Air Products: Industrial Gas and Hydrogen Infrastructure in the AI Energy Transition

Executive Summary

Air Products and Chemicals (APD), the second-largest industrial gas company globally with fiscal 2023 revenue of approximately $12.6 billion, finds itself at a peculiar strategic inflection point. Like Linde, it supplies the atmospheric and process gases that power semiconductor manufacturing, petrochemical refining, and medical systems — businesses broadly insulated from AI disruption. Unlike Linde, Air Products has made an extraordinary concentrated bet on green hydrogen infrastructure, committing roughly $15 billion to three flagship mega-projects: NEOM Green Hydrogen in Saudi Arabia, the green hydrogen facility in Alberta (Canada), and a U.S. clean hydrogen hub.

This concentrated hydrogen strategy, executed under former CEO Seifi Ghasemi and now under significant pressure from activist investor Mantle Ridge, is the defining variable in any AI-era assessment of Air Products. AI impacts this business as both a direct end-market driver (semiconductor and data center gas demand) and as an accelerator of the clean energy transition that underpins hydrogen economics. The net AI margin pressure score is 2/10 — broadly positive — but with considerably more execution risk than Linde.

Business Through an AI Lens

Air Products operates through four geographic segments: Americas, Asia, Europe, and Middle East, India, Egypt. Its gas portfolio — oxygen, nitrogen, argon, hydrogen, helium, carbon dioxide, and specialty gases — mirrors Linde's but at roughly one-third the scale. The company's long-term contract model (take-or-pay, 15-20 year terms, energy pass-through clauses) provides cash flow visibility and margin stability independent of AI's near-term trajectory.

The AI intersection is most direct in the electronics sub-segment. Air Products supplies ultra-high-purity gases to semiconductor fabs in South Korea (Samsung, SK Hynix), Taiwan (TSMC, UMC), and increasingly the United States as CHIPS Act investments materialize. AI chip demand — Nvidia's H100/B200 GPU architecture, AMD's MI300 series, and custom ASICs from Google, Amazon, and Microsoft — requires advanced node semiconductor capacity that consumes escalating quantities of specialty gases per wafer at each process generation.

The strategic inflection, however, is hydrogen. Air Products is the world's largest hydrogen producer by volume, supplying the industrial hydrogen that powers petrochemical refining along the U.S. Gulf Coast, European chemical corridors, and Asian refineries. The pivot to green hydrogen — electrolysis-derived hydrogen using renewable electricity — is the central strategic gamble. AI's role here is indirect but meaningful: AI-optimized grid management, AI-accelerated electrolyzer catalyst research, and AI-driven energy demand forecasting all affect the economics of the NEOM and Alberta projects.

Revenue Exposure

Air Products' $12.6 billion in fiscal 2023 revenue breaks down across end markets with varying AI sensitivity:

The electronics segment is growing at high-single-digit rates and should reach approximately $1.9-$2.1 billion by fiscal 2027, supported by fab expansions in the U.S. and Asia. This is straightforwardly AI-positive.

The energy and refining segment is more complex. Industrial hydrogen supplied to oil refineries for hydrotreating and hydrocracking represents significant volume that AI cannot substitute but also may not grow significantly as the long-run energy transition progresses. Green hydrogen mega-projects, if executed on schedule, would add $800 million to $1.2 billion in new revenue by 2030 at premium pricing under long-term offtake agreements.

Cost Exposure

Air Products' cost structure shares Linde's dominant characteristic: electricity is the largest variable cost driver, representing approximately 35% of cost of sales. The company's energy pass-through mechanisms in customer contracts provide margin insulation, but in energy-volatile markets (as seen in Europe during 2022-2023), volume risks emerge when customers temporarily curtail operations due to energy economics.

AI is materially affecting Air Products' own operational costs in two ways. The company has deployed predictive maintenance AI across its ASU fleet, reducing unplanned downtime by an estimated 15-20% versus five years ago. This translates to approximately $50-$70 million in avoided maintenance costs and revenue loss annually. More significantly, Air Products has invested in AI-optimized production scheduling that coordinates ASU output with real-time electricity price signals across 19 countries — a capability that management estimates saves $80-$120 million per year in energy costs.

The NEOM project introduces a different cost calculus. At $8.5 billion in capital commitment, the project's economics depend critically on renewable electricity costs, electrolyzer efficiency, and ammonia shipping costs for hydrogen export. AI-accelerated electrolyzer research (from players such as Electric Hydrogen and ITM Power) could improve efficiency assumptions, but could also enable competitors to build comparable facilities at lower cost — a competitive risk for Air Products' first-mover premium.

Moat Test

Air Products' competitive moat in its core industrial gas business is robust: pipeline infrastructure, on-site supply contracts, safety expertise, and regulatory certifications create meaningful barriers. In the hydrogen economy, the moat is less clear. NEOM is an unprecedented scale project, but the underlying technology (alkaline and PEM electrolysis) is increasingly commoditized, and project execution risk is high in a politically complex region.

The activist pressure from Mantle Ridge — which has pushed for CEO change, project review, and potential asset divestitures — reflects market skepticism about whether the hydrogen mega-project strategy creates or destroys shareholder value. This governance uncertainty is not an AI issue per se, but it affects management bandwidth to respond to AI-related opportunities and competitive dynamics.

Timeline Scenarios

1-3 Years (Near Term)

The near-term priority is execution on the NEOM hydrogen project, which faces a revised timeline targeting first production in late 2026 or 2027. Semiconductor gas demand provides a stable earnings floor with modest growth. The new management team (Mantle Ridge has secured board changes) is conducting a strategic review that may result in project divestitures or restructuring. AI-related earnings tailwinds from electronics are real but relatively modest compared to the NEOM uncertainty. Operating margins, currently approximately 22-23%, could compress to 20-21% if hydrogen project overruns require additional expense recognition.

3-7 Years (Medium Term)

If NEOM achieves commercial production, Air Products could receive a step-change in revenue — potentially $1.0-$1.5 billion in additional high-margin hydrogen supply revenue by 2030. AI's role in accelerating electrolyzer cost curves is relevant here: if hydrogen production costs fall faster than projected, NEOM's offtake pricing may look attractive relative to the market, protecting returns. Conversely, faster cost reduction by competitors could reduce Air Products' pricing power in future hydrogen contracts. The Alberta project faces similar dynamics in the North American market.

7+ Years (Long Term)

Over the long run, Air Products' position as the world's largest hydrogen infrastructure owner — if the mega-projects deliver — creates an analogy to early natural gas pipeline networks: the physical infrastructure becomes the dominant competitive advantage regardless of who produces or transports the molecule. AI's long-run impact on hydrogen demand (through AI-optimized grid storage, AI-designed fuel cells, and AI-accelerated materials science for electrolyzers) is net positive for demand volumes while creating downward pressure on pricing. Air Products must own the infrastructure, not just the production, to capture enduring value.

Bull Case

In the bull case, NEOM and Alberta achieve first production on revised timelines, green hydrogen economics improve faster than consensus forecasts due to AI-accelerated electrolyzer development, and semiconductor gas demand grows at double digits through 2028. Operating margins recover to 25-26% as hydrogen projects move from capital deployment to cash generation. The stock re-rates from current levels near 22x forward earnings toward 26-28x, as the market assigns value to the hydrogen infrastructure option.

Bear Case

In the bear case, NEOM faces further delays and cost overruns, requiring additional capital or project restructuring that impairs earnings visibility. Green hydrogen production costs fail to reach parity with grey hydrogen before 2035, limiting offtake volumes and revenue from the Alberta project. Semiconductor gas demand growth slows if AI efficiency improvements reduce hardware intensity faster than projected. Operating margins remain compressed at 20-22%, and the stock trades at 18-20x earnings — a valuation that offers limited upside from current levels.

Verdict: AI Margin Pressure Score 2/10

Air Products earns a 2/10 on AI margin pressure. Like Linde, the core industrial gas business is structurally insulated from AI disruption — physical gas production cannot be cognitively automated, and long-term contracts provide cash flow stability. AI is a net positive through semiconductor demand, data center cooling, and AI-accelerated hydrogen infrastructure economics. The elevated execution risk around NEOM and governance uncertainty from activist pressure are not AI risks; they are strategic and operational risks that exist independently of the AI cycle. Investors should assess these separately from AI margin pressure analysis.

Takeaways for Investors

Air Products is a more complex investment than Linde due to concentrated hydrogen project risk, but the AI structural outlook is equally positive. The key differentiation from Linde is execution risk on mega-projects and governance uncertainty rather than any AI-driven fundamental threat. Investors seeking AI infrastructure exposure in materials should prefer Linde for execution certainty, while Air Products offers higher optionality on the green hydrogen transition — with commensurate project and governance risk. Monitor NEOM construction milestones, the strategic review outcomes under new leadership, and U.S. hydrogen production tax credit (45V) guidance as the primary catalysts.