that includes depreciation in cost of sales and one that excludes it. All four 2024 quarters print on the inclusive basis (37.0%, 36.4%, 36.8%). The 2025 annual figure is the sum of two quarters on each basis and corresponds to no real accounting. ratios_ttm.grossProfitMarginTTM of 45.64% is the same blend and is rejected with it. No gross-margin figure from the vendor appears anywhere in this dive; we use the filing's cost-of-sales-excluding-depreciation ratio throughout. This is a direction-and-magnitude error of the JNJ class and it is the most consequential defect in the file for anyone screening on margin trend.

3. est.ebitdaAvg and est.ebitAvg carry a fixed-margin fabrication signature — REJECTED outright. ebitdaAvg / revenueAvg equals 47.43% and ebitAvg / revenueAvg equals 39.98% in every one of the eight years from 2023 to 2030, to four significant figures. Actual 2024 EBITDA per the 10-K was $12,585M (38.1% of revenue) against the row's $15.812B; actual 2025 reported operating margin was 26.3% against the row's 39.98%. These rows overstate EBITDA by roughly 25% and EBIT by roughly 50% and are not analyst output. Per the data contract, all forward valuation runs on epsAvg.

4. capitalLeaseObligations drops to $0 in the 2025 balance sheet — year-on-year debt comparison flagged. The field reads $942M (2023), $986M (2024), $0 (2025). Linde has not extinguished its leases. totalDebt of $26.989B therefore understates by roughly $1.0B and the 2024-to-2025 increase is understated by the same. We treat $26.989B as a floor.

5. The balance sheet is seven months stale and misses a €1.6B note issuance — stated, and the direction given. The most recent vendor balance sheet is 2025-12-31; the most recent filed one is 2026-06-30. Inside that gap the 10-Q discloses $1,645M of net cash provided by debt, comprising a May 2026 issuance of €1.6 billion in three tranches (€600M floating due 2028, €500M of 3.200% due 2030, €500M of 3.800% due 2036) against a $725M repayment of 3.20% notes in January 2026. The long-term-debt carrying value rose from $22,479M to $23,152M. Every leverage ratio in this dive is computed on the 2025 balance sheet and therefore understates leverage at 2026-06-30. We say so rather than estimating a figure the data does not support.

6. Vendor longTermDebt ($20.683B at 2025-12-31) does not equal the filing's long-term-debt carrying value ($22,479M) — noted, not reconciled. The $1,796M difference is almost certainly the current portion of long-term debt, which the vendor classifies inside shortTermDebt ($6.306B) and the filing's fair-value note includes in the "long-term debt portfolio". Total debt is unaffected, so no ratio in this dive changes. We record the discrepancy so that no reader takes the vendor's long-term/short-term split as the filing's.

7. seg_prod does not sum to revenue — a $866M "Other" segment is missing. The four supplied segment lines total $33.120B against reported 2025 revenue of $33.986B, a shortfall of 2.5%. The gap is Linde's Other category (global helium wholesale, surface technologies), which the 10-K's segment note carries and the vendor does not supply. The 10-Q separately discloses intersegment helium sales of $123M in the June quarter, confirming the category exists. Segment shares computed against the vendor's own sum are 2.5% too high each; we restate all shares against total revenue. Nothing double-counts; this is an omission, not the XOM-class 2.05x over-sum.

8. seg_geo ties exactly — a clean check worth reporting. The eight 2025 country lines sum to $33.986B against reported revenue of $33.986B. Verified clean.

9. lastDividend: 6.2 is the trailing-twelve-month total, not the last declared dividend — corrected from the filing. The 10-Q states the quarterly rate rose from $1.50 to $1.60, a 7% increase. Two quarters at each is exactly $6.20. The forward run-rate is $6.40 — a 1.32% yield, not the 1.28% the field implies.

Not defects, correctly reported and independently confirmed. netDebt of $21.933B is arithmetically correct because Linde carries $0 of short-term investments, so the standard omission cannot occur — this is the one name in this batch where the net-debt field passes without adjustment, and we state it because the defect class is common enough that its absence is a finding. enterpriseValueTTM of $247.309B reconciles to market cap plus net debt plus the $1.496B minority interest with a 0.13% residual, so the vendor EV is accepted and used. FY2025 and FY2024 operating income ($8.935B and $8.595B) agree with the 10-K's reported $8,923M and $8,635M to within 0.5%. Diluted EPS of $3.98 and $4.15 for the two 2026 quarters match the 10-Q exactly. The effective tax rate of 22.23% is consistent with the 23.5% and 24.0% quarterly rates the filings report. Employee count (65,034 in profile versus 65,177 at 2025-12-31 in the 10-K) differs by 0.2% and is immaterial.

quote.yearHigh/yearLow ($548.20 / $387.78) disagree with tech.hi52/lo52 ($546.64 / $389.38) by 0.3% and 0.4%. We use tech throughout, per house rule, for internal consistency with the moving averages and relative-return figures computed from the same close series.

Vendor composite rating — mechanism disclosed, partially rejected. The rating block gives B / 3 overall, with 1 out of 5 on debt-to-equity, 1 on price-to-earnings and 1 on price-to-book, against 5s on return on equity and return on assets and a 4 on discounted cash flow. The price-to-earnings and price-to-book 1s are judgements about valuation, not about the business, and we agree with them — 28.1x trailing and 5.74x book is expensive. The debt-to-equity 1 is harder to defend: debtToEquityRatioTTM is 0.717x, which for a business with 37.9x interest coverage, $6.5B of undrawn covenant-free revolvers and contracted revenue is not a distressed reading. We reject the debt-to-equity sub-score as a lens and accept the two valuation sub-scores.

Non-equity tripwire — checked and passed. LIN is ordinary shares of a UK-incorporated plc, NASDAQ-listed, reporting in USD. The price of $484.64 is not par-like; beta is 0.723, above the sub-0.3 typical of a fixed-income-like instrument; the dividend is variable and was raised 7% this year; volume was 2.83M shares (~$1.37B of turnover); the 52-week band of $389.38 to $546.64 is a 40% range. This is common equity. Currency check: profile.currency = USD and inc_a[0].reportedCurrency = USD. No ASML-class currency mixing; all multiples are like-for-like.

8. Technicals

• Price $484.64. −11.3% from the 52-week high of $546.64; +24.5% above the low of $389.38. Position within the annual range: 61st percentile.
• Moving averages, split decision: −5.6% below a 50-day average of $513.38, which is falling; +2.2% above a 200-day average of $474.36, which is rising. The 200-day is the only support between here and the low $400s and it is 2.1% away.
• RSI 32.2 — the weakest reading in this batch outside outright broken charts, and approaching oversold. MACD −7.88 — negative.
• Maximum drawdown from peak over the trailing year: −11.3%, i.e. the current price is the worst point of a shallow drawdown. This is not a crash; it is a grind.
• Relative performance, and this is the whole story: 3-month −1.8% vs SPY +7.6% and QQQ +7.7%; 6-month +5.3% vs SPY +11.1% and QQQ +15.7%; 12-month +5.5% vs SPY +24.3% and QQQ +30.8%. Linde underperformed the S&P by roughly nineteen points over twelve months while beating consensus adjusted EPS in every quarter it reported. With adjusted EPS up roughly 10% over that period, the forward multiple contracted by about 15%. That is the entire explanation for the last year of this stock and it is a valuation event, not an operating one.
• Sentiment: 25 buy, 3 hold, 0 sell across 28 analysts; consensus target $557.67 (+15.1%), median $560, high $612 (+26.3%), low $525 — 8.3% ABOVE spot. Not one published target sits below the current price. That is a well-covered name with unanimous directional agreement, and it is a reason for caution rather than comfort: there is nobody left to upgrade.

Today's move

LIN closed 2026-08-04 at $484.64, up 0.87% or $4.18 from $480.46. It opened at $481.99, traded $478.51 to $485.60, and closed near the day's high on 2.83M shares. The session was a semiconductor-led melt-up; a 0.87% move in a 0.723-beta industrial name on such a day is market beta and nothing more. No company-specific news appears in this file for 2026-08-04. The last company event was the second-quarter release on 2026-07-31, four days earlier, at which adjusted EPS of $4.50 beat the $4.49 consensus by a penny and revenue of $9,289M beat $9,016M by 3.0%.

The honest read on entry. RSI at 32.2 with the 200-day 2.1% below is the most attractive technical setup this name has offered in a year, and the shallow 11.3% drawdown means the stock has not broken. But the same arithmetic that makes the chart interesting makes the valuation uninteresting: 27.1x forward is what the de-rating has already produced, and there is no rule that says it stops here. A move to the 200-day at $474 is a 2% event. A move to the 52-week low at $389 is a 20% event and would put the stock at 21.7x FY2026E — a multiple this business has traded at before.

What would change the answer: a price near $430, roughly 20x FY2028E and 24x FY2026E, where our $539 base becomes +25% and the bear case sits only 3% below. That is 11.3% below today's close and well inside the range this stock traversed in the last twelve months.

9. Insiders

The eight transactions in the file are not a signal in either direction and we report them as such.

Every transaction is either a routine director deferred-stock-unit award at zero price, an option or unit exercise coded M-Exempt, or shares withheld by the issuer to satisfy tax on vesting (F-InKind). F-InKind shares are retained by the company and never reach the market; they are not sales. The file therefore contains zero open-market purchases and zero open-market sales. That is a neutral reading and it carries none of the information content of a discretionary purchase into a 32-RSI drawdown, which is the transaction that would have mattered here and did not occur.

10. Verdict, kill-criteria and flip conditions

Watch.

What is genuinely excellent, and none of it is in dispute: approximately $64 billion of remaining performance obligations with contract durations up to thirty years; on-site supply contracts of 10-20 years with minimum purchase requirements and price-escalation provisions; reported operating profit up in every one of the last five years, from $3.889B to $8.923B on revenue that grew 24.8%; an operating margin that went from 14.3% to 26.3% reported and 29.8% adjusted; five consecutive adjusted-EPS beats; interest coverage of 37.9x; $6.5B of undrawn revolvers with no financial maintenance covenants; a diluted share count down 12.9% in five years with $5.6B of buyback authority remaining; a dividend raised 7% to $1.60 a quarter; and a beta of 0.723 in a batch where that is scarce.

What we are declining to pay for: 28.1x trailing adjusted earnings, 27.1x FY2026 consensus, 6.3x sales and 44.1x free cash flow for a business consensus expects to grow revenue 5.6% compound. A PEG of 3.05 on the vendor's own arithmetic. A base fair value of $539 that is 11.2% above spot and 3.3% below the street's $557.67, in a coverage universe where not one of 28 analysts has a target below the current price. A cost-of-sales ratio that deteriorated 160 basis points year on year in the June quarter. Free cash flow that fell 6.2% in the first half while earnings rose 10%. And a knowledge-base lane of exactly one channel-attributed claim.

The distinction that matters. We are not saying Linde is expensive because it is a bad business. It is one of the three or four most structurally protected businesses in this batch. We are saying that a 5%-growth business at 27x forward has to convert flawlessly for a decade to justify the price, that the conversion levers (price, productivity, buyback) are all visibly maturing, and that the market has spent twelve months telling us it agrees. In the Synthos frame, a name whose fair value sits below consensus with a one-claim conviction lane is a Watch by construction.

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

• Volume growth turning negative in any quarter. Volume is +2% and two of the nine headline points are currency, which reverses. Negative volume in an industrial-gas company is an industrial-recession signal and the multiple has 30% of downside in it.
• Cost of sales excluding depreciation above 53% of sales for two consecutive quarters. It was 52.3% in the June quarter against 50.7% a year earlier. That would end the conversion story.
• A full year of declining free cash flow with capex still above 15% of revenue. The first half already declined 6.2%.
• The buyback stopping or the $5.6B remaining authority not being replaced — that removes roughly 1.5 points from a ~10% growth rate.
• EMEA revenue falling below $8.0B on a full-year basis, or Germany falling below $2.2B. The European base has already shrunk 35% since 2020.

Pre-registered FLIP conditions — what would take this to Buy — Core:

• Price at or below approximately $430 (roughly 24x FY2026E, 20x FY2028E) with the fundamental picture intact. This is the single most likely path to a Buy and it is 11.3% below today's close. At $430 the base case is +25% and the bear case is 3% away — genuine asymmetry.
• The 2026-10-30 print showing cost of sales excluding depreciation back below 51% of sales with volume still at +2% or better, which would demonstrate the margin turn was currency translation rather than structural cost inflation.
• A sustained reclaim of the 50-day average at $513.38 alongside the above, converting a twelve-month de-rating into a base.
• A new buyback authorisation materially above the $5.6B remaining, which would extend the visible share-count runway beyond twenty months.
• Genuine name-level knowledge-base conviction appearing — a forward claim about Linde rather than a trailing-return observation. There is currently none, and that alone is worth roughly one tier.

Where LIN fits in the Synthos Framework Portfolio. The defensive-industrial / low-beta sleeve, at 0% today with a 1.5-2.5% target on a fill near $430. On batch overlap: LIN is the lowest-correlation name in this batch of twelve — a 0.723-beta industrial-gas compounder sits in a different risk bucket from ADI, QCOM, IBM or SCHW, and it is the natural ballast holding if one is wanted. Its closest overlap is with UNP and NEE as regulated-or-contracted cash-flow assets, and among those three the question is which contracted stream is cheapest per unit of growth. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $484.64.

Single biggest risk: the multiple keeps compressing. Over twelve months Linde grew adjusted EPS roughly 10%, beat consensus five times out of five, raised its dividend 7%, and returned 5.5% against the index's 24.3%. Nothing operational explains that. What explains it is that the market re-rated a slow compounder downward by about 15%, and there is nothing in the data in this file that identifies where that process stops. At 27.1x forward there is a great deal of multiple left. A move to 22x FY2026E — a multiple this stock traded at within the last three years — is $394, essentially the 52-week low.

Most fragile assumption in the price: that ~9% adjusted EPS growth continues through FY2028. Every other assumption is a volume or price assumption, observable quarterly. This one is a composite of three levers that are each late in their run — price attainment of 2% against cost inflation the company calls adverse; productivity savings after five years of a cost-reduction programme that has already taken headcount down; and a buyback with twenty months of authority left. If EPS grows 6% instead of 9%, FY2028 adjusted EPS is $19.60 rather than $21.56, and at 25x that is $490 — six dollars above today's price, three years from now.

Provenance & disclosures

• Traceability: 1 tagged knowledge-base claim names Linde (raw entity hits 1, entity matches 1, text matches 0 after collision filtering, discarded for collision: one full bare-ticker sweep on LIN; breadth 1, net conviction positive-negligible). Source: we_study_billionaires (2026-03-05, bullish, conviction 68, horizon thesis, no named speaker — the trailing-compounder observation). One adjacent principle-level claim from the same channel (2023-10-12, bullish, conviction 80, no named speaker) naming industrial gases as a category is used as supporting colour only and is not counted in the lane. The single naming claim is a trailing-return statement, not a forward view, and carries no falsifiable number about 2026 or beyond — stated as a finding rather than padded. The claim's entity list contains the distillation artefact "Lindy PLC", flagged. No claim in the lane carries a management speaker_role, so the half-weight management discount does not apply; no promoter-class voice appears. Searches: Linde, Linde plc, industrial gas, industrial gases, Air Products, and a bare-ticker sweep on LIN which was discarded in full for homograph collision with ordinary word fragments. All quotes are verbatim from the stored claim text.
• Data as-of: fundamentals — annual income statement, balance sheet and cash flow to 2025-12-31 from the vendor payload; quarterly income statement to 2026-06-30; most recent filing-verified financials 2026-06-30, from the 10-Q filed 2026-07-31 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873601 = 2026-08-04T20:00:01Z ($484.64, +0.87%; 50-DMA $513.38; 200-DMA $474.36; RSI 32.2; MACD −7.88) · knowledge-base claims 2026-08-04. Fiscal note: Linde's fiscal year is the calendar year. All figures come from the Synthos vendor data file for LIN or from the SEC filings in the LIN archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2026-02-25 (fiscal 2025); 10-Q filed 2026-05-01 (March 2026 quarter); 10-Q filed 2026-07-31 (June 2026 quarter); 8-K filed 2026-05-13; 8-K filed 2026-07-30; 8-K filed 2026-07-31 (second-quarter results). The LIN archive is flagged tables: false in the filings index — the financial-statement tables were not preserved, so every filing figure in this dive comes from narrative prose (MD&A, the notes, and the risk factors) rather than from a statement table. The 2026-07-31 8-K is present as the cover document only; Exhibit 99.1, the press release carrying the second-quarter financials, was not extracted and the file contains no numbers. Second-quarter figures are therefore taken from the 10-Q filed the same day, which does carry them in prose.
• Where the filings contradicted or corrected the vendor (detailed in Section 7): Q1 2026 operating income restated from the vendor's $3.264B to the 10-Q's $2,439M — an $825M, 33.8% vendor overstatement, with every trailing operating-profit and EBITDA figure in this dive recomputed on the filing; the entire gross-margin series rejected, because the vendor shows margin rising 672 basis points from 36.61% to 43.33% between 2024 and 2025 while the 10-K states cost of sales excluding depreciation was 51.2% versus 51.9% of sales — a 70 basis point improvement — and the 10-Q states the ratio deteriorated to 52.3% from 50.7% in the June quarter, with the mechanism traced to alternating inclusion of depreciation across the 2025 quarters (37.5%, 49.3%, 38.0%, 48.1%); est.ebitdaAvg and est.ebitAvg rejected outright on a fixed-margin fabrication signature (47.43% and 39.98% of revenue in all eight years, against actual 2024 EBITDA of $12,585M / 38.1% per the 10-K); the balance sheet identified as seven months stale, missing the May 2026 €1.6B three-tranche note issuance and $1,645M of net debt provided in the first half; capitalLeaseObligations dropping to $0 in 2025 making the total-debt trend non-comparable; seg_prod failing to sum to revenue by $866M because the Other segment is not supplied; and lastDividend of $6.20 identified as the TTM total against a filing-stated forward run-rate of $6.40. Where vendor and filing AGREED — worth recording: FY2025 and FY2024 operating income within 0.5% of the 10-K's $8,923M and $8,635M; Q1 and Q2 2026 diluted EPS of $3.98 and $4.15 exactly; netDebt of $21.933B arithmetically correct because Linde carries $0 of short-term investments — the one name in this batch where that field needed no repair; enterpriseValueTTM reconciling to 0.13%; seg_geo summing exactly to reported revenue; and employee count within 0.2%.
• Basis note: analyst estimates in est.epsAvg are non-GAAP adjusted EPS — verified because the FY2025 row of $16.428 matches the sum of the four earn_cal quarterly actuals ($3.95 + $4.09 + $4.21 + $4.20 = $16.45) and does not match GAAP diluted EPS of $14.59. The GAAP-to-adjusted gap is material at 12.6% — Linde excludes Linde AG merger purchase-accounting impacts, cost-reduction charges and pension settlements. Forward multiples in this dive use consensus adjusted EPS, and the trailing multiple is shown on both bases (28.1x adjusted, 31.3x GAAP). The choice of basis moves the trailing multiple by 3.2 turns and readers should not compare our 27.1x forward figure to a GAAP-basis multiple elsewhere.
• Estimate coverage: 14 analysts on FY2026 EPS but only 8 on FY2026 revenue; 10 and 18 on FY2027; 9 and 20 on FY2028. The FY2029 EPS row rests on 3 analysts and FY2030 on 3, and the two are non-monotonic ($25.487 falling to $25.05 on rising revenue) — both are excluded from every conclusion.
• Peer note: the vendor peer set contains one genuine comparable, Air Products and Chemicals (APD, $65.6B at $294.71), alongside eight names that are not (Sherwin-Williams, Ecolab, Southern Copper, Sociedad Química y Minera, Element Solutions, Avient, H.B. Fuller, Hawkins, Sensient). No peer multiples are supplied in the file, so no peer-multiple comparison is drawn.
• Fair-value caveat: the $418 / $539 / $652 anchors are multiples of the FY2028 consensus adjusted-EPS distribution — 21x the low of $19.901, 25x the mean of $21.555, and 28x the high of $23.292. Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 23x gives $496, only 2.3% above spot; 27x gives $582, above the street. The whole answer lives in a 23-27x band and no greater precision is claimed. The bear case assumes a de-rating; the bull case assumes both a modest re-rating and an earnings beat, which is why it is the least likely of the three.
• Timing: second-quarter 2026 results were released 2026-07-31, four days before this dive, so the quarter and the accompanying 10-Q are fully reflected. The next print is 2026-10-30, 87 days away — there is no near-term company-specific catalyst, which is an explicit input to the Watch verdict. 2026-08-04 was a semiconductor-led up-day and LIN's +0.87% is treated as market beta.
• 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.