What Air Products Does

Air Products and Chemicals, Inc. is a world-leading industrial gases company. It produces and supplies atmospheric gases (like oxygen, nitrogen, and argon), process and specialty gases, and equipment for a vast range of industries, including refining, chemicals, metals, electronics, and healthcare. Its core technology involves cryogenic air separation and steam methane reforming, which are highly energy-intensive processes. The company operates on a massive global scale, with thousands of employees and a complex logistics network for delivering gases via pipeline, truck, and cylinder.

Why AI Matters at This Scale

For a capital-intensive, continuous-process manufacturer like Air Products, marginal efficiency gains translate into massive financial and operational impact. At its size (10,001+ employees), even a 1% reduction in energy consumption or unplanned downtime across its global plant network can save tens of millions of dollars annually. Furthermore, the industrial sector is under increasing pressure to decarbonize, and AI is a powerful tool for optimizing energy use and accelerating the development of low-carbon technologies like green hydrogen and carbon capture. Large enterprises have the data infrastructure, capital, and technical talent to pilot and scale AI solutions that smaller competitors cannot.

Concrete AI Opportunities with ROI Framing

1. Predictive Maintenance for Critical Assets: Deploying machine learning on sensor data from compressors, turbines, and heat exchangers can predict equipment failures weeks in advance. For a single large air separation unit, an unplanned outage can cost over $1 million per day in lost production and emergency repairs. A predictive system preventing just one major failure per plant per year offers a rapid ROI, while improving overall asset reliability and safety. 2. Dynamic Process Optimization: AI algorithms can continuously analyze real-time data from production plants to adjust setpoints for maximum efficiency. Given that energy can constitute 70-80% of the production cost for industrial gases, a 2-5% optimization in energy use per plant directly boosts gross margins. This is a continuous, software-driven improvement versus costly capital upgrades. 3. AI-Enhanced Supply Chain for Gases: Industrial gas demand is volatile. AI can improve demand forecasting by analyzing customer production schedules, weather data, and economic indicators. This optimizes production planning, bulk delivery routing, and cylinder inventory, reducing logistics costs (a major expense) and improving service levels. Better forecasting reduces waste from overproduction and costly emergency deliveries.

Deployment Risks Specific to Large Enterprises (10,001+)

The primary risk is integration complexity with legacy Operational Technology (OT). AI models developed in IT environments must safely interface with decades-old industrial control systems (ICS/SCADA) that run 24/7 critical infrastructure. Any failure or unvetted change can cause catastrophic safety or production issues. This necessitates slow, careful collaboration between data scientists and veteran process engineers, potentially slowing deployment. Organizational inertia is another hurdle. Large, successful companies have established, proven workflows. Gaining buy-in from plant managers who are measured on reliability, not experimentation, requires clear pilot results and top-down mandate. Data silos across global business units and between engineering, maintenance, and logistics departments can hinder the creation of unified data lakes needed for the most impactful AI models. Finally, cybersecurity risks multiply when connecting historically isolated OT networks to AI platforms, requiring significant investment in securing these new data pathways.