Cummins’ retreat from hydrogen electrolysers has taken an uncomfortable turn. In September 2026, its Hydrogenics subsidiaries told HyLYZER customers to stop operating their electrolysers after operating data called safety assumptions into question. Cummins’ November regulatory filing leaves the scope, remedy and costs unresolved, and warns of potentially material financial consequences.
Cummins was a substantial supplier in the PEM electrolyser market. Accelera reports more than 45 HyLYZER 500 systems with operating histories dating to 2015 and more than 20 of its 5 MW HyLYZER 1000 systems since 2020. The minimum counts and equipment ratings imply approximately 145–213 MW of historical deployments, equivalent to 11–16% of the IEA’s projected 1.3 GW global PEM installed base at the end of 2026.
In February, Cummins announced its withdrawal from new electrolyser sales, citing deteriorating market conditions. Its 2025 Accelera restructuring and impairment charges totalled $458 million, much of them noncash and associated with electrolysers. Cummins committed to serving existing customers while reducing the organisation that built their equipment. Eurofound documents plans to eliminate 100 positions in Belgium, retaining 60 to finish orders and support installations. The Guadalajara, Spain, redundancy agreement covers 88 of 104 positions. That leaves a smaller specialist organisation to manage a potentially extensive repair programme.
That recalls my July assessment of Cummins’ fuel-cell exit and Alstom’s obligations to hydrogen trains. Alstom acquired engineering and support capabilities for a fleet requiring maintenance long after the original sales case had weakened. In March 2026, Cummins made a net payment of $175 million in disposing of its low-pressure fuel-cell business, cancelling commitments and settling claims with a customer. The same transaction produced a $199 million accounting charge. The electrolyser warning creates a new, unquantified exposure. Discontinuing sales is much easier than closing out decades of installed-equipment obligations.
Richard Rumelt’s account of good strategy begins with diagnosing the obstacle rather than announcing an ambition. Cummins’ difficulties expose a wider diagnosis: hydrogen investment was premised on rapid growth beyond established industrial chemistry into transport, power generation, building heat and gas networks. The IEA’s 2026 hydrogen review finds that virtually all of the more than 100 million tons consumed in 2025 remained in industry and refining. The hoped-for new markets remain small, and the economics explain much of the shortfall. In my November 11 submission to the European Commission, Europe’s Hydrogen Strategy After Competition, I compared hydrogen with direct electrification, competing industrial processes and imported green products. The central 2050 case put European electrolytic hydrogen at €5.70–€5.81/kg, above the competitive thresholds for many heat, iron and ammonia pathways. Direct-electric technologies also have substantial cost advantages in power, heat and most transport. Firm commitments backed only about one-fifth of newly signed low-emissions hydrogen offtake volume in 2025. The obstacle is paying demand at competitive prices.
Cummins’ customers make that gap concrete. Florida Power & Light installed 25 MW of HyLYZER capacity to turn solar electricity into hydrogen for blending into gas burned at an existing power station. Air Liquide installed 20 MW at Bécancour to expand an established hydrogen-supply complex serving industrial and mobility markets. FPL’s power station already generates electricity from natural gas. Air Liquide has customers who buy hydrogen as a product and can source it through other production routes. That difference changes the value of repairing an electrolyser.
Investors and public funders have two hangovers to consider. A manufacturer leaving the market retains warranty, service and safety obligations. A buyer whose hydrogen application already struggles to compete faces a different decision: pay for restart and continued operation, or retire the equipment. Both problems reach beyond the direct buyer and seller, through lenders, infrastructure operators and public subsidies. The same market forecasts can underpin investments across that entire chain.
Which Cummins customers depend on durable hydrogen demand, and which on markets that have disappointed? What happens to other investors when a supplier or buyer withdraws? Below, I examine the customer evidence and the exit conditions worth testing before the next commitment.
