1. Core judgment in one sentence
Industrial-gas profits depend more on local delivery and contract structure than molecular scarcity; large expansion projects can still consume the earnings of established operations.
This judgment can be overturned: if new gas-production capacity consistently earns excess returns without reliable power, logistics, customer qualification and enforceable contracts, or if major-project risks do not impair invested capital, this report should reduce its emphasis on delivery systems and capital discipline.[E01][E05][E09][E12]
2. Industry boundaries and corrections
2.1 Freeze the perimeter: molecules, delivery modes and customers are different classifications
The scope includes air separation and its oxygen, nitrogen and argon products; industrial hydrogen and associated on-site supply; carbon dioxide, helium and electronics specialty gases; and gas-production equipment, engineering, purification, storage, transport, pipelines and customer qualification. Delivery modes are on-site/pipeline supply, merchant bulk liquids delivered by tanker, and cylinders or other packaged supply. Electronics customers use both on-site carrier gases and high-purity specialty molecules, so “electronics gases” cannot be treated as one uniform contract.[E01][E03][E10][E12]
Excluded are natural-gas and LNG trading, electricity sales, end-market fertilizer sales, hospital treatment services, and the full profits of steel, chemicals and semiconductor end products. Medical gases remain within company financial samples, but hospital revenues are not added to this industry. Low-emissions hydrogen is discussed only through production, supply agreements and project investment returns, not as a proxy for the entire energy transition.
Boundary correction: a simple “equipment → production → gas sales” chain misses returnable cylinders, tankers, liquid backup systems, customer purity qualification and contractual credit. These delivery and risk-bearing activities belong in the profit map. Oxygen and nitrogen in the atmosphere do not give customers costless access to specified purity, pressure, flow and continuity.[E01][E09][E12][E23]
2.2 Dates and measurement conventions
The research cutoff is 2026-11-06. FY2025 is the full-year anchor for Linde, Air Liquide and Hangyang, supplemented by the latest FY2026 disclosures available before the cutoff. Air Products' fiscal year ends on September 30. Its FY2026 full-year results were published on 2026-11-05, so FY2025 or FY2026 Q3 must no longer be presented as its latest full-year result.[E01][E02][E03][E04][E05][E06][E07][E08]
Air Liquide's FY2026 Q3 announcement primarily provides sales, not the complete quarterly profit information required here; profit observations therefore use H1. Hangyang's latest Q3 report provides consolidated quarterly results, but equipment/gas margins still use H1. Amounts retain their original currencies and are not added across currencies. Quarterly, half-year and full-year absolute profits are not compared directly or mechanically annualized.[E04][E07][E08][E19]
Gross profit, operating profit, recurring operating income (OIR), adjusted operating profit and net income are labeled separately. Geographic company segments are not value-chain stages; equipment and gas activities may also transact internally. No global profit pool with consistent eliminations was obtained, so this report does not invent value-chain profit shares.
3. Verdicts on four falsifiable hypotheses
H1: Upstream equipment and engineering consistently capture most industrial-gas profits
Verdict: the strong version is weakened; margins and absolute profit must be separated. Air Liquide FY2025 Gas & Services generated OIR of EUR 5,897 million at 22.6%, versus EUR 124 million at 14.5% for Engineering & Technologies. Linde's FY2026 Q3 Engineering segment operating margin was 18.4%, below the 35.1% reported by both Americas and EMEA, although the latter are geographic segments containing mixed gas businesses.[E02][E03]
Key evidence and counterexample: Hangyang's FY2025 equipment gross margin was 27.86%, above gas at 16.45%, contradicting the claim that gas operations have the higher margin at every company. Yet equipment revenue of RMB 5.441 billion and gas revenue of RMB 9.207 billion, multiplied by their disclosed margins, imply approximately RMB 1.52 billion of gross profit each. The higher margin did not automatically produce the larger absolute gross-profit pool. This multiplication is an author estimate using rounded revenues and margins, not the original gross-profit line in the accounts.[E06]
Definition treatment: Hangyang's gross margins cannot be ranked directly against overseas companies' operating margins. Air Liquide's two segment OIR figures cannot be treated as group profit without deducting corporate and R&D costs. The evidence challenges a uniform upstream-winner narrative; it does not establish that every on-site gas project earns more.[E03][E06][E07]
Falsification test: follow the next complete annual and interim reports published during 2027. If equipment/engineering consistently exceeds gas operations in both comparable margins and absolute profits within fixed companies, and this remains true after internal-transaction eliminations, increase the weight assigned to upstream profits. Companies without disclosed activity-level profits cannot count as supporting observations.
Confidence: medium. Exact company figures are single-issuer sourced. Cross-company observations complement one another; they are not independent audits of the same numbers.
H2: New gas-production plants are the first requirement for resolving every industrial-gas shortage
Verdict: rejected as a universal proposition; ranking must be product- and location-specific. Local oxygen/nitrogen delivery depends jointly on power, plant reliability, pipelines and cryogenic distribution. For electronics gases, purity, packaging and customer requalification can constrain supply before nominal capacity does. For helium, feedstock, refining and cross-border transport cannot be replaced by building a local air-separation unit.[E09][E12][E13][E23]
Key evidence: the European Commission's 2026-06-03 impact assessment describes how changes in process-gas supplier, production site or cylinder system often require requalification lasting weeks to months. This describes a supply-chain mechanism, not a statutory waiting period for every gas. On 2026-10-23, the U.S. Department of Energy explained that limited refining capacity and lengthy permitting constrain helium output even where resources exist. Independent AP reporting provides directional corroboration of related supply and timing constraints.[E12][E13][E14]
Counterevidence: new qualified plants can clearly be decisive where customers, power, qualification and delivery infrastructure already exist. Linde's additional on-site supply agreement with Samsung demonstrates customer commitment, but a plan for supply to start in mid-2026 is not proof that this report has verified commissioning.[E20]
Falsification test: over the 12–18 months after 2026-11-06, separately record mechanical completion, power availability, qualification, first commercial deliveries and stable operation for announced projects. If mechanical completion is followed by stable delivery over two consecutive quarters without material subsequent waits, reduce the ranking of qualification or logistics constraints in that market. Do not publish a global shortage percentage without a comparable sample.
Confidence: medium. Multiple independent source types support differentiated bottlenecks, but specific lead times and regional rankings lack harmonized measurement.
H3: The global top three's market share is sufficient to determine pricing power in every gas
Verdict: unsupported. Concentration is informative, but the economic market may be local liquid oxygen/nitrogen, national argon, or a more internationally connected helium chain. The FTC's 2018 merger analysis explicitly distinguished these markets. Its historical U.S. analysis cannot be repurposed as a 2026 global market-share table.[E09]
Key evidence: Scope Ratings' 2026-01-05 report estimated 2024 shares of 30% for Linde, 27% for Air Liquide and 11% for Air Products, totaling 68%, while describing the total as “around 70%.” These estimates draw on company reports, with attention needed to the industry denominator and inclusion of captive production. ShareAction's 2025 report says “over 70%”; its footnote uses FY2023 company revenues, a EUR/USD conversion of 1.10, and a 2024 MarketsandMarkets market-size source. These cannot be combined into a precise 2026 share or used to infer falling concentration.[E11][E25]
Conflict record: Combined global share of the three largest industrial-gas companies [conflictE11-vs-E25] - Source A (Scope Ratings, 2026-01-05): estimated 2024 shares of 30%, 27% and 11%, totaling 68%; the narrative says around 70%. - Source B (ShareAction, 2025-04): over 70%, using FY2023 company revenues, EUR/USD 1.10 and a 2024 MarketsandMarkets market-size source, rather than the same year and definition as Source A. - Nature of conflict: different time windows / incompletely comparable market denominators and estimation definitions. - Verdict and rationale: no verdict on one precise global share. Retain only the qualitative conclusion that major suppliers are concentrated. Neither percentage is labeled a 2026 fact or used to infer markups in every local market.
Falsification test: over the next four quarters, track qualified alternatives, renewal prices, delivery terms and new capacity for the same region, product and purity grade. If changes in local competition leave prices and profits entirely unchanged while these respond only to global concentration, increase the explanatory weight of global share. With insufficient public observations, keep the conclusion conditional.
Confidence: medium for local-market and switching-cost mechanisms; no verdict on precise 2026 global shares or their causal effects.
H4: Long contracts and energy pass-through make industrial-gas expansion almost risk-free
Verdict: they buffer some operating risks, not guarantee capital returns. Linde on-site contracts generally last 10–20 years with minimum purchases and escalation; merchant contracts generally last 3–7 years, and packaged contracts 1–3 years or individual purchase orders. The latter two categories generally lack minimum-volume commitments. Its FY2025 revenue mix was approximately merchant 30%, on-site 24%, packaged 35% and other 11%. Air Products independently discloses a similar distinction between on-site protections and merchant gases without minimum purchases.[E01][E10]
Key evidence: Linde's FY2026 Q3 adjusted operating margin declined 30 basis points year on year to 29.4%, but was flat excluding cost pass-through, demonstrating a denominator effect on margin. Air Products' FY2026 reported operating profit was only USD 23.3 million at 0.2%, versus adjusted profit of USD 3,175.2 million at 24.4%. Material project-exit-related charges demonstrate that contractual protection in established operations cannot prevent capital destruction in new investments.[E02][E05]
Limits of the counterexample: Air Products' losses do not establish that ordinary take-or-pay contracts universally failed. They show that new-project returns, execution and exit risks cannot be replaced by the margin of existing operations. Yara independently withdrew from a related acquisition because its own return hurdles were not met. This corroborates commercial feasibility concerns from another transaction party, not the amount of Air Products' losses.[E16][E17][E18]
Falsification test: during 2027, continuously reconcile energy reset lags, contract receivables, customer credit, project commissioning, incremental capital returns and exit charges. Reduce risk assessments if revenue declines or energy shocks are fully absorbed, receivables remain sound and new projects sustain their original return hurdles. Adjusted earnings per share alone are insufficient.
Confidence: medium. Contract mechanisms have cross-issuer support; company-specific financial figures and project losses remain single-source numbers.
4. Asset one: the profit map
4.1 Where charging rights, capital and risks sit
The table explains how each node earns money and why one stage's margin cannot be extrapolated to the whole industry.
| Node | Representative activities and charging mechanisms | Profit durability | Risks requiring verification |
|---|---|---|---|
| Equipment, engineering and plant delivery | Hangyang equipment, Linde Engineering, Air Liquide Engineering & Technologies; equipment/engineering orders and milestone delivery | Depends on order cycles and execution; not inherently more profitable than gas operations | Schedule, materials, warranties, cancellation and internal eliminations [E01][E03][E06] |
| On-site and pipeline gases | Dedicated plants, capacity fees or minimum purchases, energy resets | Contracts and networks provide structural support, conditional on customer credit and utilization | Do not double-count capacity fees and minimum purchases as separate revenue streams; outage obligations, customer shutdown or default [E01][E10] |
| Merchant bulk liquids | Production, liquefaction, tanks, tankers and local customer density | Local networks and service efficiency can persist; volumes remain cyclical | Energy/fuel, delivery radius, idle capacity, price resets and absent minimum purchases [E01][E09] |
| Packaged and cylinder gases | Gas, container rental, delivery and services | Customer density and container turns matter; not a pure molecular gross margin | Container assets, recovery, safety, labor and smaller-customer demand [E01] |
| Electronics carrier and specialty gases | Long-term on-site carrier supply, high-purity materials, testing, purification and qualification | Qualification and reliability support entry barriers; end-market utilization still matters | Protection varies by product; requalification does not establish a permanent monopoly [E03][E12][E19] |
| Helium and other by-product/rare gases | Feedstock access, refining, transport, storage and inventory management | Feedstock and logistics can generate cyclical scarcity | Scarcity does not guarantee gas available to sell; inventory, substitution and freight can offset price [E13][E14][E21] |
| Large new projects such as low-emissions hydrogen | Long-term purchase commitments, construction, energy and infrastructure combinations | Conditional on bankable demand, cost, policy and execution | Announced capacity, non-binding agreements and final investment decisions are different commitments [E15][E16][E17] |
4.2 Disclosed profit samples
This table asks “who has earned how much, on which definition,” not who owns what share of global profits. Except for explicitly identified author calculations, exact figures are single-issuer sourced with medium confidence.
| Sample | FY2025 full-year anchor | Latest usable FY2026 profit observation at cutoff | Comparability limits |
|---|---|---|---|
| Linde group | Sales USD 33,986 million; reported operating profit USD 8,923 million, 26.3%; adjusted USD 10,137 million, 29.8% [E01] | Q3 sales USD 9,404 million; reported operating profit USD 2,150 million, 22.9%; adjusted USD 2,763 million, 29.4% [E02] | Adjusted segment margins cannot stand in for the group's reported basis |
| Linde Engineering | External sales USD 2,250 million; segment operating profit USD 408 million; author-calculated margin 18.13% [E01] | Q3 external sales USD 582 million; segment operating profit USD 107 million, 18.4% [E02] | Segment and internal-transaction definitions must match; not the entire engineering industry |
| Air Liquide group | OIR EUR 5,582 million, 20.7% [E03] | H1 is the latest complete profit disclosure; segments appear below [E04] | OIR is not unadjusted EBIT; no invented Q3 profit |
| Air Liquide Gas & Services | OIR EUR 5,897 million, 22.6% [E03] | H1 OIR EUR 3,051 million, 22.8% [E04] | Includes different gases, regions and healthcare, not one on-site project |
| Air Liquide Engineering & Technologies | OIR EUR 124 million, 14.5%; group corporate and R&D costs separately EUR 439 million [E03] | No latest comparable quarterly profit series established in this report | Segment profits cannot simply be added to obtain group OIR |
| Air Products group | Reported operating loss USD 877 million, margin −7.3%; adjusted margin 23.7% [E05] | FY2026 full year: reported operating profit USD 23.3 million, 0.2%; adjusted USD 3,175.2 million, 24.4%; Q4 reported USD 633 million, 17.8%, adjusted USD 856 million, 24.1% [E05] | Fiscal year ends September 30; annual and quarterly results separated; adjusted figures exclude material charges |
| Hangyang equipment/manufacturing | Equipment revenue RMB 5.441 billion; gross margin 27.86%; estimated gross profit about RMB 1.52 billion [E06] | H1 manufacturing revenue RMB 2.684 billion; gross margin 28.78%; revenue less cost gives gross profit about RMB 0.772 billion [E07] | Gross profit is not operating profit; “equipment” and “manufacturing” preserve period-specific labels |
| Hangyang gases | Revenue RMB 9.207 billion; gross margin 16.45%; estimated gross profit about RMB 1.52 billion [E06] | H1 revenue RMB 5.821 billion; gross margin 17.72%; revenue less cost gives gross profit about RMB 1.031 billion [E07] | Exact H1 gross profit is calculated from revenue less cost in the same table, not from rounded multiplication |
Hangyang's FY2026 Q3 consolidated revenue of approximately RMB 4.7728 billion and attributable net income of RMB 424.469 million describe the latest consolidated results. They cannot fill in equipment or gas quarterly operating margins.[E08]
Three distinctions between margin, profit pool and cash: Hangyang's equipment gross margin was higher in FY2025, yet estimated absolute gross profits were similar; gas absolute gross profit exceeded manufacturing in H1 FY2026. Linde's FY2025 operating cash flow of approximately USD 10.4 billion less capital expenditure of approximately USD 5.3 billion leaves approximately USD 5.1 billion, but this is merely a simplified cash remainder, not a universally defined distributable profit. Acquisitions, financing and other uses of cash cannot be ignored.[E01][E06][E07]
No global profit-pool estimate: consistent eliminations for captive production, internal equipment delivery, molecule, geography and accounting basis are unavailable. Adding these samples would disguise perimeter errors as market shares.
5. Asset two: bottlenecks ranked by urgency
5.1 A global diagnostic order, not a common shortage percentage
The ranking asks whether a constraint blocks contracted demand from becoming reliable delivery. It is an analytical judgment, not a measured global average lead time.
| Priority | Constraint | Why it ranks here | Observable release signal |
|---|---|---|---|
| 1 | Available, qualified and reliable local delivery | Molecules without pipelines, tankers, backup or quality qualification may remain unusable, especially in electronics [E09][E12] | Actual commercial delivery, qualified batches, supply interruptions and customer acceptance, not equipment arrival |
| 2 | Power, feedstock and critical utilities | Air-separation compression consumes power; hydrogen and helium face different feedstock and refining constraints [E13][E23] | Available power, qualified feedstock, stable output and unit costs improve together |
| 3 | Bankable purchase commitments for major new projects | The IEA's announced-versus-committed gap indicates that contracting and project conditions may constrain investment before hydrogen equipment does [E15] | Binding offtake, financial close, final investment decision and commissioning |
| 4 | Engineering, construction and ramp-up | Engineering becomes effective capacity only after customer and utility conditions are satisfied; exits underline return hurdles [E16][E17][E20][E24] | On-schedule ramp-up, contained budget variance and expected cash recovery |
| 5 | Nominal capacity of general equipment | Can move to first place in some fully supported locations, but not universally across products | Shorter equipment lead times increase deliverable output rather than idle assets |
A helium feedstock or cross-border transport interruption can move priority 2 to first place. For a low-emissions hydrogen project without committed customers, priority 3 should come first. This is a diagnostic sequence, not a fixed answer for every project.
5.2 Regional and product matrix
The matrix identifies what to check first in each region, distinguishing direct evidence from analytical judgment.
| Region/cluster | Suggested order of checks | Evidence and limits |
|---|---|---|
| Established U.S. industrial clusters | Customers and pipelines → power/feedstock → engineering delivery | Historical FTC market definitions and company network projects support the mechanism. Air Liquide announced over USD 350 million for Louisiana on 2026-04-22, including an ASU and network infrastructure, with supply expected in 2028. This is neither a standalone-ASU price nor commissioned capacity [E09][E24] |
| Large U.S. low-emissions hydrogen projects | Bankable purchases and returns → capital execution → power/feedstock | Air Products and Yara's 2026 exits show that return hurdles can bind before technical capability; not a universal U.S. gas bottleneck [E16][E17] |
| China | Local customer utilization and contracts → delivery/reliability → equipment and energy cost | Hangyang demonstrates different equipment and gas economics. Nationally harmonized liquid prices, utilization and qualification lead times are unavailable; no uniform Chinese shortage is claimed [E06][E07][E08] |
| Europe | Power and cost recovery → qualified alternatives → new-project purchase commitments | Air Liquide's pass-through description and the EU assessment support the mechanisms; country power prices and contracts cannot be compressed into one number [E03][E12][E15] |
| Gulf and markets dependent on Gulf helium | Feedstock/export access → refining/logistics → customer inventory/substitution | AP, DOE and Reuters provide different perspectives on helium disruption. No unverified Gulf production share or recovery date is published here [E13][E14][E21] |
| East Asian electronics clusters | Purity/customer qualification → on-site carrier supply/backup → engineering ramp-up | Samsung supply agreements and the EU qualification mechanism are complementary; a Korean project cannot represent all Asian supply [E12][E20] |
6. Asset three: the timing of profit migration
6.1 2023–2026: reject an unsupported linear migration story
This sequence uses comparable historical disclosures from the same Linde FY2025 annual report. It describes changes inside one company, not industry-wide migration.
| Year | Sales and profits, USD million | Engineering share and interpretation |
|---|---|---|
| FY2023 | Sales 32,854; reported operating profit 8,024; total segment/adjusted operating profit 9,070; author-calculated adjusted margin 27.61% | Engineering segment profit 491, or 5.41% of total segment profit [E01] |
| FY2024 | Sales 33,005; reported operating profit 8,635; total segment/adjusted operating profit 9,720; author-calculated adjusted margin 29.45% | Engineering segment profit 410, or 4.22% [E01] |
| FY2025 | Sales 33,986; reported operating profit 8,923; total segment/adjusted operating profit 10,137; author-calculated adjusted margin 29.83% | Engineering segment profit 408, or 4.02%; not engineering's global industry share [E01] |
| FY2026 Q3 | Sales 9,404; adjusted operating profit 2,763, margin 29.4%; underlying growth of 5% comprises price 3% and volume 2% | Networks, pricing and delivery remain important; sales growth is not entirely demand-volume growth [E02] |
The precise percentages above are author calculations from disclosed amounts; annual reports and releases may show one decimal place. Engineering's internal sales are eliminated on consolidation and cannot be added again to external sales in group revenue.[E01]
6.2 Migration between contracts and uses of capital, not all molecules moving together
The table separates events that occurred from conditions for future realization, explaining when profits might shift.
| Stage | Verified event or prospective status | Profit-allocation implication |
|---|---|---|
| Established network operations, 2023–2025 | Linde's comparable profits rose while Engineering's share of total segment profit fell | Supports a larger relative role for this company's gas operations, not a general equipment-industry decline [E01] |
| Electronics demand and investment, 2025–2026 | Samsung expansion agreement targeted supply in mid-2026; Air Liquide FY2026 Q3 Electronics comparable sales grew 12.0% | Orders and sales support demand, but commissioning, utilization and cash must be verified; announced investment is not a return [E19][E20] |
| Major-project capital reassessment, 2026 | Some exits decided/announced in June; charges disclosed in July; November full-year figures show a large reported/adjusted profit gap | Return discipline, timely exits and asset losses become determinants of shareholder returns [E05][E16][E17] |
| Helium disruption, 2026 | Reuters interviewed affected supply-chain participants in March; DOE designated helium critical in October and AP covered supply constraints | Scarcity can raise the value of feedstock access while also reducing gas available for suppliers to sell; no automatic industry-wide margin expansion [E13][E14][E21] |
| 2028 and later, prospective | Louisiana network project expects supply in 2028; IEA discusses the 2030 low-emissions hydrogen pipeline | Earnings require construction, commissioning, customer acceptance and capital recovery; uncommissioned capacity is not realized profit [E15][E24] |
As of 2026-11-06, evidence does not support a uniform “equipment → ordinary gases → specialty gases → hydrogen” profit migration. Air Liquide FY2026 Q3 Industrial Merchant comparable growth was 5.3%, or 7.2% excluding helium, showing helium as a drag on realized sales growth. Helium-shortage headlines cannot directly establish higher profits for helium suppliers.[E19]
7. Asset four: three counterfactual stress tests
All inputs below are author-defined index models, not company forecasts, industry-average parameters or target valuations. Monetary amounts are abstract “units.” The models isolate transmission mechanisms and cannot be added to financial-statement amounts. Assumptions, formulas and outputs are recorded in [E22]; actual contract context is in [E01][E03][E10].
Stress test one: can pass-through protect earnings when electricity costs rise 50%?
Baseline: annual revenue 100; energy cost 30; other variable cost 10; fixed cost including depreciation 35; EBIT 25 and operating margin 25%. Assume unchanged volume and a full-year energy shock, adding 15 to costs. This cost structure is entirely hypothetical.
The table shows how different recovery terms change absolute profit and margin under the same energy shock.
| Scenario | Revenue | Total cost | EBIT | Operating margin |
|---|---|---|---|---|
| Baseline | 100 | 75 | 25 | 25.00% |
| Full, immediate pass-through of incremental energy cost | 115 | 90 | 25 | 21.74% |
| Recover only 80% of incremental energy cost | 112 | 90 | 22 | 19.64% |
| Full recovery after a 3-month lag, with no subsequent true-up | 111.25 | 90 | 21.25 | 19.10% |
The third shock scenario assumes even production through the year and recovery in only the final 9 months: 15 × 9 / 12 = 11.25. With full immediate pass-through, profit is unchanged while margin falls 3.26 percentage points. Treating that denominator effect as complete loss of pricing power would be incorrect.[E22]
Transmission: electricity rises → costs increase → index reset or repricing → recovery percentage and lag determine EBIT → collection timing further affects cash. Financing, working capital, hedging, taxes and customer behavior are excluded; this is not the net risk of an actual contract.
More resilient and vulnerable participants: suppliers with matched formulas, shorter resets and normal collections have more protection; merchant distribution unable to reset promptly is more exposed. If customers reduce demand or default, the unchanged-volume assumption must be withdrawn.
Validation and falsification: over the next four quarters, jointly examine prices, energy recovery, volumes, absolute operating profit and cash. If absolute earnings fall despite complete recovery, investigate cost mismatch, volume or outages rather than declaring contractual failure from this model.[E01][E02][E03][E22]
Stress test two: what does a long contract protect when customer consumption falls?
On-site model: baseline revenue comprises a capacity fee of 60 and a volume-linked gas invoice of 40. Variable costs are 40 and fixed costs 35, leaving EBIT of 25. The capacity fee is one hypothetical contract structure; no additional take-or-pay floor is added, avoiding double counting. Variable revenue and cost move proportionally with volume, and the capacity fee is assumed fully collectible.[E22]
The table distinguishes lower customer production from contract renegotiation.
| Scenario | Revenue | Variable cost | Fixed cost | EBIT |
|---|---|---|---|---|
| On-site baseline | 100 | 40 | 35 | 25 |
| Volume falls 20%, capacity fee unchanged | 92 | 32 | 35 | 25 |
| Volume falls 20%, capacity fee also falls 10% | 86 | 32 | 35 | 19 |
For comparison, under the merchant baseline from stress test one, unchanged prices and a 10% volume decline reduce revenue to 90 and variable cost from 40 to 36. Fixed cost stays at 35, producing EBIT of 19, down 24%, and a 21.11% margin. If volume and cost instead remain unchanged and price rises 3%, EBIT rises from 25 to 28, or 12%. These are fixed-cost arithmetic examples, not estimates of any company's operating leverage.[E22]
Transmission: customer production declines → volume revenue and cost fall → capacity fees or minimum purchases maintain contribution → profit protection exists only if the terms remain enforceable and the customer pays. Renegotiation, insolvency or outage liabilities can weaken protection.
More resilient and vulnerable participants: on-site contracts with creditworthy customers and enforceable fees better resist fluctuations; merchant and packaged businesses without minimum purchases remain exposed to volume. A dedicated plant with few alternative customers may face greater renegotiation risk; networks connecting several customers can provide a buffer.[E01][E10]
Validation and falsification: over the next 12 months, compare fixed on-site revenue, receivables, contract changes and merchant volumes. Withdraw the “unchanged EBIT” scenario if capacity fees are waived or uncollectible. The model does not treat commitments as cash received or assume fixed costs are perfectly invariant at extremely low utilization.[E22]
Stress test three: does a low-emissions hydrogen project retain positive NPV after a one-year delay and cost overrun?
Project model: initial capex 100; annual after-tax unlevered free cash flow of 12 for 15 years beginning in year 1; discount rate 8%; no terminal value. Baseline NPV = −100 + 12 × [1 − (1.08)^−15] / 0.08 = 2.714. Cash flow, duration and discount rate are assumptions with no direct mapping to a particular hydrogen project.[E22]
The table shows how schedule and capex can erode a thin initial return cushion.
| Scenario | NPV, index units | Calculation boundary |
|---|---|---|
| Baseline | 2.714 | Receipts in years 1–15 |
| Delay of 1 year | −4.895 | All 15 payments move to years 2–16; the last payment is not removed |
| Capex rises 20% | −17.286 | Initial spending 120; unchanged receipt timing |
| Delay of 1 year and capex rises 20% | −24.895 | Both shocks applied |
Transmission: insufficient offtake or infrastructure → delayed reliable deliveries → cash flows shift later → construction and financing risks rise → incremental project NPV can turn negative even if gas operating margins are positive. Extra delay costs, ramp-up losses, financing structure, subsidy changes and residual value are excluded, so this is not a worst case.[E15][E22]
Real-world connection: on 2026-06-18, the IEA reported a 2030 low-emissions hydrogen pipeline of approximately 27 Mtpa based on announcements, versus just over 4 Mtpa from committed projects and approximately another 2 Mtpa conditional on further progress. These are forward-looking project statuses, not realized 2030 production or a certain shortage. Air Products and Yara's exits show that return hurdles can stop projects, not that every low-emissions hydrogen route is unprofitable.[E15][E16][E17]
More resilient and vulnerable participants: projects with creditworthy buyers, clear energy costs and supporting infrastructure are better protected. Projects relying on uncommitted demand, permissive schedule assumptions and thin NPV cushions are more exposed. Actual delay compensation, higher contracted prices or other cash offsets require rebuilding the model rather than mechanically applying these losses.
Validation and falsification: at each project update, separately track final investment decisions, binding purchase quantities, capital budgets, start dates and actual cash flow. Announced capacity enters an earnings model only as these conditions are satisfied. If cash realization over the next 12–18 months is faster than the assumptions above, reduce the weight of the delayed case.[E15][E22]
8. Asset five: evidence gaps and unresolved questions
The table specifies what this report cannot adequately answer and how that limits its conclusions.
| Gap | What remains unanswered | Treatment in this report |
|---|---|---|
| Consistent global profit pool | What share of global operating profit belongs to on-site, liquid, packaged and electronics specialty gases? | No invented percentages; only verifiable company and segment samples |
| Independent financial verification | Are exact profits, contract mixes and project losses independently verified on the same basis? | Single-source, medium-confidence labels; SEC hosting, release republication and two researchers reading the same source do not make it independent |
| 2026 global share | Can 68%, around 70% and over 70% be reconciled? | Preserve [conflictE11-vs-E25]; no precise 2026 extrapolation |
| Local transaction prices and effective capacity | What are regional oxygen, nitrogen, argon and helium shortages and realized prices? | No invented global shortage percentage, price rise or recovery date |
| Electronics-gas detail | What are margins and qualification times for carrier gases, specialty molecules, advanced materials and delivery? | Electronics sales growth is not molecular profit; weeks to months is not a fixed qualification period for every product |
| Major-project cash and contracts | How do charges relate to sunk investment, recoverable assets and future liabilities? | Charges are not all treated as current-period cash outflow; adjusted profit does not imply capital losses disappear |
| Announced commissioning | Did the Samsung capacity initially targeting mid-2026 fully start commercial operation? | Insufficient confirmation obtained; retained as a plan rather than realized supply |
| Latest quarterly segment profits | What are Air Liquide's and Hangyang's complete comparable Q3 business margins? | Use H1 profits with explicit dates; do not infer margins from sales growth |
| Helium output units and disruption scale | Can media graphics, original statistics and units be fully reconciled? | Unverified production quantities omitted; qualitative constraints separated from profit claims |
| Extrapolating the counterfactuals | Do the index models represent average real-contract sensitivities? | All are author assumptions; apply to actual valuation only when contract and cost inputs match |
9. Evidence ledger
The ledger connects key figures, mechanisms and scenarios to their sources. “No independent second source” means the same figure lacks independent same-basis corroboration, not that it lacks a source. All websites were accessed on 2026-11-06. Reprints, multiple documents from one issuer and multiple pages from one research institution are not counted twice.
| ID | Data point/judgment | Value or conclusion | Time anchor | Source type | Source name and link | Independent second source? | Confidence | Conflict flag |
|---|---|---|---|---|---|---|---|---|
| E01 | Linde finances, contracts, revenue modes and historical sequence | FY2025 sales 33,986, reported operating profit 8,923, adjusted 10,137, all USD million; contracts and 2023–2025 figures in text; calculated ratios use disclosed amounts | FY2023–FY2025; FY2025 report issued in 2026 | Primary financial | Linde FY2025 Form 10-K, Business and Note 18 | No for exact figures, single source; E10 independently supports contract distinctions, not Linde figures | Medium | No |
| E02 | Latest Linde quarterly profit and pass-through | Q3 sales 9,404; reported operating profit 2,150, 22.9%; adjusted 2,763, 29.4%, all amounts in USD million; price 3%, volume 2%; Engineering 18.4% | FY2026 Q3; 2026-10-29 | Primary financial | Linde Q3 2026 results | No, single source; annual report is not independent corroboration | Medium | No |
| E03 | Air Liquide group/segment OIR and energy pass-through | Group EUR 5,582 million, 20.7%; Gas & Services 5,897, 22.6%; Engineering & Technologies 124, 14.5%; corporate/R&D costs 439 | FY2025; results 2026-02-20, URD 2026-04 | Primary financial | Air Liquide 2025 Universal Registration Document | No for exact figures, single source; E02 independently supports the denominator mechanism, not these figures | Medium | No |
| E04 | Air Liquide H1 gas profits | Gas & Services OIR EUR 3,051 million, 22.8%; not Q3 profit | H1 2026; 2026-07-28 | Primary financial | Air Liquide H1 2026 results | No, single source | Medium | No |
| E05 | Air Products full-year/Q4 reported and adjusted results | FY2026 reported operating profit USD 23.3 million, 0.2%; adjusted 3,175.2, 24.4%; FY2025 reported −877, −7.3%, adjusted margin 23.7%; Q4 in text | FY2026 ended 2026-09-30; 2026-11-05 | Primary financial | Air Products FY2026 results | No for amounts, single source; E17 is an independent party's return-hurdle evidence, not loss verification | Medium | No |
| E06 | Hangyang equipment and gas gross profit | Revenue RMB 5.441/9.207 billion; gross margins 27.86%/16.45%; multiplication implies about RMB 1.52 billion each, an author estimate | FY2025; 2026-04-11 | Primary financial | Hangyang FY2025 annual summary, p.8 | No, single source; recalculating arithmetic is not an independent source | Medium | No |
| E07 | Latest Hangyang segment gross profit | Manufacturing/gas gross margins 28.78%/17.72%; same-table revenue less cost gives about RMB 0.772/1.031 billion gross profit | H1 2026; 2026-08-28 | Primary financial | Hangyang H1 2026 report, p.18 | No, single source | Medium | No |
| E08 | Latest Hangyang consolidated quarter | Q3 revenue about RMB 4.7728 billion; attributable net income RMB 424.469 million; no required quarterly segment margins | FY2026 Q3; 2026-10-29 | Primary financial | Hangyang Q3 2026 report | No, single source | Medium | No |
| E09 | Geographic gas-market boundaries | U.S. liquid oxygen/nitrogen local, argon national, CO2 regional; historical merger analysis, not 2026 shares | 2018-10-22 | Regulatory | FTC Linde/Praxair merger analysis | E01/E11 independently support local-network mechanisms; exact historical definition has this official source | Medium | No |
| E10 | Independent company contract sample | Air Products large on-site generally 15–20 years, smaller 10–15; liquid/packaged generally no minimum purchases | FY2025; issued in 2025 | Primary financial | Air Products FY2025 Form 10-K | E01 supports the mechanism from another issuer; exact terms remain issuer-specific single-source facts | Medium | No |
| E11 | Global concentration estimate and local barriers | 2024 estimates Linde 30%, Air Liquide 27%, Air Products 11%, totaling 68%; narrative around 70% | 2024 estimate; 2026-01-05 | Industry/credit research | Scope Ratings Air Liquide report | E25 is directionally similar but differently defined, not same-number verification | Medium | Yes |
| E12 | Electronics-gas qualification and delivery systems | Impurities, packaging, valves/cylinders, supplier and site changes involve requalification, potentially weeks to months | 2026-06-03 | Regulatory research | European Commission SWD(2026)504, Part 2, pp.74–75 | No for exact duration range, single source; E09 supports only entry-barrier mechanisms | Medium | No |
| E13 | U.S. helium constraints and critical-material designation | Refining, permitting and supply-chain constraints; designation does not immediately add production | DOE 2026-10-23; notice 2026-10-27 | Government/regulatory | DOE announcement; Federal Register notice | E14 independently reports the direction; DOE and the register are the same policy source | Medium | No |
| E14 | Helium feedstock, refining and disruption | AP reports supply/timing constraints; no production quantity or recovery forecast used here | 2026-10-23 | Professional media | AP helium coverage | E13/E21 support mechanisms from different sources, not a single quantity | Medium | No |
| E15 | Announced versus committed low-emissions hydrogen | 2030 announced pipeline about 27 Mtpa; committed over 4 Mtpa; another about 2 Mtpa conditional; all forward project statuses | 2026-06-18; 2030 forecast window | Industry research | IEA Global Hydrogen Review 2026 executive summary; key questions | No for numbers, one IEA source; E16/E17 independently support commercialization conditions | Medium | No |
| E16 | Air Products project exits | June decision/announcement; approximately USD 2.9 billion pretax charge recognized in July results, no longer merely expected | Decision 2026-06-26; announcement 06-30; results 07-30 | Company announcement/financial | Project exit announcement; Q3 results | E17 independently confirms counterparty return concerns; charge solely issuer-sourced | Medium | No |
| E17 | Yara withdrew from related acquisition | Own return hurdles not met; withdrew from related LCEC acquisition | 2026-06-30 | Counterparty announcement | Yara U.S. ammonia and capital allocation update | E16 is another transaction party; does not verify losses | Medium | No |
| E18 | Independent project-exit reporting | Reuters on exits, counterparty and analytical context; charge figures still originate with issuer | 2026-06-30 | Professional media | Reuters Air Products project exit | Independent reporting, not independent auditing of the loss figure | Medium | No |
| E19 | Latest Air Liquide sales and helium effects | Q3 Electronics comparable growth 12.0%; Industrial Merchant 5.3%, excluding helium 7.2%; no inferred profit margin | FY2026 Q3; 2026-10-29 | Primary financial | Air Liquide Q3 2026 release | No for exact numbers, one issuer; media repetition is not independent numerical verification | Medium | No |
| E20 | Samsung on-site supply plan | Agreement for 8 additional ASUs; supply expected mid-2026; full commercial start not verified here | Announced 2025-04-29; planned mid-2026 | Company announcement | Linde Samsung expansion | No, single source; no verdict on commissioning status | Medium | No |
| E21 | Transmission of helium disruption to technology supply chains | Supply-chain effects and transport delays; graphic production figures with unverified units excluded | 2026-03-26 | Professional media | Reuters helium supply-chain interviews | Later E13/E14 support direction, not same-date quantitative verification | Medium | No |
| E22 | Three transparent index stress models | All baselines, shocks, ratios, durations, discount rates and outputs are author assumptions/arithmetic in section 7 | Prepared 2026-11-06; scenarios, not forecasts | Author calculation | Section 7 formulas; contract context E01/E03/E10, project context E15–E17 | Not applicable; assumptions are not presented as measurements, calculations reproducible | Medium | No |
| E23 | Air-separation energy and process boundaries | Compression is an important electricity-consuming step; state, scale and co-product allocation change intensity; no universal intensity adopted | 2021-03 | Industry/government research | PBL/TNO Dutch industrial gases production | E01 independently identifies energy as an important cost; process study itself single-source | Medium | No |
| E24 | U.S. network expansion capital and schedule | Over USD 350 million, including ASU and network infrastructure; supply expected in 2028 | Announced 2026-04-22; prospective 2028 | Company announcement | Air Liquide Louisiana investment | No, single source; not a standalone-plant price | Medium | No |
| E25 | Alternative global concentration statement | Top three over 70%; FY2023 revenues, EUR/USD 1.10 and a 2024 market-report denominator, not directly comparable with E11 | Report 2025-04 | Industry research | ShareAction Powering Change | E11 supports only direction; exact percentages not reconciled | Medium | Yes |
10. References and source tiers
Sources are grouped by function so that issuer announcements, regulatory analysis and media reprints are not mistaken for independent numerical verification. All were accessed on 2026-11-06; full clickable links and data periods appear above.
Primary financial disclosures
- Linde FY2025 Form 10-K, FY2025, issued in 2026: [E01].
- Linde Q3 2026 results, 2026-10-29: [E02].
- Air Liquide 2025 URD, 2026-04; H1 2026 results, 2026-07-28; Q3 sales, 2026-10-29: [E03][E04][E19].
- Air Products FY2026 results, 2026-11-05; FY2025 Form 10-K, issued in 2025: [E05][E10].
- Hangyang FY2025 annual summary, 2026-04-11; H1 2026, 2026-08-28; Q3 2026, 2026-10-29: [E06][E07][E08].
Industry and credit research
- Scope Ratings Air Liquide report, 2026-01-05: [E11].
- IEA Global Hydrogen Review 2026, 2026-06-18: [E15].
- PBL/TNO industrial-gas production research, 2021-03; historical technical mechanisms, not a 2026 cost benchmark: [E23].
- ShareAction Powering Change, 2025-04: [E25]. Its institutional perspective does not remove the need to verify numerical definitions.
Professional media and independent reporting
- AP helium coverage, 2026-10-23: [E14].
- Reuters Air Products project exit, 2026-06-30: [E18].
- Reuters helium supply-chain interviews, 2026-03-26: [E21]. Only qualitative interview findings are used, not graphic figures with unreconciled units.
Company announcements, regulatory documents and policy research
- FTC Linde/Praxair, 2018-10-22: [E09].
- European Commission SWD(2026)504 Part 2, 2026-06-03, hosted by Bulgaria's Parliament; an impact assessment, not a newly enacted gas-qualification rule: [E12].
- DOE helium critical-material announcement, 2026-10-23; Federal Register, 2026-10-27: [E13].
- Air Products exit announcement, 2026-06-30; quarterly recognition, 2026-07-30; Yara capital-allocation update, 2026-06-30: [E16][E17].
- Linde/Samsung, 2025-04-29: [E20]; Air Liquide Louisiana, 2026-04-22: [E24].
This is the Day13 research draft. All data follow the time anchors in the evidence ledger; unresolved conflicts must not be presented as established conclusions in subsequent copy. This is not investment advice.