Energy · Deep dive

Quaise Energy

Superhot-rock geothermal via millimeter-wave drilling — an MIT fusion-lab spinout using gyrotrons to vaporize basement rock, aiming to make 300-500°C geothermal a baseload power source almost anywhere on Earth.

emerging

The question that decides it: Quaise's premise is that a fusion-grade gyrotron firing millimeter waves down an argon-filled waveguide can ablate basement rock to 5-20 km depths where drill bits and electronics fail. Does that system survive contact with real geology at commercial depth and cost — plasma breakdown in the waveguide, vaporized-rock ash recondensing and choking the bore, completing and casing an open hole at 400°C+, and a penetration rate reported near one meter per hour — or does the 2-year slip already visible (steam by 2026 and a 100 MW plant by 2028 became 50 MW and first electrons by 2030) stretch into the fate of every deep-drilling moonshot: technically real, commercially beaten by boring rotary rigs drilling shallower, cooler rock?

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Quaise Energy is an MIT Plasma Science and Fusion Center spinout betting that the way to make geothermal a global baseload power source is not better drill bits but no drill bits: a gyrotron — the megawatt-class microwave source built for fusion experiments — firing millimeter waves down a borehole to vaporize granite at depths where mechanical drilling dies. Founded in 2018, it has raised $230 million through a $134 million Series B first close (July 2026) led by Prelude Ventures with Japanese energy majors JERA and Idemitsu, and is building Project Obsidian in Oregon — pitched as the first commercial superhot geothermal plant, first power by 2030. In 2025 it drilled a record 100+ meters through Texas granite — the first time the technology left the lab — and is approaching one kilometer as of July 2026. If rock at 300-500°C is reachable almost anywhere, geothermal stops being a Ring-of-Fire niche and starts rivaling nuclear in power density. That is still a very large if.

Founding story

The technology predates the company by a decade. In 2008, Paul Woskov, a senior research engineer at MIT’s Plasma Science and Fusion Center, began pointing the lab’s gyrotrons — vacuum-tube millimeter-wave sources built to heat fusion plasmas — at blocks of granite and basalt, and found they ablated rock efficiently. For years it stayed a curiosity measured in centimeters.

Carlos Araque was built to find it. A Medellín-born mechanical engineer with MIT bachelor’s and master’s degrees, he spent 15 years at Schlumberger developing downhole tools — the equipment that fails when wells get too hot. In 2017 he became Technical Director at The Engine, MIT’s tough-tech venture fund, screening lab research for commercial potential. He knew from Schlumberger exactly why drilling hits a wall around 10 km and 200°C+ — bits wear, electronics cook, trip times eat the economics — and Woskov’s method had no downhole moving parts. He crossed from investor to founder in 2018. Matt Houde supplied the geothermal-industry half: Wisconsin geological engineering, a Stanford MS, stints at Ormat and EGS pioneer AltaRock Energy, where he helped win the 2019 ARPA-E grant that scaled Woskov’s experiments at Oak Ridge — then managed that $5M grant inside Quaise. One founder knows why wells fail; the other knows how geothermal projects get permitted and funded. Neither, skeptics note, changes the physics.

How it works

The gyrotron stays on the surface — that is the entire trick. A conventional rig drills the first 2-3 km of sedimentary cover, where rotary bits are unbeatable. At basement rock, Quaise swaps the drill string for a metallic waveguide carrying a megawatt-class millimeter-wave beam (roughly 100-300 GHz) from the surface gyrotron to the rock face. The beam heats rock dielectrically — the microwave-oven principle at vastly higher power density — until it ablates into fine, volcanic-style ash. The waveguide is filled with argon, which stays transparent to millimeter waves at downhole pressures; a purge gas lifts the ash out. The melted wall cools into a vitrified glass lining the company hopes acts as built-in casing. In its 2024 Houston demo, a 100 kW gyrotron burned granite at roughly 2 cm per minute; the 2025 Texas campaign used the same machine class, with a 10x-power system next.

With nothing complicated downhole, there is nothing to cook — no bits, no heat-limited electronics, no trip time — which makes 5-20 km and 500°C conceivable. At those depths water turns supercritical — carrying 5-10x the energy of conventional geothermal fluid — so one superhot well can in theory do the work of five to ten ordinary ones, and its steam matches the turbine inlet conditions of existing coal and gas plants, hence the repowering pitch. The catches are equally concrete: the beam can accidentally ionize gas into plasma, wasting energy and damaging the waveguide; vaporized rock must not recondense and choke the bore; and a reported penetration rate near a meter per hour at 1 MW implies months of beaming per deep well.

Product and business overview

Quaise calls itself both technology innovator and project developer — three components. First, the drilling system itself — gyrotron, waveguide, surface handling — developed with Nabors-supplied rigs at a Central Texas granite quarry, past 100 meters in July 2025 and approaching 1 km in July 2026. Second, Project Obsidian: a company-developed superhot plant on federal geothermal leases near Newberry Volcano in Oregon’s Deschutes National Forest — construction underway in 2026, ~50 MW initially, expandable toward 250 MW, first electrons promised by 2030. Third, repowering: a December 2024 agreement with Nevada Gold Mines (the Barrick-Newmont JV) to study retrofitting the 242 MW TS Power Plant in Eureka, Nevada with deep geothermal heat — the first pilot of its steam-for-fuel swap thesis. Note what Obsidian quietly concedes: Newberry is one of America’s most-studied geothermal sites, with superhot rock unusually shallow. The first plant does not test the drill-anywhere claim; it tests the machine where geology is friendliest.

Business model and pricing

Revenue is zero today; the model is capital structure. The $134M Series B (July 2026) is corporate equity; Obsidian is financed separately with project-level equity and debt (BofA and Goldman advising), and undisclosed offtake partners are slated to provide first revenues — presumably PPAs in the Fervo-Google and Sage/XGS-Meta pattern. The pricing story is the LCOE curve: Quaise projects below $40/MWh even for conservative supercritical systems at 10-20 km, and $20-30/MWh where superhot rock sits shallower (company analyses, 2024-26); the Clean Air Task Force’s 2022-23 technoeconomic work similarly modeled mature superhot rock at $20-35/MWh against a ~$40/MWh US average wholesale price. Every one of those numbers is a model, not a bill. Fervo signs real PPAs today; Quaise must undercut conventional rigs’ cost per meter at merely-hot depths — argon, gyrotron power, and months of beam time per well included.

Traction over time

The depth curve is real — five orders of magnitude in six years, and the 2025 Texas run was the first field penetration of basement rock by millimeter waves. It is also the whole story: no revenue series, no customers, no steam yet. Headcount sits in LinkedIn’s 51-200 band (2026). The strategic-investor progression is its own signal — Nabors (drilling), Mitsubishi (2024), then JERA and Idemitsu (2026), buying optionality on a technology that would matter enormously to resource-poor Japan.

Market analysis

Geothermal today is roughly 17-18 GW installed globally (2026), about 0.3% of world electricity — trapped where hot rock meets the surface. The prize is categorically different: DOE analyses see 90+ GW of US enhanced-geothermal potential by 2050, the IEA (2024) projected geothermal could meet up to 15% of global demand growth to 2050, and the Clean Air Task Force’s superhot-rock mapping argues 300°C+ rock under most continents makes geothermal terawatt-scale. The structural forces are strong: next-gen geothermal investment hit $1.7 billion in Q1 2025 alone, up 85% (industry tallies), driven by AI-datacenter demand for clean firm power — Google (Fervo, 115 MW) and Meta (Sage and XGS, ~300 MW, 2026) already contract years ahead of delivery. The catch: that demand is being met today with existing tools. The superhot TAM only becomes Quaise’s if conventional-depth EGS proves insufficient — or dearer than gyrotron holes.

Competitive intel

The shape of the set: Fervo Energy ($1.4B+ equity and debt through early 2026; 500 MW Cape Station targeted online by 2028) leads the category precisely because it imported oil-and-gas horizontal drilling instead of inventing physics — it will sell power before Quaise finishes its first well field. Sage Geosystems ($97M in January 2026, co-led by incumbent Ormat; Meta PPA) and XGS Energy (Meta 150 MW PPA, 2026) crowd the mid-depth field. Eavor ($530M raised through mid-2025) owns the closed-loop, seismicity-safe niche, finishing a commercial plant in Germany. Ormat and the conventional industry keep compounding drill-bit economics — and pocket the fee for Quaise’s own top 3 km. Quaise’s angle is depth: rivals are confined to geographies where existing drilling works; superhot wells would deliver 5-10x the energy per hole almost anywhere. Its weakness is symmetrical: it alone has never operated at commercial depth, and each rival PPA converts the demand Quaise needs into someone else’s backlog.

History and evolution

• 2008 — Paul Woskov begins millimeter-wave rock ablation experiments at MIT’s Plasma Science and Fusion Center.
• 2018 — Araque, then Technical Director at The Engine, founds Quaise with Houde to commercialize the work.
• 2019 — ARPA-E grant (~$5M, via AltaRock) moves experiments to Oak Ridge National Laboratory.
• Jun 2020 — $6M seed led by The Engine, with Vinod Khosla and Collaborative Fund.
• Feb-Jun 2022 — Series A: $40M led by Safar Partners, extended to $52M by TechEnergy Ventures. Public roadmap: field demos 2024, first steam 2026, a repowered 100 MW fossil plant 2028.
• Dec 2023-Mar 2024 — TechCrunch reports a ~$25M raise in progress (Dec 2023); it closes as a $21M Series A1 (Mar 2024) led by Prelude and Safar — a bridge, not a step-up, in a brutal cleantech market. Mitsubishi joins.
• 2024 — Full-rig surface demo in Houston (~10 ft through granite/basalt at ~2 cm/min); Dec 2024, Nevada Gold Mines pilot agreement for the TS Power Plant retrofit.
• Jul 2025 — 100 m (ultimately 118 m) drilled through Texas quarry granite — a record for the method; public demos follow in August 2025.
• Apr 2026 — Company announces it is on track for the first superhot plant; a Stanford Geothermal Workshop analysis backs a 50 MW first phase — half the 2022 promise, two years later.
• Jul 7, 2026 — $134M Series B first close led by Prelude with JERA and Idemitsu; $230M total; Project Obsidian construction underway, first electrons promised by 2030.

What people say

The case for. The scientific and trade press (MIT News 2022, MIT Technology Review July 2025, ThinkGeoEnergy 2025-26) treat the Texas results as a genuine breakthrough — the first non-contact drilling of basement rock at field scale, after seventeen years of lab work. Investor behavior is the loudest praise: nearly all existing investors returned for the Series B (July 2026), and the strategics are not tourists — Nabors is a drilling contractor; JERA and Idemitsu are operators in a resource-poor, volcanically rich home market. Supporters also credit the pragmatic hybrid conventional-then-millimeter-wave architecture.

The complaints. Stanford geothermal program head Roland Horne, in MIT Technology Review (July 2025), delivered the canonical critique: burning holes in rock is impressive but not the whole of drilling — the system must survive the heat and pressure at the bottom of a real well, and 100 m is a fraction of commercial depth. Hacker News threads (2022-2026) recur on specific mechanisms: a former drilling geologist flagged the cost of the argon waveguide fill at depth; others note ~1 m/hour penetration at 1 MW implies very long well times; hole stability, aquifer isolation, and casing a vitrified bore remain publicly unanswered; one much-upvoted comment said skipping any near-term product for 10 km holes suggests the real product is the stock. The literature adds a precedent: Iceland’s IDDP-1 well (2009) reached superhot conditions and produced record steam so corrosive it destroyed the surface equipment — making supercritical fluids survivable is an unsolved materials problem independent of drilling. And the record shows slippage the company does not advertise: 2022’s roadmap said steam by 2026 and 100 MW by 2028; mid-2026 reality is no steam, a 50 MW first phase, and 2030 first electrons — at a site chosen where superhot rock is shallowest.

Outlook: the open question

Quaise works if, and only if, millimeter-wave cost per meter through basement rock beats conventional drilling at depth — proven first by the 1 km Texas well, then by an Obsidian production well flowing supercritical steam near the 2030 schedule. The bull case: ablation is demonstrated at field scale; no-downhole-hardware dissolves the constraint that has capped well depth for a century; $230M and two of Japan’s largest energy companies are behind it; and the prize is a redraw of the world energy map, with 5-10x per-well energy density and drop-in reuse of the existing thermal fleet. The bear case needs no new facts, only extrapolation: meter-per-hour penetration, argon logistics, plasma breakdown, ash recondensation and 500°C completion each compound cost; the timeline has already slipped two years and halved in initial scale; the first site is a geological best case; and every year of slippage lets Fervo, Sage, XGS and Eavor lock up the clean-firm PPAs with drilling that works today. Watch three dated tests: the 1 km Texas milestone (promised 2026), Obsidian’s project equity and debt actually closing (promised later in 2026), and named offtake partners. Each lands, or the question starts answering itself.

How a challenger would attack it

The wedge. Quaise’s exposed flank is time. Its own roadmap has slipped two years and halved in scale — steam-by-2026 became first-electrons-by-2030 at 50 MW — and every quarter of slippage converts its addressable demand into rivals’ backlogs: Google’s 115 MW is with Fervo, Meta’s ~300 MW with Sage and XGS. A challenger doesn’t out-invent the gyrotron; it out-executes on the buyer. The play is Fervo’s, refined: proven rotary drilling, hyperscaler PPAs signed years ahead of delivery, and a deliberate march down the learning curve toward hotter, deeper wells — capturing the market Quaise needs while conventional-depth EGS keeps proving “sufficient.” A second, nastier vector: attack the unsolved back half of Quaise’s system. The IDDP-1 precedent shows supercritical steam destroys surface equipment; whoever solves 400°C+ completions, corrosion-resistant materials, and casing — sellable to every deep-geothermal player — owns the choke point Quaise must pass through anyway, without betting on millimeter-wave penetration rates near one meter per hour. Third: the first 3 km of every Quaise well is drilled by incumbents like Nabors, who see the economics firsthand and can option the technology’s success while keeping the revenue either way.

Same playbook, new buyer

Quaise’s playbook — repurpose fusion-grade gyrotron hardware into a no-downhole-parts drilling system — has applications that don’t require winning the US power market against Fervo. The most promising shift is geography and offtake: Japan. JERA and Idemitsu didn’t invest as tourists — resource-poor, volcanically rich, import-dependent Japan is arguably a better first market than Oregon, with superhot rock, national energy-security motivation, and no Fervo-equivalent domestic competitor; a licensed or JV deployment there monetizes the strategics’ appetite Quaise itself can only partially serve while proving Obsidian. Second: sell drilling-as-a-service for non-power buyers before the power thesis matures — the Nevada Gold Mines TS Power Plant retrofit already sketches this, and industrial heat, mining decarbonization, and repowering stranded coal assets pay for steam, not electrons, at smaller scale and earlier dates. The incumbent EGS players can’t follow into superhot repowering because their conventional-depth wells don’t produce turbine-grade steam for existing coal and gas plants — that inlet-condition match is uniquely Quaise’s, and it works at every stranded thermal plant on Earth if the drill does.

Sources and further reading

• Quaise Energy Raises $134 Million in First Close of Series B (Quaise Energy / BusinessWire, July 7, 2026)
• This startup wants to use beams of energy to drill geothermal wells (MIT Technology Review, July 22, 2025 — includes Roland Horne’s skepticism)
• Quaise Energy Achieves Drilling Milestone with Millimeter Wave Technology (ThinkGeoEnergy, July 2025)
• Tapping into the million-year energy source below our feet (MIT News, June 2022 — original 2024/2026/2028 roadmap)
• Quaise Energy demos millimeter-wave drilling rig in Houston (New Atlas, 2024)
• Quaise Energy Expands Series A to $52M (BusinessWire, June 8, 2022)
• Geothermal startup Quaise is raising $25M as it gears up for drilling (TechCrunch, December 12, 2023)
• Quaise Energy and Nevada Gold Mines Partner on Deep Geothermal Pilot (BusinessWire, December 3, 2024)
• Quaise looks to advance superhot geothermal power plant in Oregon (Canary Media, 2026)
• A Preliminary Techno-Economic Model of Superhot Rock Energy (Clean Air Task Force, 2023)
• Millimeter wave technology drills 100 meters into granite — discussion (Hacker News, 2026 — technical skepticism thread)