Iridium Supply: The Real Limit on Green Hydrogen
The world produces about seven tonnes of iridium a year. That number sets the ceiling on green hydrogen.
Green hydrogen has a cost problem, and the industry has settled on an incomplete explanation for it. Almost every discussion of why hydrogen made from water costs more to produce than hydrogen made from gas lands on the price of renewable electricity. Electricity is the largest single input, so that focus is reasonable. It also skips the constraint that decides how many electrolysers the world can physically build.
That constraint is iridium. Every proton exchange membrane (PEM) electrolyser needs it at the anode, the side of the cell where water gives up its oxygen. Nothing else has yet matched iridium's combination of catalytic activity and survival in the brutally acidic, high-voltage conditions inside the cell. And the world produces almost none of it.
Johnson Matthey's May 2026 PGM Market Report puts total iridium supply at 229,000 ounces in 2025, with the same figure forecast for 2026. That is a little over seven tonnes for the entire planet, across every industrial use. For comparison, the world mines roughly 3,000 tonnes of gold a year.
Supply that cannot answer the phone
The scarcity is unusual, and the reason behind it should shape how anyone models a hydrogen project.
Iridium is not mined. It is recovered in trace quantities during the refining of platinum group metal concentrates, typically occurring at concentrations below 0.1 gram per tonne of ore, and over three-quarters of global output comes from South Africa's Bushveld Complex (SFA Oxford). No mine opens because iridium demand rose. Mines open when platinum and palladium economics justify them, and iridium arrives as a rounding error in the refinery output.
This is what separates iridium from lithium or nickel. When lithium demand climbed, capital went into lithium mines, and supply followed with a lag of years. Iridium has no equivalent response. A hydrogen developer signing an offtake in 2030 is drawing on a supply pool whose size was decided by the platinum market.
Demand, meanwhile, has arrived. Johnson Matthey forecasts total iridium demand at 240,000 ounces in 2026 against supply of 229,000 ounces, leaving the market in deficit, and electrochemical applications are now the largest single category at a forecast 109,000 ounces. The report also notes that 2026 should bring the first commercial-scale use of iridium in PEM electrolysis, as two large European green hydrogen projects near completion. The price has responded: Johnson Matthey records iridium hitting all-time highs of around $8,000 an ounce in early 2026, against an average of $4,682 in 2023 (SFA Oxford).
Read those figures together and the picture is a metal already in deficit, priced at a record, entering its first genuinely large-scale application.
The lever is loading, and the targets are public
One number decides whether this becomes a hard ceiling or a manageable cost: catalyst loading, meaning the milligrams of precious metal in each square centimetre of the cell. Halve the loading and each tonne of iridium builds twice the electrolyser capacity.
The published targets are unambiguous about how far this has to move. A 2024 review in International Materials Reviews states that iridium loadings in current PEM electrolysers sit at 2 to 3 milligrams per square centimetre, and must fall below 0.4 milligrams per square centimetre to enable gigawatt-scale deployment (Wang et al., 2024).
The US Department of Energy is more demanding again. Its technical targets for PEM electrolysis record a 2022 status of 3.0 milligrams per square centimetre of total platinum group metal loading, a 2026 target of 0.5, and an ultimate target of 0.125. On performance, the same table records a 2022 status of 2.0 amps per square centimetre at 1.9 volts, a 2026 target of 3.0 amps at 1.8 volts, and an ultimate target of 3.0 amps at 1.6 volts. Those two rows have to be met together, because thinning the catalyst usually costs performance, and a cell that produces less hydrogen per unit of electricity moves the cost problem rather than solving it.
Translated out of laboratory units, the goal is straightforward. More hydrogen from the same electricity, using less of a metal the world cannot make more of.
Where Bspkl comes in
Third Hemisphere client Bspkl manufactures the component this all turns on. A catalyst coated membrane, or CCM, is the part inside a PEM electrolyser where water splits into hydrogen and oxygen: a thin polymer membrane with catalyst layers on each side, iridium on the anode and platinum on the cathode. Most people working in hydrogen have never had to look closely at one, which is part of why the iridium question stays buried in the technical literature rather than in the investment case.
Bspkl's approach starts from a different manufacturing process. The established route mixes catalyst into an ink and coats it onto the membrane, a technique descended from photographic film manufacturing, and a meaningful share of the metal in an ink layer ends up buried where the reaction cannot reach it. Bspkl uses ion beam sputtering, a deposition technique developed for semiconductor manufacturing, in which a beam of ions dislodges atoms from a target and those atoms settle onto a surface one layer at a time.
The adaptation is the interesting part. Conventional ion beam sputtering produces a dense, flat film with almost no surface structure. Bspkl has tuned the process to build a highly nanostructured layer instead, and to run cool enough to deposit directly onto the membrane rather than onto the metallic transport layer beside it, which is where rival thin-film processes have to put their catalyst because they run hot. Nanostructuring creates surface area, surface area makes more of the catalyst available to drive the reaction, and that raises output for a given quantity of metal.
The company publishes a total platinum group metal loading of 0.122 milligrams per square centimetre, of which 0.089 milligrams is iridium (Bspkl). Set against the DOE table, that sits below the 0.125 ultimate target, and well below the 0.4 threshold the International Materials Reviews authors identify for gigawatt-scale deployment. On performance, Bspkl reports 3 amps per square centimetre at 1.9 volts, which matches the DOE's 2026 current density target while remaining above its voltage target, and compares with the 2.0 amps at 1.9 volts the DOE records as 2022 status.
Bspkl is a component supplier, so the honest boundary of the claim is worth stating. The catalyst layer sets how much electricity a cell needs and how much hydrogen it makes, which drives the electricity share of the levelised cost of hydrogen. It does not touch the cost of the electricity itself, the balance of plant, or the capital structure of a project. The company also has catalyst coated membranes in development for anion exchange membrane electrolysers and fuel cells.
Substitution is the lever nobody has pulled
The obvious escape from a scarce metal is to stop using it. Decades of research have yet to produce a replacement that matches iridium for oxygen evolution in acid, because the anode of a PEM cell combines a strongly oxidising environment with an acidic one, and most cheaper oxides dissolve there. Alkaline electrolysis sidesteps the problem entirely by running in a caustic solution with nickel-based catalysts, which is why it remains the cheaper and more mature technology, and why it holds the larger share of installed capacity. The IEA put global electrolyser capacity at 1.4 GW at the end of 2023, of which alkaline accounted for 840 MW and PEM for 300 MW (IEA).
That is a real alternative rather than a full answer. PEM systems hold their position because they respond quickly to variable renewable output and deliver hydrogen at high purity and high pressure, which suits projects pairing an electrolyser with wind or solar generation. As long as those characteristics are worth paying for, iridium demand keeps rising, and loading remains the only lever that moves.
What the cancelled projects tell us
The obvious objection to any of this is that green hydrogen has had a difficult two years, so the constraint hardly bites. Fortescue cancelled two green hydrogen projects in July 2025 after both had passed final investment decision, in Arizona and at Gladstone in Queensland (Argus). Others have pulled back on similar grounds.
Those cancellations were about cost, which makes the materials question more relevant rather than less. The demand that supports hydrogen is not speculative. The world used over 100 million tonnes of hydrogen in 2025, overwhelmingly for refining, fertiliser, and chemicals, and low-emissions production accounted for close to 1 million tonnes, or roughly 1 percent of the total (IEA Global Hydrogen Review 2026). The customers already exist and already buy the product. The open question is the price at which a cleaner version competes, and the answer runs through efficiency and materials.
The single takeaway
Green hydrogen will be decided in two places: the price of electricity, and the amount of iridium each electrolyser needs. The first is a market that many people are working on. The second is a materials science problem with about seven tonnes of annual supply behind it, and the companies solving it are working at a scale most investors have never had a reason to look at. Ask a hydrogen project developer how many milligrams of iridium sit in each square centimetre of their stack. The answer tells you more about the project's cost than the power purchase agreement does.