Most hydrogen today is made from natural gas, with the carbon dioxide vented. "Green" hydrogen from electrolysis is clean but expensive, because it inherits the cost of the electricity and the electrolyser. Natural hydrogen, generated underground by geological processes, skips both. If it can be produced at scale, it may be the cheapest low-carbon molecule available.
The proof of concept exists. In Mali, a water well drilled in the 1980s turned out to be leaking almost pure hydrogen, which since 2012 has been used to generate electricity for the village of Bourakebougou. Exploration licences for natural hydrogen have followed in Europe, the United States and Australia. The excitement is justified. The questions that will decide which discoveries become resources are, however, the ones reservoir engineers have been asking about oil and gas for a century, with a few twists.
1. Is it a stock or a flow?
Oil and gas are stocks: a fixed volume generated over geological time, which we deplete. Hydrogen is generated continuously by reactions such as the serpentinisation of iron-rich rocks and the radiolysis of water. If a reservoir is recharged at a meaningful rate, it behaves partly like a renewable flux rather than a tank. That changes everything about the development plan: well spacing, plateau length, even the definition of reserves.
The engineering question is the ratio between the generation rate and the planned offtake. Very few datasets can yet answer it, and appraisal should be designed around measuring it.
2. Will the trap hold the smallest molecule?
Hydrogen is tiny and diffuses through seals that hold methane comfortably. Salt and other evaporites are excellent seals; many shales are less certain. A seal that is "proven" for gas is not proven for hydrogen, and the leakage rate over geological time sets how much can have accumulated in the first place.
3. Who else is eating it?
Underground, hydrogen is food. Microbes consume it to make methane, or to reduce sulphate to hydrogen sulphide. The same microbial chemistry that sours oil reservoirs during waterflooding can consume a hydrogen accumulation, or contaminate it. Any hydrogen resource assessment that ignores the biology is incomplete, and souring models built for oil fields are a useful starting point.
4. Is the measurement real?
Drilling itself can produce hydrogen: steel and drilling fluids react at the bit and generate gas that shows up on the mud log. Soil-gas surveys are sensitive to many shallow sources. Before a show is counted as a discovery, the measurement has to be separated from the artefacts of making it.
5. Can the well survive it?
Hydrogen embrittles some steels, and hydrogen sulphide (see point 3) makes it worse. Material selection, cement integrity and the gathering system have to be specified for hydrogen from day one, not converted later.
Natural hydrogen will reward the operators who appraise it like a new resource class, not those who drill it like a gas prospect with a different label.
The decision lens
Early natural-hydrogen projects are small, uncertain and information-rich. That is precisely where staged, modular development earns its keep: an appraisal well and a small power unit generate both revenue and the data that decides the next step. Each stage should carry its own kill criterion, written before the results arrive. I explain why that framing matters in modular energy as a real option.