Making Deep Heat Bankable

Abstract

Deep heat has an economic asymmetry. A project spends heavily before the subsurface proves temperature, flow, impedance, seismic acceptability, and decline. Once a productive field exists, it can provide firm service with modest fuel exposure. I develop a financing architecture that separates exploration, well-field creation, surface conversion, and operation into evidence gates. Cost is represented as a probability distribution rather than a single levelized number. Drilling learning, standardized pads, modular plants, shared data, heat offtake, power contracts, insurance, and portfolio development reduce different risks. Public support is justified where early projects create transferable evidence, but it should require transparent data and competent execution. A bankable institution does not hide geology. It allocates each uncertainty to the party with the best capacity and incentive to manage it.

Capital arrives before certainty

Surface solar and wind resources can be measured extensively before major construction. A geothermal developer can survey a site, yet a deep well remains the decisive observation. Drilling may reveal lower temperature, poor permeability, difficult rock, unexpected fluids, or unacceptable seismic connection.

This front-loaded uncertainty raises the cost of capital and encourages small developers to risk survival on one campaign. Project Lithos needs an evidence structure that allows capital to enter in stages while preserving the integrated mission.

I divide development into regional exploration, site characterization, confirmation well, well-field demonstration, commercial expansion, and operation. Each stage produces evidence with defined acceptance criteria. The next capital tranche funds a narrower uncertainty set.

Model distributions, not one cost

Levelized cost can be written schematically as

$$\mathrm{LCOE}=\frac{\sum_t \frac{I_t+O_t+F_t+D_t}{(1+r)^t}}{\sum_t \frac{E_t}{(1+r)^t}},$$

where investment $I$, operations $O$, fuel or consumables $F$, decommissioning $D$, and delivered energy $E$ are discounted at rate $r$. For Project Lithos, many terms are uncertain and correlated.

A difficult well raises drilling cost and may also reduce productivity, delay revenue, require redesign, and increase workover risk. A high-impedance reservoir raises pump cost and can reduce net output. A seismic limit can strand part of the planned volume. One expected-value LCOE hides the probability of project failure and the capital required to survive it.

The economic model samples well duration, success probability, temperature, flow, decline, drilling learning, plant performance, outages, water, treatment, seismic curtailment, workovers, interest during construction, customer demand, and closure. It reports probability of completion, cost distribution, downside liquidity, and which observation changes the result most.

Drilling dominates the learning opportunity

DOE reports that drilling can comprise half or more of geothermal project capital. Cost per meter is useful, but total well cost also follows nonproductive time, trips, casing, cement, logging, testing, failures, and mobilization.

Learning curves can be represented as

$$C_n=C_1n^{-b},$$

where $C_n$ is unit cost after cumulative production $n$ and $b$ reflects learning. This relation is not guaranteed. Geology changes by site, and a company can repeat mistakes. Standard pads, common designs, automated handling, better bits, stable crews, open post-well review, and predictable procurement turn repetition into learning.

The institution records normalized drilling performance by formation and operation. It distinguishes geological luck from process improvement. Failed tools and invisible downtime remain in the dataset.

Portfolio development can spread geological risk. Several prospects share engineering, procurement, staff, data, and equipment. A successful site does not erase failed-site cost, so portfolio economics should report both.

Separate subsurface from surface commitment

Permanent surface plant should follow confirmation of fluid and reservoir conditions. Temporary test equipment serves early circulation. Once temperature, flow, chemistry, impedance, and seismic envelope are credible, modular conversion trains can be ordered around measured inputs.

This sequence reduces the chance of stranding a turbine or heat network. It can lengthen the apparent schedule, so stage planning and early customer agreements must reflect it.

Modularity supports expansion. One well pair and one conversion train establish operation. Later wells share roads, grid connection, control, workshops, monitoring, treatment, and trained staff. Expansion proceeds when observed thermal decline and demand justify it.

ExploreBuy information cheaply ConfirmTest temperature and rock directly DemonstrateProve bounded circulation ScaleReplicate wells and service trains

Revenue should value what geothermal supplies

Geothermal can provide firm power, capacity, grid services, industrial heat, district heat, and resilience. A flat energy price may underpay availability or flexibility. A contract should compensate the verified service without encouraging reservoir abuse.

Power purchase agreements can specify energy, capacity, availability, escalation, curtailment, interconnection, and performance. Heat contracts add supply and return temperature, flow, backup, network losses, and customer credit. Long-term offtake supports debt but can punish a project if early geological uncertainty is translated into rigid guarantees.

I favor staged contracts: demonstration volumes first, then expansion options after evidence gates. Thermal storage and multiple wells can improve reliability. Backup supply can cover customer criticality while the reservoir ramps conservatively.

Carbon value, clean-energy credits, and public procurement may add revenue. The project should remain understandable without assuming one temporary subsidy persists for thirty years.

Allocate risk to management capacity

Risk	Best-positioned manager	Possible instrument
Drilling execution	Developer and drilling contractor	Performance contract with balanced incentives
Pre-drill resource uncertainty	Developer, portfolio, public exploration program	Cost share, insurance, portfolio capital
Equipment performance	Supplier and operator	Warranty, availability guarantee, service agreement
Market price	Customer, utility, developer, hedge provider	Long-term offtake, floor, contract for difference
Seismic and environmental control	Operator under public authority	Permit limits, assurance, insurance, stop rights
First-of-kind transferable learning	Public and private sector	Milestone support tied to open data

No contract can transfer physical responsibility. A drilling contractor may carry schedule incentive, but the operator retains well-control governance. A community should not carry an uncapped seismic downside because investors prefer a low insurance premium.

Public funding is most defensible when it purchases evidence that later projects can use: open drilling data, high-temperature qualification, reservoir tests, seismic methods, and standardized interfaces. Support should be conditional on data quality and disclosure, including failures.

EvidenceReduce what can be known

Characterization, confirmation wells, tests, open data, and independent review.

ControlManage what remains uncertain

Modularity, limits, redundancy, intervention, insurance, and reserves.

ContractPay for delivered value

Firm power, useful heat, availability, flexibility, and transparent conditions.

LearnSpread gains across the portfolio

Common designs, crews, supply chains, models, and failure evidence.

Price closure and failure from the start

Financial models include failed wells, sidetracks, suspension, remediation, damage claims, decommissioning, and post-closure monitoring. These are not remote footnotes. They determine whether the institution can act responsibly after a disappointing result.

Contingency should follow quantified risk rather than a flat percentage applied for appearance. Management reserve is controlled separately from contractor budget. Liquidity planning asks whether the project can reach a safe stopped state if the next financing round never arrives.

Financial assurance protects the public from orphaned wells and environmental liabilities. Its amount updates with installed assets, condition, and closure estimate. Release follows verified work, not merely elapsed time.

Build a market around institutions

The first site will be expensive because it creates methods, data, approvals, workforce, supply, and trust. Later sites should reuse them. Standard well families, shared regional subsurface data, prequalified tools, modular conversion, common heat interfaces, and trained operating organizations reduce cost without pretending geology is standardized.

Project Lithos should publish a post-well cost and performance account with normalized definitions. Investors need credible evidence. Engineers need failed-run detail. Communities need environmental and seismic performance. Suppliers need a visible future market before investing in high-temperature products.

An economically successful deep-crust institution is not the project with the most aggressive spreadsheet. It is the one that survives geological disappointment, pays for safe closure, delivers a valued service, and makes the next responsible project easier to finance.