BankableReal Bet? · The Green Hydrogen Teardown · one-page verdict

Green hydrogen: overhyped and a real bet

Not a hedge — two different questions with two different answers. As the fuel of everything, the physics already decided. As the cleanup of the hydrogen we already make, the demand is sitting there today. Companion to the video episode; every number below traces to the cited research file.

Swipe the table to see every column →

Ladder framing after Michael Liebreich [S10] · IEA Global Hydrogen Review [S1] · Lazard LCOH [S4]. Full [S#] mapping in the workbook's Sources sheet.

The number that decides everything

• 50–55 kWh of electricity per kilogram in a commercial electrolyzer. The thermodynamic floor is 39.4 kWh — you approach it from above. There is no Moore's law here; the remaining headroom is roughly 20%.
• That kilogram carries only ~33 kWh as fuel. Convert it back to electricity and you keep 30–45% (~37% typical) — against 85–95% from a battery.
• Electricity is 60–70% of the cost of every green kilogram. So this is not a bet on hydrogen. It is a bet on absurdly cheap, absurdly available power — and most of the planet does not have it.
• Sensitivity: every $10/MWh of power price moves roughly $0.50/kg. That is the lever; everything else is second order.

Does it pencil? Not yet, and the gap is specific

• PEM, US, unsubsidized: $4.33–6.05/kg (midpoint ~$5.25). Grey hydrogen from steam methane reforming: $1.5–2.5/kg. Same molecule, two to three times the price.
• Take the full $3/kg US credit and green lands at $2.36–4.08 — the best case only just reaches the top of grey's range.
• Grey emits 9–10 kg CO₂ per kg of hydrogen. For carbon pricing alone to close a ~$3 gap you need roughly $300/tonne. No major carbon market is close.
• Read that plainly: the sector does not currently run on economics. It runs on policy.

Read your grid before you choose the machine

• Solar alone is the trap. Cheapest power on paper, but a solar plant delivers roughly a quarter of nameplate over a year — so the expensive stack idles three quarters of the time, and idle capital is what wrecks $/kg.
• Two ways out, both cost money: oversize the array and add storage to buy running hours, or take grid power around the clock and accept a higher price plus a dirtier carbon intensity. The real constraint is not the cost of cheap solar — it is the availability of cheap solar, plus the battery bill.
• Firm hydro changes the answer. Steady output means the cheap alkaline machine runs near flat out and you can skip the fast-ramp premium. Quebec, Manitoba, British Columbia, Washington State, Norway: roughly $3/kg rather than $5 — still above grey's $2, but inside the range where today's subsidy or a modest carbon price closes it.
• Which is why the projects actually operating cluster where power is cheap and firm. Green hydrogen has never been one price. It is a bet on your grid.

Follow the money — the discipline gap

• Announced clean-hydrogen capacity for 2030: 37 Mt/yr, down from 49 Mt/yr a year earlier.
• Share of announced projects that reached final investment decision: ~9%.
• The bottleneck is almost quaintly old-fashioned — nobody signs the offtake. New agreements fell from 2.4 to 1.7 Mt/yr, roughly a third. Meanwhile ~80% of the projects that are advancing have a committed buyer behind them.
• Hence 2025's exits: ~60 large projects cancelled, including a 26 GW Australian megaproject and three US projects from one of the world's largest hydrogen producers. That is not market failure. That is capital repricing hype at cost.
• Policy shapes the map: the US credit pays up to $3/kg on a carbon-intensity ladder but shuts to projects not under construction by end-2027; Canada pays a refundable 40/25/15% on equipment by carbon intensity; Europe auctions the subsidy and lets producers bid their own floor. China simply won the machines — around 60% of global electrolyzer manufacturing, roughly two-thirds of installations, alkaline systems at $600–1,200/kW against ~$2,000 in the West.
• But winning the factory is not the same as winning the molecule: the largest green plant in the world has run at roughly 20% of nameplate. The utilization trap is real, and it is the same trap as the solar one.

Scored

• Maturity 6/10 — the machines are commercial and for sale; gigawatt-scale reliable operation is not yet proven.
• Economics 2/10 — two to three times the incumbent, with only two exits: ~$300/t carbon, or a subsidy with an expiry date.
• Scale-up 3/10 — enormous addressable market, but 9% FID is 9%.
• Moat 4/10 — serious players and serious capital, but equipment is commoditizing fast and long-term contracts are not being signed.
• Timing 3/10 — the shakeout is not over and the largest subsidy is on a countdown.