A · Power
Generate it on-site
Enhanced geothermal now delivers 24/7 carbon-free baseload at costs competitive with the grid. Rhodium projects it could meet two-thirds of new data center demand by 2030 — at or below what operators pay today.
The AI buildout just hit a wall — and it isn't land, silicon, or capital. It's grid access. Khora develops data centers that generate their own 24/7 clean geothermal power and stack compute vertically — one integrated asset, sited where heat and demand finally meet.
We're not selling power and we're not renting racks. We're building the thing that fuses them: generation, structure, and cooling designed as a single asset from the ground down.
For a decade, siting a data center was a real-estate problem. In 2025 that broke. AI demand is now colliding with a grid that physically cannot connect it fast enough — and the binding constraint moved from acres to amps.
Projected growth in global data center power demand by 2030 vs. 2023, driven primarily by AI.
Goldman Sachs Research, 2025
Average wait in U.S. interconnection queues — now the single biggest constraint on new builds, ahead of land and capital.
Lawrence Berkeley National Lab
U.S. counties and cities that have passed data-center moratoria, zoning limits, or new restrictions since 2023.
Food & Water Watch, 2025
In Northern Virginia, Texas, and Ireland, utilities are now telling new campuses to secure their own power before construction can begin.
Sources: PJM filings · Ireland CRU conditional-connection framework · ERCOT large-load reports, 2025
Three trends are usually treated as separate bets. The alpha is in refusing to separate them.
Enhanced geothermal now delivers 24/7 carbon-free baseload at costs competitive with the grid. Rhodium projects it could meet two-thirds of new data center demand by 2030 — at or below what operators pay today.
When power is co-located and land is scarce, going up beats sprawling out. Modular compute cells inside a braced exoskeleton put several times more deployable load on a constrained parcel — without bespoke, un-financeable construction. Silicon swaps out without touching the structure.
Direct-to-chip liquid cooling is where 100kW+ AI racks have forced the industry. We run cooling and geothermal off one integrated fluid system — the well that powers the building anchors the loop that cools it.
Geothermal developers (Fervo, Sage, XGS) sell power and are creeping toward the customer. Data-center operators are signing power deals and creeping toward generation. Both are well-funded — and both are optimized to do one side well. Neither is built to design generation, structure, and cooling as a single asset from the ground down.
That integration is the company. It's hard to copy, hard to unbundle, and it turns a 10-year-disposable warehouse into a 50-year platform you swap silicon through. And it only works if the people building it have actually drilled the wells and built the data centers — which is exactly the founding team we're assembling.
Modeled on a constrained 5-acre urban parcel, holding IT density constant. Conventional 2-story warehouse vs. vertical build. Planning figures — not engineering specs — but the gap is structural, not marginal.
2 stories · 45% site coverage · the standard hyperscale shed, hard-capped by its own footprint.
Modular cells · 25% footprint, freeing the rest of the parcel for the well array and substation.
Internal model · 25% spire footprint vs. 45% warehouse; 65% usable area per floor (core + exoskeleton); 2.0 kW/m² IT density held constant. Conservative, order-of-magnitude planning estimates.
This is infrastructure, not a magic box. Here's what serious diligence will flag, and our honest read on each.
The first asset is a multi-hundred-million-dollar project. Comparable: Fervo took $421M in project finance for a single site.
This round doesn't build — it makes the first project financeable. We unbundle the well (infra debt) from the building (real estate) and originate the asset like developers, not like a venture bet.
The best gradients are out West where land is cheap; the land-pinched markets have poor near-surface heat.
We don't fight that — we site into the overlap that already exists: Texas and the Mountain West, where grid pressure and strong gradients genuinely coincide today. The map widens as drilling costs fall.
Geothermal developers want the customer; data-center operators want the power. Someone could try to fuse the two.
They're structurally optimized for one side. Our edge is a founding team that spans both disciplines and an integrated design that's our product from day one — not a bolt-on retrofit.
This domain rewards specialists who've drilled wells, operated plants, or built hyperscale data centers.
We're explicit about it: this round is gated on closing a specialist co-founder. The vision is necessary but not sufficient — execution credibility is what we're funding first.
What's better, what's a tradeoff, and what we'll quantify.
Geothermal is 24/7 carbon-free baseload — unlike solar or wind, it needs no fossil backup for when conditions change. Where Khora displaces the natural gas that's otherwise being built to power data centers, that's a real carbon reduction, on a fraction of the land a solar or wind farm would need for the same output.
Conventional data centers are water-intensive because many rely on evaporative cooling — literally evaporating millions of gallons to shed heat, the source of community water fights across the Southwest. Khora uses closed-loop, direct-to-chip liquid cooling, which circulates the same fluid instead of evaporating it away — dramatically lower ongoing water consumption, and a decisive advantage in the water-stressed Western regions where our siting makes sense.
A tall structural build is concrete- and steel-intensive — both carbon-heavy. Our answer is longevity: by designing the structure, power, and cooling to outlast many generations of silicon, that embodied carbon amortizes over 50 years instead of the ~10-year life of a disposable warehouse.
The servers draw what they draw — Khora makes that power clean, firm, and local, not smaller. We change where the energy comes from, not how much is used.
Enhanced geothermal uses directional drilling and stimulation techniques with water use and a documented, managed risk of induced (minor) seismicity. Milder than oil and gas, but not zero.
We use closed-loop geothermal — the low-water option that recirculates its working fluid — but it's still a water system a grid-powered facility wouldn't have. The cooling-water savings very plausibly outweigh it, especially in arid siting.
All-in water usage effectiveness (cooling + generation, in liters per MWh) and lifecycle carbon, benchmarked against an evaporatively-cooled, grid-powered comparable. We'd rather publish the real numbers than claim "green."
This round doesn't build a data center. It converts the vision into the one thing that unlocks project capital: a specific site, a validated design, and a buyer at the table.
Sign the specialist co-founder and senior engineering hires, then screen candidate sites where strong gradient and grid pressure genuinely coincide — Texas and the Mountain West. Ends with 2–3 ranked sites and a complete founding team.
Full subsurface assessment and an engineer-reviewed reference design for the integrated cell + exoskeleton + loop. Begin the unbundled capital structure and open offtake conversations. Ends with a costed, independently-reviewed feasibility study.
Convert the lead site to a signed option, secure a non-binding offtake LOI from a named compute buyer, and finalize the capital stack. The result is a bankable feasibility package that supports a $20–50M, project-framed Series A.
An infrastructure-development round, not a software round. It funds 18 months of senior specialist work and the feasibility package that unlocks a project-framed Series A. You're backing a founding team that spans subsurface, structures, and compute — and an integrated design that's the product from day one.
Founding team — the specialist co-founder and senior drilling, structural, and power-finance hires that make this credible.
Subsurface & validation — geological assessment of candidate sites and independent technical review.
Design, site & structuring — reference engineering, site origination toward LOIs, and the unbundled capital stack.