Exposing the Learning-Curve Myth: Why Enhanced Geothermal Cannot Follow Solar’s Cost Trajectory – CleanTechnica

---

Support CleanTechnica’s work through a Substack subscription or on Stripe.

---

Enhanced geothermal systems have been positioned by advocates as a scalable source of firm, low carbon electricity that can complement wind and solar. The idea has appeal. The deep heat of the crust is available everywhere and geothermal energy is not intermittent. Policy agencies have published scenarios showing EGS costs falling over time on learning curves similar to solar and batteries.

EGS solutions aim to create engineered reservoirs in deep, hot rock where natural permeability is low. The basic approach adapts directional drilling techniques from the oil and gas sector to reach targeted zones several kilometers underground, then uses fracking to open pathways through the rock so water can move through the heated region. Once the reservoir is created, injection wells send water into the fractured rock and production wells bring the heated fluid back to the surface, where a surface power plant transfers the heat into a working fluid that boils at a lower temperature and drives a turbine. This type of power plant is called a binary or Organic Rankine Cycle (ORC) plant, but in simple terms it is a closed loop system that turns moderate geothermal heat into mechanical and electrical power. EGS is an engineered geothermal method built on mature drilling and hydraulic fracturing practices rather than the naturally permeable hydrothermal resources used by conventional geothermal power.

Outlooks optimistic for falling EGS costs assume that as more wells are drilled and more projects are completed, costs will follow the familiar Wright’s Law pattern and decline with each doubling of cumulative production. This raises a basic question. Will EGS actually behave like solar or batteries on a cost curve?

When we look at the characteristics of the technology and the realities of heavy infrastructure projects, the answer is no. EGS will improve as experience accumulates but it will not experience the deep cost declines that made solar and batteries inexpensive. It also begins at a cost position that is many times higher per MW than widely deployed thermal, nuclear, wind, solar or hydroelectric facilities. Even decades of learning and a 40% improvement would still leave it far more expensive than mainstream options. That creates at best a niche role for EGS rather than a broad one.

In the recent report I published on geothermal, Beyond the Hype: Geothermal in Context, I assessed enhanced geothermal systems, including its many fiscal challenges to achieve competitive prices, but didn’t assess its experience curve potential explicitly. While reviewing a draft set of materials from the Energy Transitions Commission to which they had invited my input, I noticed a reference, one surrounded by cautionary data, for the potential for EGS to see learning curve cost reductions similar to other energy technologies. Implied was solar, batteries and wind energy. Because I had been asked by the same group to provide feedback on their draft perspectives on nuclear power and geothermal in earlier cycles, I knew my published work on both subjects has begun to inform some of the directions of their investigations, scenario models and policy conclusions. The contrast between their suggested EGS optimism in that one reference and my cost-curve assessment for electrolysis facilities vs solar and batteries compelled me to provide an emailed response to them on why that mattered for EGS, and then to expand it for this article. Given the interaction, I’ve added this article to the whitepaper, slightly modified, to ensure it’s a more evergreen resource.

The Energy Transitions Commission was formed as a high level coalition of industry leaders, financial institutions and research organizations focused on guiding global decarbonization pathways. It brings together serious players from sectors that shape energy systems, including companies such as Orsted, BP, Shell, Schneider Electric, Siemens Energy, HSBC and Bank of America. It also includes influential institutions such as the Rocky Mountain Institute, the World Economic Forum, the European Climate Foundation and Mission Possible Partnership. The group’s purpose is to build analytically grounded roadmaps for reaching net zero while balancing economic competitiveness, investment requirements and practical deployment constraints across regions. Because the Commission represents both incumbent energy companies and emerging clean energy leaders, its reports carry weight in policy circles and among investors who use the analyses to benchmark technology pathways and system planning assumptions, hence my willingness to provide at least some pro bono insights and suggestions to the analysts. My focus is on getting the trillions needed for the decarbonization transition spent more quickly and wisely, and the Commission is in an influential position.

Learning curves are a simple idea. As cumulative production doubles, cost drops by some percentage. This relationship shows up in many manufactured goods. It works best when the product is standardized, production cycles are fast and factories can repeat the same process thousands or millions of times with incremental improvement. Screws and washers are the ultimate example, seeing doubling discounts in the 27% range.

Solar panels are a good example in energy. Early panels were expensive because factories had little experience with wafer production, metallization and module assembly. Over time each doubling of output led to process improvements, tooling improvements and supply chain gains. Solar modules experienced a learning rate of about 20% per doubling. Batteries followed a similar pattern with about 19% learning as cumulative output of cells increased. These technologies are made in highly automated environments with short cycle times. They have very large global markets. A doubling of volume has often occurred every one or two years. Wind turbines, although more complicated and bulky, still include manufactured blades, nacelles and electronics that scale through factories. Their learning rate is lower, around 10% to 15%, but the principle is similar. When the technology is simple to replicate and can be produced at industrial speed, cost reduction follows naturally.

These high learning rate technologies all share one feature. They are repeatable manufactured products. Solar panels follow the same architecture regardless of location. Batteries follow standardized chemistries and formats. Wind turbine components are assembled in factories. In each case the learning events are frequent because factories can run thousands of cycles per year. The feedback loop is tight. Engineers adjust designs, supply chains refine processes and costs decline.

Where construction complexity enters the picture, learning slows. Solar construction learning curves remain strong because the majority of solar is constructed on relatively flat ground with simple mountings and wide geographic dispersal of good sites. Turbine tower erection, foundation work and transmission interconnection do not follow as steep learning curves because these activities depend on location, resource conditions and site conditions. This mix of fast factory learning and slower field work still yields material cost decline for wind. EGS does not have any of these conditions.

EGS assembles mature technologies rather than refining relatively new ones. Directional drilling has been refined over many decades. The global oil and gas sector has drilled millions of wells. Complex directional drilling from a pad is a mainstream practice. The shale oil and gas boom pushed these techniques to a high degree of refinement. Hydraulic stimulation is also mature. Operators in North America have completed tens of thousands of frac stages every year. The equipment, materials, logistics and workforce processes for stimulation are well known. ORC plants and the balance of plant associated with thermal power have also been refined over decades. Substations and grid equipment are standard industrial products. EGS combines these elements but does not create any new fundamental technology. Because each component is already on a flat part of its cost curve, the combined system inherits that flatness.

Construction cycles put another boundary around potential learning. EGS projects require drilling several deep wells, stimulating the reservoir, circulating fluids, testing the thermal performance, building an ORC plant and securing a transmission interconnection. These steps take several years. In practice an EGS project may need four and a half to seven years from first drilling to commercial operation. When each project takes years to complete, the opportunities to double cumulative output are limited. Even in a very optimistic scenario the global industry might complete another doubling every decade after the first commercial systems go into operation, assuming that they get there. Solar panels and batteries doubled output every one or two years. EGS might double two to three times in a full generation. Slow cycles prevent steep learning because the repetition rate is low and the process is driven by field construction rather than factory production.

Drilling in particular will see slow learning for EGS. The global drilling industry has already drilled millions of wells. The learning curve has flattened. Efficiency gains during the shale boom came from thousands of wells drilled annually in each basin, pad drilling, advances in bits and motors and supply chain coordination. EGS will never drill at that scale. It may eventually drill a few thousand wells globally. That volume is a rounding error in the context of the existing well base.

Early EGS wells showed improvements in drilling days as teams gained familiarity with hotter and harder rock. These gains are normal first of a kind to next of a kind transitions. They do not represent a long term slope. Comparisons to the early shale fracking learning curve, which some analysts point to, are not appropriate for EGS because shale operators began with a clean slate and climbed a steep learning path by drilling thousands of wells per year in basins with no prior horizontal drilling history. EGS begins from a completely different position, entering a global industry that has already drilled millions of wells and refined directional drilling and fracking over decades, which means the steep early learning seen in shale is long gone and not available to EGS. Mature industries such as deepwater drilling, mining and conventional geothermal tend to show learning rates around 3% to 7% per doubling, providing the real range of discounting with doubling after the basics are overcome.

The cost structure of EGS starts at a position that is far higher per MW than every mainstream generation option. EGS wells are expensive because they are deep, technically challenging and low throughput compared to oil and gas wells. The ORC plant and balance of plant add further cost. EGS projects reported today often sit at $10,000 to $15,000 per kW. Solar farms are in the range of $800 to $1,200 per kW in many locations. Onshore wind often lands between $1,300 and $1,700 per kW. Combined cycle gas plants are around $1,000 to $1,200 per kW. Hydrothermal geothermal plants often sit between $3,000 and $5,000 per kW. Advanced nuclear projections in the 2030s are in the $4,500 to $6,500 per kW range. Even if EGS saw a 40% cost reduction over several decades, it would still sit at $6,000 to $9,000 per kW. That remains higher than mainstream options. Because the learning cycle is slow and the number of doublings is small, that 40% is spread over decades. This cost gap will not close by 2040 or 2050. EGS remains an expensive form of generation even when optimistic assumptions are applied.

The operational characteristics of EGS create further economic pressure. EGS plants require high annual output because fixed costs are so high. The wells, pumps, ORC plant and monitoring systems continue to incur cost regardless of dispatch level. Curtailment does not reduce O&M materially. If an EGS plant is asked to load follow wind and solar, its effective capacity factor drops while its costs remain steady. There is some flexibility in pre-injecting water in advance of perceived demand, increasing the artificial reservoir pressure, but that equates to refracking the same volume over and over—deeply contraindicated in the literature as I discovered in assessing a fracking for underground pumped hydro proposal—and potential for reservoir loss of cohesion with bleed of working fluids out of the reservoir. It’s much less technically concerning to not circulate fluid through the created reservoir than to overpressure it, meaning it’s easier to turn down than turn up past its design point. The cost per MWh rises quickly. An EGS plant designed as a 100 MW baseload generator looks uneconomic if it is curtailed regularly. Even though EGS can ramp technically, the economics push it toward continuous output. It behaves more like hydrothermal geothermal or nuclear. Only high utilization keeps the cost per MWh manageable. Using EGS as a flexible generator alongside renewables is not a viable economic strategy.

The combination of slow learning, high initial cost, limited repetition and inflexible economics creates a narrow role for EGS. It is not a mainstream firm power solution in decarbonization scenarios. It is more similar to large hydro or nuclear in that it requires long range planning, high capital investment and stable operation. It does not have the cost profile needed for broad deployment. Where special conditions exist, such as remote communities with poor solar and wind resources, or where geothermal gradients are exceptional, EGS might have a justified role. These would be small niches. The only realistic guidance is that EGS may be useful if a grid needs a very small fraction of firm power at almost any cost. That is not a common requirement in most power systems.

The broader system solutions for firm power remain cheaper. Storage, transmission expansion, overbuilt renewables, hydropower and a limited amount of flexible thermal generation can meet most needs at much lower cost. EGS may contribute in limited circumstances but it is not a silver bullet. When we examine the characteristics of learning curves and the underlying technologies of EGS, the conclusion is clear. EGS will improve but the cost gap with mainstream generation will remain large. It will remain a niche technology rather than a central pillar of future grids.