White Hydrogen: The Oil Rush for a Fuel Nobody Has Proved

The 83% headline met a production test

In 2025, HyTerra reported mud-gas samples from its McCoy-1 exploration well in Kansas containing as much as 83% hydrogen. That was a legitimate exploration signal, and the company was explicit that production testing would be needed to establish deliverability and commerciality. On August 4, 2026, it published the receipt: a completed 14-day production test.

At the endpoint, total gas flow was approximately 0.1 Mscf/d, or 100 standard cubic feet per day. Water production was about 300 barrels per day. After correcting for air, the gas contained 0.4% hydrogen, 2.8% helium, 36% methane and 59% nitrogen. The interval had low gas saturation, and the planned test of a second zone was cancelled.

On those endpoint figures, hydrogen flow was roughly 0.4 scf/d—about 4.1 cubic metres per year. A 2026 federal-geoscience paper estimated that a commercially useful natural-hydrogen well may need around 10 million cubic metres per year for two to three decades. That screening figure is not universal, but it puts the receipt in scale: McCoy-1's endpoint was roughly 2.4 million times smaller.

A concentration reading can prove a hydrogen show. Only stabilized purity multiplied by sustained flow can begin to prove a resource.

Real geology is not the same as a repeatable resource

Natural hydrogen is not fictional. Water can react with iron-rich rocks through processes such as serpentinization, and natural radiation can split water through radiolysis. Hydrogen can migrate, accumulate and reach the surface. The mistake is to collapse that chain into a reserve.

A functioning subsurface system needs a source, generation and charge, migration, a reservoir, and a seal that preserves the molecule long enough to recover it. Hydrogen is small and reactive. It can leak, dissolve, feed microbes or be consumed by mineral reactions. A favorable source rock or seep therefore answers only the first part of a five-part project question.

The USGS estimates that the Earth's crust may contain an enormous in-place endowment, but it also warns that most could be too deep, too offshore or too dispersed to recover economically. Its lower-48 prospectivity map shows relative geological favorability, not reserves. A map can tell an explorer where to ask better questions; it cannot tell a lender how much product a completed well will deliver.

Mali proves more—and still not enough

Bourakébougou in Mali is the strongest real counterweight to easy scepticism. A shallow well has supplied a small electricity installation since 2012. A 2023 field study reported gas at roughly 98% hydrogen, a test flow near 1,500 m³/day, and pressure that had not declined after eleven years. The authors interpreted the pressure behaviour as consistent with recharge.

That matters. It proves accumulation, useful flow and a real local application. It does not yet prove a repeatable industrial development model. Annualized test flow is about 0.55 million m³, roughly eighteen times below the 10-million-m³ screening figure, and public long-term production data remain limited. The IEA's 2026 review counts projects in almost 30 countries but only one operating well, in Mali; it does not expect large-scale deployment before the 2040s.

Canada, Europe and China are climbing different rungs

Canada now has evidence stronger than a simple gas show. A 2026 PNAS paper reported more than 140 tonnes of hydrogen per year discharged at a Canadian Shield mine site over a decadal record. InnoTech Alberta has also published the province's first systematic natural-hydrogen prospectivity study, and the Geological Survey of Canada is working on source, migration, reservoir and seal questions. The crucial boundary is that a mine discharge is not a stabilized production test from a completed hydrogen well.

Europe has the most organized public research portfolio. The EU's H2-QUEST program is building models, exploration guidance and field campaigns across several countries. France's Lorraine work found dissolved hydrogen increasing to 18% at depth, paired with 74% methane and unresolved source questions. Albania's Bulqizë mine vents at least 200 tonnes per year in gas containing 84% hydrogen. These are serious scientific records; they are not proved reserves or controlled well deliverability.

China has moved from isolated shows toward national target selection and purpose-built detection. The Songliao scientific-drilling program documented hydrogen-rich intervals, the China Geological Survey is studying the Chuxiong Basin, and Chinese research has mapped prospective regions including Songliao, Bohai Bay, Junggar and ophiolite belts. In the public primary sources reviewed for this episode, however, no independently witnessed Chinese production test reported stabilized hydrogen flow, pressure and decline. That is a documented public-evidence gap, not a claim about private work that may exist.

The cost estimates are not contradictory

White-hydrogen cost claims range from spectacular to sobering. A 2025 model estimated $0.54/kg under optimal conditions. A 2024 study estimated $1.99/kg for a hypothetical facility receiving a very large stream at 83 mol% hydrogen. A 2026 Mali-based model estimated $6.82/kg for ten wells, falling to $2.46/kg for sixty wells only if production did not diminish; a simplified decline case raised cost by almost 70%, and transport added $0.30 to $5/kg.

Each result can be internally correct because each describes a different subsurface and logistics system. Purity sets separation duty. Flow sets how many wells and gathering lines are needed. Exploration success sets the cost of dry or weak holes. Decline sets replacement drilling. Distance sets compression and transport. Until fields report those inputs repeatedly, white hydrogen has assumption curves rather than an operating cost curve.

The cheapest hydrogen model is often a high-flow reservoir assumption wearing a dollar-per-kilogram label.

Low carbon and policy support must be earned site by site

A prospective lifecycle study put a favorable natural-hydrogen case near 3 g CO₂e/MJ, but the result moved with gas composition, depth, productivity, waste-gas handling and energy supply. Mixed streams may require methane management, nitrogen removal, drying, compression and water handling. Those operations consume energy and can create emissions.

The United States' final 45V rules provide a provisional-emissions-rate route for pathways absent from the standard model and discuss geologic hydrogen specifically. That creates a route to qualify; it is not an automatic credit. The well-to-gate ledger still has to count purification and production emissions. The same discipline applies to the market: global hydrogen demand exceeded 100 million tonnes in 2025, but a field needs a nearby customer that can actually use its rate and specification.

Four prospects require four decisions

Appraise a high-purity trapped accumulation near demand when pressure supports a connected reservoir. Combine the ledger for mixed gas containing valuable helium or another co-product, but include all separation and handling costs. Treat high-purity, low-flow or recharge-supported systems as a local niche sized to nearby demand. Stop when the case rests on a mud-gas anomaly, water-dominated production or a modeled resource without stable hydrogen flow.

Across Bankable's five gates, white hydrogen scores 8/10 for physical existence, 2 for deliverability, 3 for economics, 1 for repeatability and scale, and 5 for market and policy fit. The weighted verdict is 3.4/10: real geology, a rational exploration option, and not yet a bankable industrial resource class.

What would change the verdict

The next cheque should buy information, not a commodity forecast. Require an independently witnessed 30- to 90-day stabilized test where the reservoir warrants it; total gas and hydrogen flow; purity corrected for air; wellhead and downhole pressure; decline and shut-in recovery; water and full gas composition; a reservoir model and recoverable range; processing energy; local offtake; site-specific lifecycle carbon; and policy qualification.

When several prospects clear that package and repeat it across wells, the market can begin to draw a cost curve. Until then, the most bankable move is not to dismiss white hydrogen or finance it as proved supply. It is to buy a tightly bounded learning option with a hard stop.

Research cut-off: August 8, 2026. Technology and project-economics commentary only; not investment advice. Primary sources include the McCoy-1 production-test filing, Bourakébougou field study, IEA review, USGS geologic-hydrogen program, Canadian Shield record, Bulqizë study, H2-QUEST and China Geological Survey.