Geothermal energy projects rarely fail because the heat is missing. They fail because risk is taken in the wrong order.
Capital is committed before the subsurface is understood. A conversion technology is chosen before the reservoir is proven. Development accelerates before the uncertainty that decides the outcome has been reduced. The hard problem in geothermal is not finding heat. It is turning geological uncertainty into financeable infrastructure, and that is a problem the petroleum industry is already built to solve.
It is worth being clear about why this became a petroleum problem at all. For most of its history, geothermal was a conventional business waiting on favorable geology. That changed around the middle of the last decade. By then the shale sector had spent 10 years refining two techniques, horizontal drilling and hydraulic fracturing, that turned out to be close to what geothermal needed in order to reach heat where nature never supplied the permeability to move hot fluids. Drill a long lateral into hot rock, create flow paths through it, and circulate water through formations that were previously worthless for power.
Geothermal stopped being only a question of where nature left an obvious reservoir and started to become an engineering problem the world’s most capable drilling industry already knew how to attack. The people who understand that drilling are, in large part, already in this industry.
That engineering comes in three broad forms, and the distinction matters for how a project is derisked. A conventional hydrothermal resource has all three ingredients in place: heat, fluid, and the permeability to move it. You find it, drill it, and produce it. An enhanced geothermal system has the heat but not the permeability, so wells are drilled into hot, tight rock and stimulated until water can circulate. In contrast, a closed-loop system, also known as an advanced geothermal system, runs fluid through a sealed pipe in hot rock and never contacts the formation at all.
None of these is categorically better than the others. The discipline is to match the technology to the reservoir, not the reservoir to a preferred technology. A high-enthalpy conventional resource does not need an engineered reservoir; where the rock is hot but tight, stimulation earns its place. Treating every play as a proving ground for one favored method is how good acreage gets developed badly.
Whatever the method, geothermal shares one financial shape: it is front-loaded.
Plot risk and capital against time and the two curves run opposite each other. The first phases, pre-survey through exploration, test drilling, and planning, cost a few million dollars and take 2 to 3 years, and that cost is largely independent of how large the resource turns out to be. This is where technical uncertainty is resolved. The build phases that follow, full-field drilling through construction and start up, run several million dollars per megawatt and scale with the size of the resource. This is where capital expands quickly.
Then comes operation: stable cashflow, low operating cost, and a plant life measured in decades. Larderello, in Tuscany, is the world’s first geothermal project and has produced power for more than a century.
The consequences of not aligning operational goals are unforgiving and specific to this business. The money is spent, and the geological risk is taken, before the plant produces a single kilowatt. You cannot correct a disappointing reservoir after the fact the way you can refinance a merchant power plant or retool a production line. That is what front-loaded really means, and it is why sequencing is everything.
However, this asymmetry is also a big opportunity. Most of the technical uncertainty is retired in the cheap early phases; the largest expenditures come later, after the reservoir question has been answered. A project that respects that order is financeable (Fig. 1). A project that inverts it, drilling a field or ordering long-lead equipment before the resource is bounded, is not, no matter how much heat sits under the lease.
Respecting the order depends on reading the subsurface honestly, which starts with separating signal from noise. Early exploration produces plenty of encouraging data: a thermal anomaly, a magnetotelluric conductor, hot springs at surface. None of those, on its own, is a reservoir.
Every petroleum engineer knows the difference between a show and a producer. The geothermal equivalent is the convergence of heat, permeability, and fluid. A high temperature gradient with no flow path is a dry hole with a good story. Only when several independent data sets point to all three does an anomaly become a repeatable well design, and eventually a plant. Even then, the surveys have only narrowed the range. A confirmation well is what retires the bulk of the technical risk, and the purpose of all the work before it is to reach that well as informed, and as cheaply, as possible.
Financeability, then, is not something that appears at the end of a successful project. It is engineered from the first survey. The work runs in a deliberate order. Assemble and reconcile the existing geological, geophysical, and legacy well data into one coherent picture before spending on new acquisition. Take a well-chosen land position, because poorly sized or poorly located acreage constrains every decision downstream. Then acquire the field data, gravity, magnetotellurics, temperature gradients, and structural interpretation, integrated into a single subsurface model rather than read one survey at a time.
By the time a rig arrives, drilling should be validation, not exploration. Each step is built to answer a short list of questions before the next tranche of capital is committed: is there enough heat, is there a viable flow path, and can the system be developed safely and repeatably. Each stage either produces converging evidence or it does not, and the honest answer sometimes ends the project. Discipline means killing a weak prospect early and cheaply, on data, rather than drilling to confirm a hope.
A portfolio is what makes that discipline affordable. A single project is fragile, because one underperforming well can stall an entire development. Across a portfolio, risk becomes statistical, learning compounds from basin to basin, and capital moves to the assets that are earning it.
This is what changes the cost of capital, and it is the part the finance side of our industry recognizes immediately. Early, unbounded geological risk is venture risk, and it prices accordingly, with required returns above 30%. A proven, permitted, drill-ready resource is infrastructure, and it prices in single digits. The task of a geothermal developer is to walk a project across that gap on purpose, retiring risk in the right order so the asset re-rates from a speculative exploration play into financeable infrastructure. Put plainly, the job is not to raise cheap capital early. It is to earn cheap capital later.
That gap is crossed by rotating the type of capital as the risk profile changes. The earliest work, screening and exploration, is venture risk, and it is funded like venture risk. Once a confirmation well and a resource assessment bound the reservoir, the project can carry growth or private equity at a lower hurdle. When it is permitted and drill-ready, infrastructure equity will underwrite it, and once it is generating, project-finance debt will refinance the build at a fraction of the early cost of capital. Each class of capital enters when the remaining risk matches its appetite, and no sooner.
A developer who tries to fund a drilling program at the cost of capital of an operating plant will not find the money, and one who funds early exploration with infrastructure equity is wasting it. Matching the capital to the risk at each step is as much a part of the engineering as the well design (Fig. 2).
For the petroleum professional moving into geothermal, the practical steps are specific. Bring the subsurface discipline you already have and apply it in order: characterize before you commit and treat the confirmation well as the event that earns the right to scale, not a formality on the way to a plant that has already been designed. Be skeptical of any pitch that starts with a technology and works backward to a resource.
Develop across a portfolio rather than betting a company on a single well, so that one disappointing result is a data point and not an ending. And price capital to the risk that has actually been retired, not to the risk you hope to retire. None of this is exotic. It is ordinary upstream practice, applied to a resource the industry is only beginning to treat as its own.
Geothermal’s opportunity is real, and so is the temptation to treat it as a technology race, with each new method promising to leapfrog the last. The projects that reach commercial scale will be the ones that treat it as a subsurface risk problem and solve it in the right order: identify the risk early, reduce it with data, and hand it to infrastructure capital once it is bounded. That is not a new skill set. It is the one the petroleum industry has spent a 100 years building. Geothermal is where it gets applied next.
Marcus Oesterberg, SPE, is the chief operating officer of Ignis Energy, a geothermal development company he joined as its first employee in 2022 and where he leads development of a geothermal portfolio across the US, Türkiye, Italy, and Indonesia. He has worked in the oil and gas industry for 32 years and holds 19 US patents. He began his career at Baker Hughes in 1995, holding roles across technical services, deepwater project management, global marketing, and operations, and served as director of Gulf of Mexico drilling operations and as global product management executive for directional drilling. Before Ignis he was president of Evolution Engineering, a downhole measurement-while-drilling technology company acquired by Baker Hughes in 2021. He also serves on the board of GeoAlaska and has published with SPE, Geothermal Rising, and the Unconventional Resources Technology Conference. He holds an MSc in mechanical engineering from the Technical University of Clausthal in Germany.