Geothermal energy extracts heat from the Earth’s crust to generate electricity or supply heat, and the live question in the supplied sources is whether “advanced” versions that engineer their own permeability can be built cheaply enough, and in enough places, to matter at grid scale.
Summary
Conventional geothermal power works where nature has already assembled three things in one place: hot rock at drillable depth, natural fractures or pore space that let fluid move through it, and water to carry the heat to surface. Wells are drilled into that reservoir, hot fluid is produced, its energy is converted to electricity through a steam or binary cycle, and the cooled fluid is reinjected. Because the required combination of heat and natural permeability is geologically rare, conventional geothermal has historically been a regional resource rather than a general one.
The frontier that the sources point at is “advanced” or “deep” geothermal: going after hot rock that lacks natural permeability, and creating the flow path by engineering it, so that the resource becomes a function of drilling capability and heat gradient rather than of local volcanic geology. A think tank assessment published in May 2026 frames the technology as widely available and clean, with cost the explicitly hedged variable ref. That hedge is the whole investment case in one word.
The parameters that decide it are: how temperature rises with depth at a given site, how much drilling costs as depth increases, how much fluid can be circulated through the engineered rock volume and for how long before the flow path degrades or short-circuits, and how accurately the subsurface can be predicted before capital is committed. The research awards in this source set attack precisely that last parameter. One project is building a three-dimensional predictive model of how folding and faulting fractures sandstone and thereby controls fluid flow, explicitly aimed at geothermal potential, petroleum production and groundwater. Another extends an open-source multiphysics simulation framework built for fission and fusion systems into geothermal applications. A third funds microstructural characterisation instrumentation that includes geothermal systems among its target uses.
The reader should know that the supplied evidence base is thin and indirect. There are no wells, no flow rates, no cost figures, no capacity numbers and no project outcomes in these sources. Most of what follows is therefore scored null rather than guessed at.
Viability (unscored)
Nothing in this source set reports a drilled well, a flow rate, a capacity factor, a levelised cost, or a completed project. The one 2026 assessment that speaks to viability does so in a headline that is itself conditional: advanced geothermal is described as widely available and clean and “maybe cheap enough” to make a big impact ref. A “maybe” on the decisive variable is not evidence of viability either way.
What the sources do show is that key physical uncertainties are still being reduced by publicly funded science in mid-2026: the geometry and intensity of deformation-induced fracturing that controls subsurface fluid flow is still the subject of new predictive-model development, and simulation capability for geothermal is being added to a framework built for other energy systems rather than already existing in mature form. That is consistent with either a technology on the verge of engineering scale-up or one still resolving fundamentals. The sources do not let a reader distinguish between those two states, so this dimension is left null.
TLDR: The sources contain no performance, cost or project data, so no viability judgement is earned.
Drivers (2/5)
Supply side: geothermal recurs as a named beneficiary across three separate NSF awards dated within a month of each other in mid-2026, covering subsurface fracture and fluid-flow prediction at $210,670, open-source multiphysics simulation at $1,500,000, and geoscience characterisation instrumentation at $234,089. One award explicitly frames workforce development for geothermal exploration and production careers, and the news channel carrying the 2026 policy assessment is an ARPA-E SCALE-UP and OPEN awards feed ref, indicating agency-level programme attention. These are real but modest signals, and in two of the three awards geothermal is a secondary application rather than the purpose.
Demand side: the sources are silent. There is no offtake, no utility procurement, no industrial heat customer, no siting or interconnection data, and no price signal. A score of 2 reflects visible, sustained but low-magnitude supply-side push with the demand half of the picture entirely unevidenced here.
TLDR: Small but repeated public research funding on the supply side in 2026; the sources say nothing at all about demand.
Novelty (3/5)
The comparison the sources set up is advanced geothermal against conventional geothermal, and the axis is availability. A 2026 policy assessment characterises advanced geothermal as widely available and clean ref, which is meaningful only against the baseline that conventional geothermal is not widely available. If that holds, the step change is in addressable geography rather than in the conversion technology, which is conventional thermal power engineering. The enabling novelty implied by the research awards is predictive rather than mechanical: knowing in advance how faults and folds have fractured rock and therefore where fluid will flow, and having validated multiphysics tools to model heat transfer and flow in these systems.
What the sources do not provide is any magnitude. There is no figure for how much of the land area becomes viable, no cost comparison against solar, wind plus storage, gas or nuclear, and no efficiency or output comparison against conventional geothermal fields. The same 2026 source that asserts wide availability withholds judgement on cost. A 3 records a credible and specific novelty claim that remains contested and unquantified in this evidence base; it is not a 4 because no demonstrated result is supplied.
TLDR: The claimed advance is geographic: heat available almost anywhere rather than only where nature supplied permeability, with the cost premium unquantified.
Diffusion (unscored)
Adoption is where this source set is emptiest. There is nothing on drilling rig availability or day rates, nothing on permitting or subsurface rights, nothing on induced seismicity acceptance, nothing on interconnection queues, and no named developers or projects. The only adoption-adjacent signal is a workforce one: an award that expects participants to gain skills applicable to careers in geothermal exploration and production, which implies an anticipated hiring market but proves nothing about it.
One structural point can be made without going beyond the sources. The skills, models and instrumentation being funded here are shared with petroleum production and groundwater work and with fission and fusion simulation, which suggests borrowed rather than bespoke supply chains. That usually helps diffusion, but the sources do not measure it, so no score is given.
TLDR: No deployment, permitting, drilling supply chain or grid-integration evidence in the sources.
Impact (3/5)
The upside case is straightforward and is the one the sources gesture at: a clean generation resource that is not confined to particular geology would be a general-purpose addition to power systems rather than a regional niche ref. The same source ties the size of the impact explicitly to cost, so the value is stated as contingent, not demonstrated. This publication’s own earlier horizon scanning listed deep geothermal among its three most underrated technologies in 2023, contrasted against fusion which it called overrated ref ref; that is an opinion from the same house, and is treated as such rather than as independent corroboration.
No source here supplies a market size, a deployable capacity, an emissions displacement figure, or a system-value estimate. A 3 records that two sources independently frame the potential as substantial while neither measures it. Anything higher would be scoring an assertion.
TLDR: Asserted as potentially large and clean by a 2026 policy assessment, but conditional on cost and unquantified in every source here.
Timing Unclear
The only dated assessment in the set, from May 2026, leaves the cost question unresolved ref. In the same window, funding is going into building three-dimensional predictive models of how rock deformation controls fluid flow and into extending simulation capability towards geothermal for the first time within a given framework. Research awards of this type inform projects on multi-year lags, so they are a marker of an unfinished toolchain rather than of imminent deployment.
At the same time, nothing here rules out near-term commercial activity, because the sources simply do not cover projects, wells or offtake. Assigning any band from Now to Later would be inference rather than assessment, so this is recorded as Unclear. The absence of project-level evidence in a source set drawn from 2026 news and awards is itself a mild negative signal about how visible commercial-scale activity currently is.
TLDR: As of mid-2026 the sources still treat cost as an open question and are funding fundamental subsurface prediction work.
Overrated or underrated? Too early to say
The direction of travel in these sources is mildly positive and nothing here contradicts it: a 2026 policy assessment calls advanced geothermal widely available and clean ref, public science funding keeps naming geothermal as a target application across geology, instrumentation and simulation, and this publication has been long the sector since 2023 ref. But every one of those is either an opinion or an input, not an output.
The decisive number, cost per unit of delivered energy in rock that had no natural permeability, is absent from this source set and is hedged in the only source that raises it. Until that lands, calling geothermal underrated or overrated would be borrowing conviction from outside the evidence. The honest position is that this concept is under-evidenced here rather than genuinely uncertain in the world, and the page should be revisited as soon as project-level cost or flow data appears.
Prediction
By 31 December 2028, the Cardinal open-source multiphysics ecosystem funded under NSF award 2550125 will have published at least one publicly documented geothermal tutorial, validation case or application, as promised in its July 2026 Phase II scope.
Evidence base
- May 2026: a think tank assessment characterises advanced geothermal as widely available and clean, but frames cost conditionally as “maybe cheap enough” to make a big impact ref.
- 17 June 2026: NSF awards $210,670 to build a predictive 3D model of how progressive folding fractures sandstone and controls fluid flow, explicitly to improve understanding of geothermal energy potential, and to train participants for geothermal exploration and production careers.
- 2 July 2026: NSF awards $1,500,000 to extend the Cardinal open-source multiphysics framework, built for fission and fusion systems, into geothermal energy among other areas, with verification and validation suites.
- 1 July 2026: NSF awards $234,089 for EBSD and EDS detectors whose enabled research explicitly includes studies of geothermal systems.
- May 2023: this publication listed deep geothermal among its three most underrated technologies ref, while listing nuclear fusion among the three most overrated the following week ref.
- June 2023: continued solicitation of contacts working in deep geothermal in the publication’s mailbag, indicating a sector with limited visible activity at that time ref.
Open questions
- What is the actual drilled and completed cost per well at the depths advanced geothermal requires, and how does it scale with depth and rock type?
- Can engineered flow paths sustain commercial circulation rates over a project lifetime without thermal short-circuiting, and is there any long-duration field data to show it?
- Do the predictive fracture models being funded in 2026 actually reduce subsurface risk enough to change project financing terms, or do they remain academic tools?
- Who is the buyer? The sources contain no demand-side evidence at all, so is there procurement, industrial heat or data centre offtake behind the enthusiasm?
Assessment drafted 2026-08-31 from up to 10 KB sources using the technology-scorecard framework; scores are a draft read pending review.