CPH2 did not merely lose an experimental electrolyzer. Its destructive test failure exposed a mismatch between the product the company promoted and the business it was actually trying to operate. After more than a decade of development, a public listing and roughly £50 million in disclosed financing before its 2026 rescue raise, the company was still treating its membrane-free stack as though it represented most of the commercial problem. The damaged equipment demonstrated that the difficult part was the industrial system wrapped around it. The advertised breakthrough was attractive. CPH2’s electrolyzer avoids membranes, platinum-group metals and PFAS, potentially removing cost, supply-chain and degradation concerns associated with conventional designs. But the architecture generates hydrogen and oxygen together in the stack, passes the mixed gas through dryers and then separates it cryogenically. Once that process choice is made, the commercial proposition includes every pipe, valve, sensor, dryer, chiller, separator and control sequence required to manage a potentially explosive gas mixture under every operating condition. The 1 MW MFE220 unit was approaching the end of factory acceptance testing when an unexpected error initiated its normal automated shutdown procedure. During depressurization, according to CPH2’s initial assessment, a hydrogen-and-oxygen mixture ignited and caused a loss of containment. The equipment was damaged too severely to continue testing, and the company concluded that the mixed-gas system required substantial redesign before it could operate safely in all conditions. Later disclosures identified possible contributors including moisture passing through the cryogenic separation unit, particles in the system and gas velocity through a valve apparently exceeding European Industrial Gases Association guidance. These are not esoteric electrochemistry problems. They are ordinary industrial-process details, which is why their appearance matters so much. A commercial hydrogen system has to remain safe when a valve moves too quickly, a sensor produces a bad reading, moisture reaches the wrong part of the plant or the equipment moves from steady operation into shutdown. Shutdown is not an obscure edge case that can be deferred until after commercialization. Industrial equipment must be validated through startup, shutdown, loss of power, abnormal pressure, failed sensors, maintenance states and foreseeable operator errors. CPH2’s system encountered its decisive failure during a standard automated sequence, and the company subsequently acknowledged that it lacked the financial, engineering and technical resources required for the redesign. The issue was therefore not simply that a prototype broke, but that the failure revealed a product and capital requirement larger than the company was prepared to carry. This is a recurring problem in hydrogen startups because investors and corporate presentations naturally concentrate on the differentiated component. The promoted object might be a fuel-cell stack, an electrolyzer, a pressure vessel, a refuelling module, an aircraft powertrain or a novel production pathway. The customer receives that component inside a much broader operating boundary that includes power conditioning, water quality, cooling, compression, storage, hazardous-area design, ventilation, leak detection, purging, permitting, training, spare parts, maintenance and insurance. “Known components” provide less reassurance than the phrase suggests. A familiar valve does not remain a routine valve merely because it appears in a catalogue. Oxygen service changes materials, cleanliness and ignition requirements, while hydrogen introduces leakage, ventilation and detection concerns. The safety case emerges from how the complete collection of equipment behaves together, especially during transitions and failures, rather than from whether each individual item has been used somewhere before. That distinction should have shaped CPH2’s strategy much earlier. A company whose core capability was membrane-free stack and process intellectual property could have maintained a narrow boundary, placing mixed-gas handling, cryogenic separation, oxygen engineering, plant integration and field support under experienced industrial partners with accountable design authority. The alternative was to become a full industrial original-equipment manufacturer, which required deeper process-safety expertise, manufacturing quality systems, extensive abnormal-operation testing, a service organization and enough capital to honour warranties through a difficult deployment cycle. Instead, CPH2 appears to have occupied the dangerous middle. It was simultaneously inventor, process integrator, manufacturer, deployer, licensor and listed growth company, but it did not possess the industrial depth or financial resources needed when the integrated system required substantial redesign. Its board did not need to second-guess the electrochemistry. It needed to understand that a membrane-free electrolyzer deliberately producing mixed hydrogen and oxygen was hazardous process equipment, not merely a differentiated stack. The wider sector shows why this boundary decision matters. Plug Power chose to own much more of the system, including production, delivery, customer-site equipment, service and fuel logistics, which gave it operating substance while placing compressors, cryogenic systems, field maintenance and customer uptime on its own balance sheet. Ballard historically kept the boundary narrower, selling stacks and modules while integrators, transit agencies and governments carried much of the fuel, infrastructure and operating burden. Its acquisition of GeoPura moves more production, logistics, refuelling and field operations inside the company, making the offer more complete but also more capital-intensive and operationally exposed. Hydrogen vehicle and aircraft companies encountered the same problem in a harsher form. Nikola delivered trucks and dispensed hydrogen before entering Chapter 11, Hyzon moved toward liquidation, and Universal Hydrogen flew a demonstrator before failing to finance the industrial system required for commercialization. ITM Power offers the more useful counterexample: its recent focus on product simplification, manufacturing bottlenecks, factory-acceptance pass rates, capital discipline and full-scope containerized plants is less dramatic than another electrochemical breakthrough, but much closer to the work required to turn a component into dependable industrial equipment. Hydrogen remains an important industrial molecule, and replacing fossil-derived hydrogen in fertilizer, refining, methanol and other established uses is necessary decarbonization work. That does not lower the burden of proof for startups claiming a new production technology or energy application. It makes the system questions more important: who owns the dryer, separator, compressor, valve logic and safety controls; who validated startup and shutdown rather than only steady operation; who services the equipment outside ordinary business hours; and who has enough balance sheet to stand behind the warranty when an ordinary component creates a system-level failure? CPH2’s incident does not show that hydrogen innovation is impossible. It shows that the relevant diligence boundary is the complete industrial system, because that is where safety, uptime, service cost, warranty exposure and capital adequacy meet. The breakthrough may be located in the stack, but the business succeeds or fails in everything surrounding it. Read the full TFIE Strategy Briefing diligence analysis of CPH2 and the hydrogen system-boundary problem.
+12 others
