APD’s beat is not the story; the $4 billion capex reset is

The print matters because it was just clean enough to avoid an immediate numbers reset, but not clean enough to validate a rerating. What was priced in was a narrow EPS outcome around consensus, since Q4 FY2025 EPS came in at $3.39 versus the street at $3.38, a +0.3% surprise. What actually surprised was that revenue missed despite the modest sequential recovery: actual revenue was $3,166.9 million versus estimate $3,176.2 million, a -0.3% surprise, and the company’s own trajectory still shows Q4 FY2025 revenue down -0.6% YoY after Q3 FY2025 had turned +1.3% YoY. That is the variant perception: the headline EPS beat is being misread as evidence that the base business is already compounding again, when the data instead says the FY2026 bridge depends on below-the-line cleanup, headcount savings, and delayed capital normalization. The stock can work only if investors accept $4 billion of FY2026 capital spending as a temporary bridge to roughly $2.5 billion per year; if not, the +0.3% EPS beat is too small to carry the multiple.

The revenue and margin path is the first place to separate signal from noise. Q4 FY2025 revenue of $3,166.9 million improved +4.8% QoQ from Q3 FY2025 revenue of $3,022.7 million, but that came after a fiscal year that included Q1 FY2025 revenue of $2,931.5 million, Q2 FY2025 revenue of $2,916.2 million, and only a partial recovery in Q3 FY2025. The year-on-year comparison is less forgiving: Q4 FY2025 was -0.6% YoY against Q4 FY2024 revenue of $3,187.5 million, and the reported gross margin of 32.3% was below Q4 FY2024 gross margin of 34.0%. The market could reasonably have expected a better operating setup because Q4 FY2024 gross margin had reached 34.0%, but the actual Q4 FY2025 margin landed closer to Q3 FY2025 at 32.5% and below the Q3 FY2024 to Q4 FY2024 expansion that took gross margin from 32.8% to 34.0%. In other words, the fourth quarter did not show a margin snapback; it showed revenue stabilization with gross margin still below the high-water mark in the recent series.

That margin context explains why management’s FY2026 EPS guide is more important than the quarterly beat, because the earnings bridge is explicit about self-help. Melissa Schaeffer’s guidance language is the anchor: “For the full year, we expect to deliver earnings per share in the range of $12.85 to $13.15, an improvement of 7% to 9% from the prior year.” The commitment is clear, but the building blocks are not all volume-led. The company’s own call basis for fiscal 2025 was EPS of $12.03, which Eduardo Menezes said was “above the midpoint of our full year fiscal guidance range,” while Schaeffer said fiscal year EPS of $12.03 decreased $0.40 or 3% from prior year because of a 4% headwind from the LNG divestiture and a 2% headwind from project exits. That is not a demand acceleration story. It is a guide that asks investors to underwrite recovery from portfolio headwinds while management continues to shrink the cost base and rationalize projects.

The self-help component is not small enough to ignore, and that is precisely why the EPS guidance can be both credible and lower quality than a cyclical upturn. Since 2022, APD has identified 3,600 headcount reductions, which management says translates to 16% of peak workforce, and expects those reductions to contribute approximately $250 million in annual cost savings or $0.90 per share in earnings once complete. That $0.90 per share is material next to FY2026 EPS guidance of $12.85 to $13.15, and it gives the company a controllable lever if volumes remain mixed. The issue for portfolio managers is not whether $250 million of annual cost savings helps; it plainly does. The issue is multiple quality. Earnings generated by 3,600 headcount reductions and 16% workforce compression should not receive the same multiple as earnings generated by accelerating merchant gases demand, especially when Q4 FY2025 revenue was still -0.6% YoY and Q4 FY2025 gross margin was 32.3% rather than the prior-year 34.0%.

The capital spending discussion is where the bull case becomes more fragile, because management raised the near-term capital intensity while promising a lower steady state later. Menezes said, “In 2026, we expect our capital expenditures to be about $4 billion,” and later framed the post-project run rate as roughly $2.5 billion per year following completion of several large projects. Schaeffer added that FY2026 capital spending includes approximately $1 billion on traditional industrial gas projects and ongoing maintenance, while another call exchange captured investor pushback that the fiscal 2026 capex forecast had moved from about $3.1 billion to $4 billion. The debate is not semantic. If capex is $4 billion in FY2026 and only later falls to roughly $2.5 billion per year, the free-cash-flow inflection is pushed beyond the earnings guide investors are buying today. The variant view is that the market is over-crediting the later roughly $2.5 billion steady-state claim and under-discounting the immediate $4 billion cash outlay required to execute the backlog.

That backlog risk is amplified by the large-project and hydrogen comments, because management is asking investors to fund strategic optionality before the returns are visible in the quarterly P&L. The company expects to spend about $4 billion as it executes the project backlog, including approximately $1 billion on traditional industrial gas projects and ongoing maintenance, and Schaeffer said that as long as APD is executing against backlog it will see $3.5 billion to $4 billion. Menezes also said the company has a long-term commitment for almost 50% of the volume with a major customer on one project, language that helps de-risk offtake but leaves half of the volume outside that commitment. His green ammonia comment was pointed because he stressed that bringing green ammonia from Saudi Arabia for dissociation in Europe is competitive in pricing and requires 0 public subsidies. The lack of public subsidies is strategically useful if true in realized customer economics, but the financial statement still shows Q4 FY2025 revenue of $3,166.9 million, down -0.6% YoY, not a project ramp already lifting the base.

The tone of the call supports that cautious reading rather than contradicting it. In the tone history, Q4 FY2025 transcript sentiment was 0.05, down from Q1 FY2025 sentiment of 0.14 and only slightly above Q3 FY2025 sentiment of 0.02; guidance_tone was 0.09, below Q3 FY2025 guidance_tone of 0.20 and far below Q2 FY2025 guidance_tone of 0.43. The more important number is uncertainty, which reached 71.7 in Q4 FY2025 versus 36.3 in Q3 FY2025 and 52.7 in Q2 FY2025. That spike matches the substance of the call: management gave investors an EPS range, but the Q&A focused on capex escalation, NEOM assumptions, monetizing Darrow costs, helium headwinds, and whether the headcount base had changed. Q4 FY2025 qa_evasiveness was 27.6, above Q3 FY2025 qa_evasiveness of 22.8, while qa_sentiment was 0.03 versus Q3 FY2025 qa_sentiment of 0.11. This was not a call where management tone resolved the backlog debate; it surfaced the debate.

That call delivery matters for the semis read-through because APD is a process-input supplier to the fabs, and the quarter does not point to a broad specialty-gas restocking wave. APD supplies TSMC with bulk gases including N2, O2, Ar, He, plus NF3, SiH4, and specialty gases; Intel with bulk gases, He, NF3, and specialty gases; Samsung with NF3, SiH4, N2O, C2F6, and bulk gases; SK Hynix with helium, NF3, and specialty gases; and Micron with NF3 and specialty gases. The direct implication is that fab-related gas demand is not yet overpowering APD’s portfolio and project noise: company revenue in Q4 FY2025 was $3,166.9 million, missed the street by -0.3%, and remained -0.6% YoY despite semiconductor customers continuing to consume critical gases. Schaeffer’s helium comment also matters for memory and leading-edge fabs because she said APD had forecasted between $0.50 and $0.55 headwind on helium for the full year. That magnitude frames the customer read-through: helium remains a meaningful earnings variable for APD, not just a pass-through detail, and it touches Intel, TSMC, Samsung, and SK Hynix through bulk gas and helium supply exposure.

The peer comparison does not rescue the APD story, because the company sits in the middle of materials margin performance while lagging the stronger revenue-growth cohorts. APD’s Q4 FY2025 gross margin of 32.3% matches 3407.T at 32.3% and is below 6367.T at 32.9% and 4901.T at 40.6%, while its Q4 FY2025 revenue YoY of -0.6% is behind 6367.T at +16.4%, 5201.T at +7.7%, 4901.T at +6.8%, 3407.T at +4.5%, 3402.T at +4.1%, and SHECY at +3.2%. It is better than 4188.T at -10.1% and 4005.T at -11.3%, but that is not the right benchmark if investors are paying for a backlog-driven earnings recovery. The comparative point is straightforward: APD’s margin is not differentiated enough, at 32.3%, to offset a revenue line that was still -0.6% YoY when several materials peers were posting positive growth.

The EPS optics also require discipline because the print basis and the company’s call basis are different, and mixing them would overstate the message. The street-comparison print says Q4 FY2025 EPS was $3.39 versus estimate $3.38, a +0.3% surprise. The quarterly history table shows diluted EPS of $0.02 for Q4 FY2025, after -$7.77 in Q2 FY2025 and $3.20 in Q3 FY2025, reflecting reporting items that are not the same basis as the street-comparison EPS line. The company’s call basis says fiscal year EPS was $12.03 and guides FY2026 EPS to $12.85 to $13.15. The conclusion is not that earnings quality is bad; it is that investors must be precise about which earnings stream they are underwriting. On the clean guide, management is offering 7% to 9% improvement. On the actual quarterly path, the business just delivered revenue of $3,166.9 million, gross margin of 32.3%, and a Q4 FY2025 diluted EPS line of $0.02 in the history table. Those facts justify a valuation discount until the cash and margin evidence catches up with the guidance.

The near-term setup therefore comes down to whether Q1 FY2026 confirms that the cost program can bridge the revenue softness without further capex creep. Schaeffer guided Q1 FY2026 EPS to $2.95 to $3.10, representing a 3% to 8% improvement from the prior year, and the quarterly history already shows Q1 FY2026 revenue of $3,102.5 million, gross margin of 32.1%, revenue QoQ of -2.0%, revenue YoY of +5.8%, and diluted EPS of $3.04. The confirming pattern would be Q1 FY2026 EPS staying within $2.95 to $3.10, gross margin holding near 32.1%, and the FY2026 EPS range remaining $12.85 to $13.15 while management keeps FY2026 capex at about $4 billion rather than letting the $3.5 billion to $4 billion language drift higher. The break point would be any revision that weakens the $12.85 to $13.15 guide, any indication that the $250 million annual cost savings or $0.90 per share benefit is slipping, or any move away from the roughly $2.5 billion per year post-project capex target. For the next quarterly call, the numbers to watch are the Q1 FY2026 EPS range of $2.95 to $3.10, the FY2026 capex level of about $4 billion, the post-completion capex target of roughly $2.5 billion per year, and whether uncertainty in the tone history falls from Q4 FY2025’s 71.7; if uncertainty stays elevated while capex remains at $4 billion, the market is still paying too much for a cost-led EPS bridge.