The Underground Hydrogen Rush: A Cheap Green Fuel or Just a Pipe Dream?

Posted on 20 September, 2026 by Mehrdad Fathi

For most of the past century, the idea that hydrogen gas could be sitting underground in commercially viable deposits simply did not register as a possibility worth entertaining. Geologists knew the Earth produces hydrogen constantly through reactions between iron-rich rock and water, and through the splitting of water molecules by radioactive decay. They also knew that microbes eat it, that geochemical reactions consume it, and that its molecules are small enough to slip through rock formations that trap oil and gas. The logical conclusion was that none of it sticks around.

That assumption has started to fracture — and with it, a quiet but accelerating race to find out whether the Earth’s subsurface holds enough hydrogen to matter.

How the Assumption Broke

The shift began in earnest in 2018, when researchers documented a substantial hydrogen deposit in Mali — not a trace amount, but a reservoir substantial enough to support a small power plant. The discovery was reported in the International Journal of Hydrogen Energy and caught the attention of geologists who had spent careers thinking about hydrogen from a microbiology angle rather than an energy one.

“The first time it hit the scientific literature, it caught our attention,” says Chris Ballentine, a geologist at the University of Oxford.

Supporting evidence has accumulated since. In 2024, researchers reported that a chromite mine in Albania releases at least 200 tonnes of hydrogen annually — one of the highest natural flow rates ever recorded. If hydrogen can accumulate in a mine shaft, the argument that it cannot accumulate under cap rock starts to look like an assumption rather than a fact.

A sweeping 2024 analysis in Science Advances put some numbers on the possibility: Earth’s subsurface probably contains tens of trillions of tonnes of hydrogen. Even if a fraction of a percent of that were recoverable at a profit, it could cover projected clean energy demand for 200 years — potentially at costs competitive with or below fossil fuels.

The Rush to Drill

That estimate is why, since 2023, venture capitalists have poured nearly $500 million into geological hydrogen exploration. Companies are now drilling or prospecting in nearly 30 countries: northern Canada, eastern Africa, South Australia, and the heart of North America, where the continent began tearing apart nearly a billion years ago and then stopped, leaving behind iron-rich rock formations in Nebraska, Kansas, and Iowa. H2Au, a hydrogen exploration company based in Epsom, UK, started drilling in Kansas in August 2026.

Pinpointing where to drill is not simple. Geophysicist Mengli Zhang at the University of Georgia describes a layered remote-sensing process using satellite and aircraft data. One reliable indicator is the magnetic signature left when olivine — an iron-rich rock type — reacts with water and forms magnetite. The most promising targets appear to be where that kind of iron-rich geology sits in close proximity to impermeable cap rock, mirroring the structural logic of conventional oil and gas traps.

Eric Gaucher, who co-leads the natural hydrogen taskforce at the International Energy Agency, summarizes the basic geology: the Earth produces hydrogen at large volumes and also destroys it at large volumes. The question is whether, in specific geological configurations, enough survives to be worth producing.

What Still Needs to Be Answered

The fundamental unknowns are substantial. Researchers do not yet have a confident picture of how quickly hydrogen forms under different geological conditions, how it migrates through different rock types, where and under what structural configurations it accumulates, or how long it persists once trapped. Ballentine’s group is working on formation rates across different iron-containing rock types and investigating how dissolved minerals can accelerate hydrogen generation via radioactive decay.

“With that fundamental information, we can start refining exploration strategies,” he says.

Environmental risks also require attention. Hydrogen can contribute indirectly to warming by interacting with atmospheric chemistry, so production at any scale would require tight containment. Many natural hydrogen deposits are likely to contain mixed natural gas, and processing that impurity would produce some greenhouse emissions — though the carbon footprint, Ballentine says, should be modest.

The broader clean-energy framing holds: every zero-carbon energy technology involves mining or resource extraction somewhere. Solar panels, wind turbines, and batteries all require minerals. Geological hydrogen would not be different in kind, though the specific tradeoffs are not yet quantified.

What the Next Few Months Could Reveal

The practical horizon for a first answer is closer than it might seem. Several of the exploratory wells now being drilled could produce results within months — either flows of hydrogen gas sufficient to establish proof of concept, or dry holes that redirect the search. Ballentine estimates that if geological hydrogen is going to demonstrate viability at scale, there should be flowing gas within two to three years.

“At the moment, we still don’t know if it’s almost nothing or completely everywhere,” says David Waltham, a geophysicist at Royal Holloway, University of London.

That candor captures where the field sits. The potential is large enough to justify serious investment. The evidence base is thin enough that serious investment carries real risk. The companies drilling now are, in effect, running the experiment that determines which of those descriptions turns out to matter more.

Source: Nature 657, 590–591 (2026). doi: https://doi.org/10.1038/d41586-026-02850-5

References:

1. Ellis, G. S. & Gelman, S. E. Sci. Adv. 10, eado0955 (2024).

2. Ballentine, C. J. et al. Nature Rev. Earth Environ. 6, 342–356 (2025).

3. Prinzhofer, A. M., Tahara Cissé, C. S. & Diallo, A. B. Int. J. Hydrog. Energy 43, 19315–19326 (2018).

4. Truche, L. et al. Science 383, 618–621 (2024).