Hydrogen's Problem Was Never Production

So when IEEE Spectrum reports that researchers can stimulate hydrogen production by running electrical current through iron-rich rock underground, my first instinct is the usual one. And then the geology gets in the way of my cynicism, which is annoying.

Here is the part that is actually true: the planet makes hydrogen. It has been making it for four billion years. When water meets iron-rich ultramafic rock — olivine, peridotite, the stuff of the upper mantle — the iron oxidizes, the water gives up its oxygen, and hydrogen comes off. The reaction is called serpentinization and it runs continuously, at scale, in ocean crust and continental basement, requiring permission from nobody. In 1987 a drilling crew in Bourakébougou, Mali, looking for water, hit a pocket of it. A driller leaned over the hole with a cigarette and the well caught fire. They capped it for twenty-four years. When it was finally reopened, the gas turned out to be about 98 percent hydrogen, and it has been running a generator for the village ever since.

That well is, four decades on, still essentially the only geologic hydrogen well anywhere that has actually powered anything. Hold that thought.

In 2024 the USGS published a model putting something like 5.6 trillion tonnes of hydrogen in the Earth's crust, with the immediate caveat that most of it is too deep, too diffuse, or too far offshore to ever touch. Even a couple of percent, they noted, would cover global hydrogen demand for centuries. That is the kind of number that moves capital, and it did — Koloma, Hy-Terra, a scatter of others, funded substantially by people who made their money in oil and gas, because those are the people who know how to drill a hole.

The stimulation research is the genuinely interesting middle. Not extraction, not manufacture — catalysis. If serpentinization is a reaction that wants to proceed and is merely slow, then the engineering problem isn't producing hydrogen at all. It's turning up the rate on a process that is already running. Nudge the rock. Come back for the gas.

I want to be careful here, because this is precisely the shape of thing I would normally take apart. "It's already down there, we just have to get it out" is a sentence with a history. It's the shale sentence. The apparatus — the land leases, the seismic surveys, the press releases about a resource "the size of Saudi Arabia" — always arrives well before anyone demonstrates a flow rate that sustains itself for a year.

And even if all of it works, hydrogen's problem was never production. It was the molecule. Hydrogen is the smallest thing there is. It leaks through metal, embrittles the steel it leaks through, and carries about a third the energy per unit volume of natural gas even after you chill it to twenty degrees above absolute zero — which itself burns roughly a third of the energy you were trying to store. Leaked hydrogen is an indirect greenhouse gas, something like eleven times CO2 over a century, because it extends the atmospheric lifetime of methane. A free wellhead fixes none of that. Most of the applications hydrogen has been marketed for — cars, home heating, grid storage — are better served by electrons in a wire, and always were.

But not all of them. Around a hundred million tonnes of hydrogen get made every year, nearly all of it from fossil fuel, and most of it goes to two places: ammonia for fertilizer, and refining crude. Haber-Bosch feeds something like half the people currently alive. Steelmaking wants hydrogen as a reducing agent for the same chemical reason. Those are jobs electrons cannot take. If there is cheap non-fossil hydrogen sitting in the basement rock, that is where it would matter, and it would matter enormously. Credit where it's due — that's a real thing to care about.

The coherenceist read is almost too tidy, which is the first thing that should worry anybody about it. Electrolysis is force: shove current through water against the gradient, accept your seventy percent, compress the result to seven hundred atmospheres, and call the losses a rounding error. Geologic hydrogen is alignment: find where the planet's own four-billion-year reaction is already running and position yourself downstream of it.

Except that's two different things wearing one name. Bourakébougou is alignment — the reaction ran on its own, the gas accumulated on its own, and a drill bit arrived four billion years late to a process that had never required anything from anyone. Stimulated hydrogen is not that. You are injecting electrical current into rock to drive a reaction faster than it would otherwise go. Spend energy, get hydrogen: structurally, that is the electrolyzer's move performed in a worse-characterized reactor on a longer timescale. The number that decides which it is would be energy in per kilogram of hydrogen out, stimulated basement rock against a commercial electrolyzer — and as far as I can find, nobody has published it. That silence is itself informative. If stimulation lands worse than electrolysis on that ratio, "alignment" isn't a philosophy of energy. It's a less efficient electrolyzer wearing four billion years of planetary chemistry as a credential.

Which is this piece's own accusation, turned around and aimed at its own thesis. Crediting a current-driven industrial technique with the virtue of the spontaneous process it merely speeds up is the shale sentence in a nicer coat, and I nearly wrote it without noticing.

There's a larger thing underneath, and I buried it twice as a punchline — once as "people who made their money in oil and gas," once as a slide about mineral rights. It isn't color. It's the structure of the whole story.

Electrolysis makes hydrogen a manufactured good. Anyone with water and electricity produces it, at any scale, on a roof, without asking. Geologic hydrogen makes it an extracted good: subsurface, leaseable, enclosable, concentrated in whoever holds title to the rock. That is not an efficiency change. It is an inversion of who is permitted to participate, and it runs the wrong direction — the inefficient option is the distributable one, and the efficient option may turn out to be the re-enclosing one. The same landholding apparatus that owns the fossil economy, handed a cleaner molecule and a considerably better story.

Bourakébougou is the tell. I won't claim the well is unique purely because nobody worked out how to own it — that accumulation is a genuine geological outlier, and finding a second one is hard work, not merely unglamorous work. But this much is fair: a village-scale well running a generator for a few hundred people generates no lease, no royalty, and no press release, and so nothing in the current apparatus is aimed at reproducing it. Mali is not the model the industry is walking toward. It is what happened the one time the chemistry got there before the leases did.

So, the prediction. The geology is real. The chemistry is real. And by the time anybody proves a commercially sustained flow rate outside of Mali, there will be four hundred press releases, eleven companies, three acquisitions, two bankruptcies, and one village in West Africa that has quietly been doing it since 2012 — not because the village solved something the companies can't, but because no one has yet worked out how to charge it rent.