ZEMANIUMD3 · The Business
The business case · for capital and commercial partners

The business

One audited model, five rungs, one named invention per rung. The commercial band is $30–45/MWh; today's best coupled result is $61.65. This is the arithmetic of the gap — and the company that closes it rung by rung.

1 · The economics

The ladder below is five project definitions inside one recomputed, audited cost model — not additive savings stacked on one plant. Model rules are held constant across every row: 25-year life, 90% availability, 8% real discount, 5% gross auxiliaries, 2%/yr fixed O&M of capital, $8/MWh variable. The arithmetic was independently recomputed from the stored physical trajectories and published as a QA record model. Geology and costs remain unvalidated — that is what field runs are for — but no number here is hopeful. Each row is a requirement with a bench gate, and each transition names the invention that must deliver it.

LCOE spec ladder 61.65 to 29.80 dollars per MWhFive audited model rows from 61.65 to 29.80 dollars per MWh with the 30-45 band and the invention that delivers each rung.The spec ladder — $61.65 to $29.80 per MWhone named invention per rung$30–45/MWh commercial band$20$40$60$61.65Baseline240 °C · $37m subsurface · $1,800/kWtoday's coupled study$55.05Subsurface cost target240 °C · $25m subsurfaceE1 Shift-Latch$47.46Hotter resource300 °C · $25m subsurfaceE4 Ember-BHCT + E5 Flex-Seal + E2 Clear$34.06Conversion cost300 °C · $25m · $1,000/kW blockE3 Conversion$29.80Full stack320 °C · $15m · $1,000/kW · ~46.3 MW netall five + Relay controls$0[MODEL] Five different project definitions in one audited model — not additive savings in one plant (non-additivity warning stands). Model rules held: 25 yr · 90% availability · 8% real discount · 5% gross auxiliaries ·2%/yr fixed O&M of capital · $8/MWh variable. Arithmetic verified — qa/thirty-mwh-audit-2026-09-14.json. Geology and costs unvalidated; nothing here is measured plant performance.
Spec ladder: $61.65 → $29.80/MWh. Row 4 ($34.06) is already inside the commercial band at 300 °C if conversion cost lands. Row 5 is the $30 mission model.

Read the rungs as a work order. Row 2 is E1 Shift-Latch — commandable reservoir reconfiguration puts more of each well pair's rock to work and takes subsurface cost from $37m to $25m per pair. Row 3 is access and survival at 300 °C: E4 Ember-BHCT drilling, E5 Flex-Seal integrity through cycling, E2 Clear-Chemistry keeping the hot branch conductive. Row 4 is E3 Conversion — an installed power block at or below $1,000/gross-kW. Row 5 is all five, with Relay controls holding the margins at 320 °C.

Zemanium · Earth Power SystemsD3 · page 1/4

2 · Margin discipline — designing to $27 to hold $30

The sub-$30 row is thin at nominal: it survives 0.90% of lifetime-electricity shortfall or 0.91% of capital overrun — about $0.68m of total capital headroom. So we do not sell the nominal row. The engineering target is $27/MWh nominal, which holds $30 under a combined −5% electricity and +10% capital stress. At 320 °C and $15m subsurface that means an installed surface allowance near $831/gross-kW. We treat that number as a design contract, not a hope.

Sensitivity of the 29.80 row to six stress casesDeviation chart of six stress cases around 29.80 dollars per MWh with the 30 dollar mission line and 27 dollar design-to target.Margin discipline — designing to $27 to hold $30Nominal spec row$29.80Lifetime electricity −1%$30.02Variable O&M +$1/MWh$30.80Lifetime electricity −5%$30.95Capital + linked O&M +10%$31.98Electricity −5% and capital +10%$33.24$30 mission$27 design-to targetHeadroom at nominal: 0.90% lifetime electricity · 0.91% capital (~$0.68m).Holding $30 under −5% electricity and +10% capital needs the nominal design at $27 (≈ $831/gross-kW installed surface at 320 °C / $15m subsurface).[MODEL] Illustrative engineering allowances, not measured uncertainty distributions. Same model rules as the ladder — qa/thirty-mwh-audit-2026-09-14.json (arithmetic verified).
Six stress cases on the $29.80 row — illustrative engineering allowances, not measured uncertainty model.

And the stress is now quantified, not asserted: 20,000 cases across a declared uncertainty box — capital, subsurface, electricity, financing — under pre-registered falsifiers simulated. The band is robust: the $34.06 row holds ≤$45/MWh in 95.4% of the box, the $29.80 row in 100%. The point is honest work: the "design to $27" cushion holds $30 in 42.5% of the box, and the design that holds $30 in 80% of it is ≈$684/gross-kW. Two falsifiers breached; both became published redesign levers — and the top lever turned out to be the discount rate, which is why de-risked long-tenor capital sits inside the engineering plan (qa/ladder-stress-2026-09-30.json).

The size of the required invention is also audited, and it is honest about what optimization alone cannot do. At the coupled target's electricity, discount and operating rules, $30/MWh needs total capital of $40.68m where today's design carries $99.21m — a 59% reduction. Make the subsurface work free and the model still prints $41.64/MWh. Eliminate all circulation pumping in a thought experiment and it prints $53.45/MWh. With today's $1,800/gross-kW surface cost the arithmetic floor is $34.48/MWh even with free wells and zero pumping. The gap therefore closes only through the named inventions — more electricity per well pair, cheaper access to hotter rock, and conversion equipment at half the installed cost. That is exactly the series we have designed and virtually qualified.

What would make a skeptic's objection

Nothing in this model is measured plant performance; the spec rows are requirements, and each one expires unless its bench gate closes. If a gate fails, the published failure re-prices the row — and the ladder re-computes. The model is a living contract between the physics and the plan.

Zemanium · Earth Power SystemsD3 · page 2/4

3 · The market — a funded frontier that cannot see itself

Three demand signals point the same way. Capital arrived at the superhot frontier in 2026: Quaise $180M (with Nabors), Mazama $135M (Khosla), Hephae $17.8M Series A, Fervo's Cape Station first power — all verified press. Load arrived behind it: data-center electricity grew 17% in 2025 toward ~945 TWh/yr by 2030, and hyperscalers are contracting firm geothermal directly (Google–Fervo framework up to 3 GW through 2033; Meta–Sage up to 150 MW — frameworks and partnerships, cited as such). And the public target arrived: $45/MWh by 2035 (DOE Enhanced Geothermal Shot).

Every one of those programs drills wells whose transient and hot-section behavior nobody can currently measure. Demand for instrumentation and qualification scales with wells drilled regardless of which developer wins — the picks-and-shovels position in a category being dug by four well-funded teams. The layer those teams share is unfunded: $100M+ rounds price plants; the tools and standards layer is exactly where a defensible component-and-service business lives.

4 · The model — three lines, one ladder

LineWhat sellsWhen it turns on
InstrumentsCool-Brain profilers and the series — the E1 latch arrays, E5-rated well-integrity engineering, the Ember-BHCT control systemon the bench record (R3)
Witnessed qualificationduty-basis testing and witnessed records at 450 °C-class — the service every tool maker in the segment needs and nobody offerson the first record, Q1 2027
Data & basisthe qualification dataset and the standard itself; duty-cycle rating services (E5's product regardless of any patent posture)as records accumulate

Product lines in the field: Relay (heat routing and control — the opening move that holds auxiliaries down), Shift (reversible flow treatment — the reservoir repair and redirect), Flex (thermal movement — the moving pressure boundary). The gigascale ladder converts lines into manufacturing: G1 bench proof → G2 a three-well repeatable-service program ($55.05 row) → G3 the first 45–50 MW block at the band, built from factory power-block modules → G4 the fleet at $30, where 25-year operating records make the position undroppable.

Zemanium · Earth Power SystemsD3 · page 3/4

5 · The raise — $2–3M pre-seed against a dated schedule

RungSizeWhat it buys
Sprint (open now)$10–30kprototype materials, Tier-0/1 fixtures — terms memo ready
Pre-seed (this round)$2–3M
lead $1–1.5M
the R1–R3 numbers, the Rev B field string, design-partner wells, qualification-service stand-up
Non-dilutive (parallel)$300k–$4.5M classDOE UTR via a university PI, the next ARPA-E FOA (its QA-infrastructure category is our lane), EU SHiFT via the Iceland labs
Series A$10M+ classrepeatable service scaled: field pilots, the manufacturing line
Project finance$100M+ / blockfleet rungs at measured LCOE

Use of proceeds is named milestones, never runway: (1) chain proof → first hot data, (2) the 500-hour qualification record at 450 °C, (3) design-partner wells with the Shift latch and Cool-Brain strings, (4) first commercial subsystems, (5) the witnessed-qualification service in revenue. The schedule: R1 within 30 days, R2 within 90, R3 in Q1 — SGW week (Feb 8–10) is the coming-out venue for the record and the recruiting.

The honest status line, printed because diligence will find it anyway and it is better as our sentence than theirs: zero physical runs and zero meters today — design, software, digital qualification and audited economics are done, and that is precisely what this round converts into measurements. Bench number, publication and named attachments land inside the schedule above. Money conversations open on the demo, the audited economics and this results calendar — and every number in the room keeps its label.

The ask: lead $1–1.5M of a $2–3M pre-seed. First step is thirty minutes — we will bring the model, the gates and the falsifiers, and you can attack all three.
Zemanium · Earth Power SystemsD3 · page 4/4