Scaling the Industrial Transition 2025

2.4 Infrastructure is expanding but strained While clean-energy generation is accelerating, delivery systems are falling behind. Grid congestion,53 constraints with CO2 transport, hydrogen logistics and inadequate port capacity are now the main bottlenecks to scaling industrial transition. Rapid electrification – driven by transport, heavy industry and the expansion of AI – is colliding with a surge in power demand from data centres. The emerging “AI–energy paradox” exacerbates system strain,54 forcing industrial players to compete directly with digital infrastructure for limited, low-carbon grid capacity. Yet, the same technologies contributing to higher demand can also hold part of the solution: AI applications across hard-to-abate sectors can improve efficiency, optimize operations and reduce emissions (Table 5). Still, these efficiency gains cannot compensate for the scale of physical infrastructure required. Grid expansion, CO2 transport networks, hydrogen corridors and port upgrades are advancing too slowly to meet rising demand. Energy storage is the bright spot: BloombergNEF projects 35% growth from 2025 to 94 GW/247 GWh (gigawatt-hour), with large projects advancing across the US, China, Saudi Arabia, South Africa, Australia, Chile, the United Kingdom and the Netherlands (Figure 5).55 Shared CO2 and hydrogen infrastructure – such as pipelines, storage hubs and ammonia-ready ports – is emerging as the most practical and cost- effective model, allowing multiple industries to share networks and reduce investment risk. Yet progress remains fragmented. Without faster investment in grids, CO2 hubs, hydrogen corridors and ports, ready-to-build projects will stay stuck between pilot and scale, making infrastructure alignment the real test of transition readiness. AI industrial applications in hard-to-abate sectors Sector AI use cases Decarbonization impact Benefit Aviation –Flight path optimization56 –Demand analytics57 –Predictive maintenance58Lower emissions per passenger- km,59 increased fleet efficiency, operational stability,60 potential reduction in non-CO2 impacts3–8% fuel savings61 Shipping –Route optimization62 –Digital twins of vessels63 –Anomaly detection64Minimized fuel consumption,65 less downtime, lower CO2 emissions6610% fuel savings per voyage 20% GHG emission reduction67 Trucking –Logistics and route optimization –Load management –Predictive vehicle maintenance68Reduces idle time, improves load factor, cuts per tonne of CO2 emissions10–15% fuel savings, emission reduction69 Steel –Real-time furnace optimization70 –Asset health prediction71 –Process control72Reduces energy consumption, targets hotspots, improves operational efficiency735–10% emission reduction74 Aluminium –Load balancing –Predictive process control –Energy management75Reduced electricity consumption, higher smelting efficiency, fewer failures7610% energy savings in smelters Cement –Kiln process optimization77 –Smart energy management78Lower energy use, clinker reduction, emission intensity drops79 10–15% energy reduction in kilns80 Primary chemicals –Process AI for safety81 –Catalyst design82 –Waste minimization83Decreases waste, flaring, energy use, boosts plant throughput 5% reduction in CO2 emissions,84 predictive maintenance improvements Oil and gas –Energy optimization –Methane leak detection85 –Predictive maintenance –Demand forecasting86Cuts methane leaks by up to 95%,87 improves energy efficiency, reduces Scope 1 and 2 emissions88-40% methane reduction by 203089TABLE 5 Clean energy generation is accelerating, but infrastructure is lagging – grid limits, CO2 transport gaps and rising AI demand now test system readiness. Scaling the Industrial Transition: Hard-to-Abate Sectors and Net-Zero Progress in 2025 19