What Happened to Hydrogen?

In the early 2020s, hydrogen carried an unusual burden of expectation: it was to fuel our cars and trucks, lift our aircraft, heat our homes and decarbonise our heaviest industry. Governments published strategies and a wave of announcements implied that a hydrogen economy was a question of when, not if. Several years on, the picture is far more sober. This issue asks why the promise has so badly outrun delivery, what green hydrogen really costs, why the artificial intelligence boom now works directly against it, and where hydrogen still has a genuine future.

The Everything Fuel

The appeal was clear. Hydrogen reacts to produce only water, carries far more energy per kilogram than a battery, and refuels in minutes. Made by splitting water with renewable electricity, green hydrogen promised a fuel that was clean and, eventually, cheap. The whole case rested on one analogy: that green hydrogen would track the cost curve of solar and wind, where scale and learning drove prices relentlessly down. On that basis, capital queued behind it. The Hydrogen Council has tracked more than 1,500 projects across over 70 countries, some 680 billion US dollars of proposed investment, and the United States set a target of one dollar per kilogram within a decade. The trajectory looked inevitable.

The Reckoning

Inevitability did not survive the balance sheet. Of that 680 billion dollars announced, only around 75 billion has reached a final investment decision — the point of genuine commitment. That gap now defines the sector. In 2025 the industry committed to or began building roughly one million tonnes a year of low-emissions capacity while cancelling more than 4.9 million. For the first time the International Energy Agency cut its outlook: potential 2030 production from announced projects fell from 49 to 37 million tonnes a year in a single annual review.

The casualties span the most committed players. ArcelorMittal shelved a 2.5 billion euro German green-steel conversion despite 1.3 billion euros of offered subsidy; Iberdrola cut its 2030 target by two thirds; Fortescue cancelled projects in both the United States and Australia; Queensland withdrew from a 12.5 billion Australian dollar plant; and the Whyalla green steel project collapsed alongside its steelworks. These were flagship projects backed by serious operators and public money, and they still did not clear.

Nowhere is the retreat clearer than in transport, where much of the promise lived. The passenger-car experiment is effectively over: Toyota, hydrogen's most determined champion, sold just 210 Mirai sedans in the United States in 2025, down nearly 58 percent, and now faces a class action over fuel scarcity. California recorded its first ever decline in fuel-cell vehicles on the road. Shell exited US retail refuelling in 2024, leaving about 70 public stations nationwide where hydrogen costs around 36 dollars a kilogram. Aviation has followed the same arc: in early 2025 Airbus delayed its ZEROe hydrogen aircraft by five to ten years beyond 2035, cut the budget by a quarter, and cited both immature technology and the absence of green hydrogen at scale.

Why the Economics Broke

The solar analogy was flawed in one decisive respect. With solar, the panel is the power. With green hydrogen the electrolyser is only part of the cost; electricity is typically 70 to 80 percent of it. Green hydrogen is therefore not really an energy source but a way of converting electricity into a storable molecule, at a meaningful efficiency loss. It can only ever be as cheap as the abundant, low-cost power it consumes, and it must outbid every other buyer for that power. Worse, the costs meant to fall did not. Electrolyser prices rose as interest rates climbed and cancellations denied the industry scale. The US Department of Energy noted in late 2024 that green hydrogen costs had risen by two to three dollars a kilogram since early 2023. The cost curve did not bend down. In many places it bent up.

Into this picture has arrived a powerful new competitor for the input that matters most. The IEA expects global data-centre electricity demand to roughly double from about 485 terawatt hours in 2025 to 950 by 2030, with AI-specific demand tripling. The technology companies driving this are not passive buyers: they signed roughly 40 percent of all corporate renewable power purchase agreements in 2025. They are buying up precisely the cheap, clean electrons green hydrogen depends on, and they can pay more for them than a hydrogen producer can.

The effect is already in the price. In the PJM market, serving 67 million people across the eastern United States, wholesale power in the first quarter reached 136.53 dollars per megawatt hour, up from 77.78 a year earlier — a jump of roughly 76 percent. PJM's own market monitor attributes most recent capacity-price increases to data-centre demand. For a fuel whose economics are dominated by electricity, a structural rise in the price of electricity strikes the core of the case.

The Next Ten Years

Forecasters have absorbed this. In December 2024 BloombergNEF more than tripled its 2050 cost estimate. It now places current green hydrogen at 3.74 to 11.70 dollars a kilogram and the 2050 range at just 1.60 to 5.09. Even in the cheapest modelled US market, Texas, costs are seen falling only from around 7.22 dollars today to 4.82 by 2030. BNEF expects parity with conventional grey hydrogen — made from gas at roughly one to two dollars a kilogram and broadly stable — only in China and India, and only by 2040.

Costs will fall over the decade, but slowly, unevenly, and from a far higher base than the original thesis assumed. The one-dollar target now looks like a destination for the 2040s, and only in select geographies.

A Fuel for Industry, Not for the Everyday

Hydrogen will not be an everyday fuel. It will not power the family car, heat the suburban home, or, for at least a generation, fly the commercial airliner; in each role it is beaten on cost, convenience or efficiency by electricity used directly. Its future is narrower but real, and it lies in industry. Global hydrogen demand already runs close to 100 million tonnes a year, in oil refining, ammonia for fertiliser, and methanol — almost all of it made today from unabated fossil fuels. These markets are not aspirational. They exist, they are large, and they need a low-carbon molecule that electricity cannot directly supply.

The projects actually proceeding confirm it. Recent final investment decisions cluster in refining and fertiliser, not in cars or planes. Where hydrogen is built at scale it sits beside a customer who already needs it, frequently as blue hydrogen made from gas with carbon capture rather than green from electrolysis, and increasingly in China — now around 65 percent of installed electrolyser capacity. Hydrogen is becoming an industrial feedstock decarbonised in place, not a consumer fuel distributed to the masses.

Where Batteries Reach Their Limit

There is, however, one corner of transport where hydrogen's future looks more assured, and it is defined by a hard physical constraint: a battery is a heavy and bulky way to store energy. Hydrogen holds around 143 megajoules per kilogram against roughly 45 for diesel, and compressed to 700 bar it stores some five to seven times the energy per litre of a lithium-ion cell. Where range, payload and refuelling time matter most, that gap is decisive.

A battery train carrying enough charge for 200 kilometres needs roughly 33 tonnes of cells and the better part of an hour to recharge; a hydrogen train covers 800 to 1,000 kilometres on a single fill in minutes. A fuel-cell truck frees up to three and a half tonnes of payload over its battery equivalent. Battery power has won the short, urban, frequently-stopping duty cycles — in city buses and light rail — on cost. But for long non-electrified rail, where overhead wires cost one to three million euros per kilometre to install, for heavy buses on long or cold-climate routes, and above all for deep-sea shipping, batteries simply cannot carry enough energy. In those segments hydrogen and its derivatives are not one option among several. They are the only zero-emission option that scales.

This is where the investment logic turns, and it turns on time. A ship or a locomotive ordered today will still be working in twenty to thirty years, and the cost of running it on diesel is set to rise for the whole of that life. In April 2025 the International Maritime Organization approved the first sector-wide regime to combine mandatory emissions limits with a global price on carbon, reaching for net-zero shipping by around 2050, with cuts of 20 percent by 2030 and 70 percent by 2040. The order book for alternative-fuel ships is now nearly as large as the fleet in service, set to almost double between 2024 and 2028.