Aalo Atomics makes 50 MWe modular nuclear plants for AI data centers. Too big to be a microreactor, too small to be a traditional SMR. The pitch is behind-the-meter power sited next to data centers.
Founded late 2022 by Matt Loszak and nuclear engineer Yasir Arafat, who was chief designer of the DOE's MARVEL reactor at Idaho National Lab. The design borrows from that government work rather than starting clean-sheet.
| Round | Date | Amount | Lead | Key Participants |
|---|---|---|---|---|
| Seed + Series A | 2023-2024 | $33M | 50Y, Fine Structure | $6.26M seed + $27M Series A |
| Series B | Aug 19, 2025 | $100M | Valor Equity Partners | Hitachi Ventures, NRG Energy, Tishman Speyer, Fine Structure, Crosscut, Kindred, Harpoon, 50Y, MCJ, Nucleation Capital, VamosVentures, Gaingels, Crescent Enterprises, Alumni Ventures, Perpetual VC |
Valor Equity Partners (Lead) - Growth firm; portfolio includes Heron Power, Eight Sleep, and Defense Unicorns. Valor puts operators into its companies, not just money. Multi-billion AUM, exact figure undisclosed.
Hitachi Ventures - CVC arm of Hitachi Ltd., $1B across four funds. Hitachi has decades of nuclear engineering behind it, so this reads as a strategic bet, not just a financial one.
NRG Energy (NYSE: NRG) - One of the largest U.S. power companies. A utility on the cap table, not just a financial investor.
Tishman Speyer - One of the world's largest real estate developers. Their interest points to co-locating reactors with data center real estate.
Previously co-founded Humi, a Canadian HR/payroll SaaS company that raised $31M led by Kensington Capital Partners. Built the platform, scaled it, then moved into nuclear. Engineering and physics background.
Arafat was the chief designer and project lead for the DOE's MARVEL microreactor at Idaho National Lab - a 100 kW sodium-potassium cooled reactor that Aalo's commercial design is based on. Born in Bangladesh, trained in the U.S. Few people have actually designed and built a working advanced reactor for the federal government; he's one of them.
The CEO brings startup execution, the CTO brings the nuclear engineering. Most nuclear startups have one or the other, not both.
In April 2026, Aalo added Michael So as VP of Programs (Apr 27, 2026), previously head of Tesla's Cybercab launch, to run manufacturing and program execution.
Aalo coined its own reactor category, XMR (Extra-Modular Reactor), for something between a microreactor and an SMR. The commercial product is the Aalo Pod: five 10 MWe reactors bundled into a single 50 MWe plant.
| Spec | Detail |
|---|---|
| Output | 50 MWe (5 × 10 MWe Aalo-1 units) |
| Reactor Type | Sodium-cooled, thermal neutron spectrum |
| Fuel | Low-enriched uranium dioxide (LEU, ~5%) - NOT HALEU |
| Coolant | Liquid sodium |
| Manufacturing | Factory-built modules, assembled on-site |
| Water Required | None (air-cooled) |
| Safety | Passive shutdown, walk-away safe |
Aalo's main edge. Nearly every other advanced reactor startup (Kairos, Oklo, X-Energy, Radiant) needs HALEU - high-assay low-enriched uranium at 5-20% enrichment. Aalo runs on standard 5% LEU, the same grade conventional reactors use, with a supply chain that already exists.
HALEU is in short supply. The DOE is spending billions to build domestic production, but it won't be widely available until the late 2020s. Aalo doesn't have to wait for it.
Aalo-X is being built at INL, next to the DOE's MARVEL reactor. DOE granted a siting MOU in 2025 and Aalo broke ground in September 2025. INL gives it regulatory-pathway support, technical help, and government credibility.
Urenco - Aalo says this is the first commercial enriched-uranium delivery contract signed by any U.S. advanced nuclear startup. Fuel delivery scheduled Q1 2026 for the Aalo-X reactor.
Global Nuclear Fuel (GNF) - Contracted for fuel rod fabrication (March 2026). GNF is a GE Hitachi joint venture.
The supply chain is contracted, not hypothetical. Fuel, fabrication, turbines, and site access are all under contract.
| Milestone | Target Date | Status |
|---|---|---|
| Final Design Review (DOE + NRC) | Jan 22, 2026 | Complete |
| Aalo-X / Critical Test Reactor construction | Mar 2026 | Complete |
| Critical Test Reactor (CTR) unveiled at INL | Mar 19, 2026 | Complete |
| LEU feedstock secured (Urenco) | Apr 30, 2026 | Complete |
| DOE-Idaho approves Documented Safety Analysis | May 5, 2026 | Complete |
| Graphite loaded into reactor core | May 2026 | Complete |
| Aalo-X criticality | Jul 4, 2026 | Complete |
| NRC commercial licensing for Aalo-1 | 2026-2027 | Pending |
| First commercial deployment | 2028 | Projected |
Most advanced nuclear startups sit at one of two extremes: microreactors under 10 MW (Radiant, Oklo's Aurora) or SMRs above 100 MW (TerraPower's Natrium at 345 MW, X-Energy's Xe-100 at 80 MW). Aalo's 50 MW slot has little competition and lines up with hyperscale data center campus power needs.
| Advantage | Why It Matters |
|---|---|
| LEU fuel (not HALEU) | Sidesteps the industry's biggest supply chain bottleneck |
| 50 MW data center fit | Right-sized for hyperscale campus power blocks |
| MARVEL heritage | Built on proven DOE technology, not a clean-sheet design |
| Factory manufacturing | Mass production potential vs. custom one-off construction |
| No water required | Air-cooled design opens up site options other reactors can't touch |
Kairos Power is further along than any other advanced reactor company in the United States. They hold the only NRC construction permits ever issued for a non-light-water reactor, they're building at two sites, and they have a signed power purchase agreement with Google.
Founded in 2016 by three nuclear engineers from UC Berkeley and the Electric Power Research Institute, Kairos has built and tested hardware instead of publishing papers.
| Source | Date | Amount | Type | Details |
|---|---|---|---|---|
| Private Rounds | 2016-2025 | $100M+ (est.) | Equity | Limited public disclosure; institutional + strategic investors |
| DOE ARDP | Jan 2024 | $303M | Milestone-Based | Performance-based, fixed-price milestones over 7 years. Total project value: $629M ($303M DOE, $326M Kairos). |
| Samsung C&T | Sep 21, 2026 | up to $100M | Strategic Investment + In-Kind | Binding term sheet: direct investment plus in-kind engineering services, subject to regulatory approvals, earmarked for Hermes 2 delivery. Samsung C&T also joins Kairos's integrated EPC team for Hermes 2 and future KP-FHR plants - a step up from the advisory role it was exploring in the Aug NuCAMP MOU. Term sheet signed; the money hasn't closed yet. |
The DOE contract is milestone-based: Kairos only gets paid when it hits specific technical milestones. The $303M is government co-investment, with payment tied to proving the technology rather than handed over up front.
Several milestones have already been hit: NRC construction permits, ETU reactor vessel installations, and HALEU fuel supply agreements.
Kairos also holds a $27M partnership with Oak Ridge National Laboratory (ORNL) and anchors the KP-OMADA alliance (Google, TVA, and participating utilities) for fleet deployment.
Stanford-educated, co-founded Kairos in 2016. Leads business strategy and commercial deployment. Has steered the company through its DOE award, NRC construction permits, and Google partnership - three things no other advanced reactor CEO can claim.
PhD in nuclear engineering from UC Berkeley. Former project manager at EPRI (Electric Power Research Institute), then tenure-track professor at University of New Mexico. Deep expertise in thermal hydraulics and reactor design. Technical architect behind the KP-FHR.
Professor and a leading expert in fluoride salt-cooled reactor technology. The scientific basis for the KP-FHR program traces to his work.
Named to the role in early August 2026 (Teknovation.biz, Aug 3). Responsible for delivering Hermes 2 at Oak Ridge - construction through installation, commissioning, and startup - and for coordinating operations across Albuquerque, Oak Ridge, and future deployment sites. Previously senior director of launch and recovery operations at SpaceX; before that, 25 years in U.S. Army Special Operations Command. PhD in nuclear engineering from MIT, master's in management from Stanford. The hire reads as Kairos shifting from a company that designs and licenses reactors to one that has to actually build and run them.
The KP-FHR (Kairos Power Fluoride Salt-Cooled, High-Temperature Reactor) combines two proven technologies:
| Reactor | Type | Thermal | Electric | Purpose | Target |
|---|---|---|---|---|---|
| Hermes | Demo | 35 MWt | - | Validate systems, supply chain | 2027 |
| Hermes 2 | Commercial demo | 140 MWt | 50 MWe | Google/TVA PPA delivery | 2030 |
| Commercial | Full-scale | 320 MWt | 140-150 MWe | Grid & data center deployment | 2030s |
Walk-away safe. If every operator left and every system failed, the reactor shuts itself down through physics alone. No pumps, no power, no intervention required.
Kairos is past the PowerPoint stage. Its Engineering Test Unit (ETU) program has logged 2,000+ hours of pumped molten salt operations at its Albuquerque facility.
They're the first company to perform online measurements of FLiBe salt chemistry throughout operations. Data that didn't exist before.
2026 manufacturing milestones: the Alameda facility earned ISO 9001:2015 certification (Jun 3, 2026), joining the 2025-certified Albuquerque campus; the ETU 2 vessel (Kairos's first fully in-house build) had its top head installed with startup expected later in 2026; and Kairos completed multi-year qualification of its ET-10 graphite with Ibiden (Apr 29, 2026).
NuCAMP (Aug 18, 2026): Kairos announced the Nuclear Center for Advanced Manufacturing and Precast at its Oak Ridge campus - an initiative to qualify new manufacturing and construction methods for nuclear use and to build a skilled-trades pipeline in East Tennessee. The signed MOU pulls in ORNL, UT-Knoxville, IACMI, Barnard Construction, Roane State, Chattanooga State, and Oak Ridge Schools; Samsung C&T E&C and Cambridge Vacuum Engineering are exploring advisory roles under an initial 12-month exploration phase. It's a methods-and-workforce consortium, not a funding round - no money was disclosed.
ETU 3 reactor cavity mockup (Sep 8, 2026): Kairos completed the first phase of testing on a non-nuclear, reduced-scale mockup of the Hermes 2 reactor cavity at ETU 3 in Oak Ridge (Interesting Engineering). The company installed 12 modular precast concrete shielding wall panels in four different joint designs - including stepped and sinusoidal geometries meant to stop radiation streaming through panel seams - and tested shielding plugs for the openings where process lines pass through the cavity, plus tooling to connect and disconnect those lines remotely so workers stay out of high-radiation areas. The panels were cast in 3D-printed polymer-composite forms developed with ORNL's Manufacturing Demonstration Facility, fabricated by Tindall Corporation in South Carolina, and installed by Barnard Construction. CTO Ed Blandford says the point is standardizing shielding interfaces and remote-maintenance features across the future commercial KP-FHR fleet; ETU 3 will also double as an operator training platform. Non-nuclear mockup work - useful de-risking, but it's concrete and robotics, not reactor physics.
ETU 2 modular build (Sep 10, 2026): Kairos published a writeup on how its second Engineering Test Unit went together in Albuquerque - the first ETU built as modules rather than stick-built in place. ETU 2 is assembled from 30+ truck/rail-shippable skids, each piped, wired, and insulated in the shop and tested there before shipping; crews laser-mapped the enclosure, dropped each skid onto reference marks, and only welded the joints between modules on site. Most of the hardware was built in-house, including the primary salt pump, the reactor shutdown safety system, and the ASME U-stamped vessel. The stated goal is for 80% of ETU 2's cost to come from raw materials or off-the-shelf parts - a target, not a published result. Status: installation and system testing underway, with integration and hot argon testing still ahead - later than the "close to startup" line CTO Ed Blandford gave ANS in April. Kairos says Hermes 2's reactor equipment modules will be fabricated the same way in Albuquerque and shipped to Oak Ridge. No fuel in ETU 2, ever - it's a build-method demonstrator, not a reactor.
Kairos has more advanced reactor test data than any other startup.
In August 2025, Kairos, Google, and the Tennessee Valley Authority signed the first commercial deal between Big Tech and an advanced nuclear company for data center power.
| Detail | Value |
|---|---|
| Reactor | Hermes 2 - 50 MWe |
| Location | Oak Ridge, Tennessee |
| Grid | TVA transmission network |
| Off-taker | Google data centers in AL & TN |
| Commercial Operation | 2030 |
| Broader Commitment | 500 MW of advanced nuclear by 2035 |
The structure: Google isn't buying the reactor - they're buying the electricity through TVA, which is a federally-owned utility. TVA manages grid integration; Kairos builds and operates the plant; Google gets 24/7 carbon-free power for its data centers.
500 MW by 2035 means Google is planning to buy power from multiple Kairos reactors. The PPA is the first deal in what's intended to be a fleet deployment.
On April 17, 2026, Kairos broke ground on the Hermes 2 Demonstration Plant in Oak Ridge, TN - the first commercial-scale reactor and first power-producing Gen IV reactor to receive an NRC construction permit. Hermes 2 is the first delivery under Kairos's Master Plant Development Agreement with Google: up to 50 MW to the TVA grid under the country's first binding advanced-nuclear PPA. Barnard Construction is general contractor; reactor modules are factory-fabricated in Albuquerque, NM.
Kairos holds two NRC construction permits - the first ever issued for non-light-water reactors in the United States:
| Permit | Issued | Reactor | Status |
|---|---|---|---|
| Hermes CP | Dec 2023 | 35 MWt demo | Construction started Jul 2024 |
| Hermes 2 CP | Nov 2024 | 50 MWe commercial | Groundbreaking Apr 17, 2026 |
Kairos has 14 topical reports approved by the NRC (with 1 under review). Topical reports are standalone safety analyses that establish regulatory precedent. Once approved, they don't need to be re-argued for future deployments.
The next gate is the operating license for Hermes. The NRC review is expected to follow the two-step process (construction permit - operating license), with the OL application anticipated in late 2026 or 2027.
Kairos is 2-3 years ahead of its nearest competitors on regulatory and construction milestones. No other advanced reactor company has:
TerraPower (Bill Gates, Natrium, 345 MW) is probably the closest in terms of funding and momentum, but they're further from construction permits. X-Energy has DOE ARDP backing and an Amazon partnership, but their timeline is behind Kairos. Oklo is public (NYSE: OKLO) but had to withdraw and resubmit their NRC application.
Heron Power isn't building reactors. It's building the infrastructure that connects power to data centers - replacing 100-year-old transformer technology.
Founded in 2025 by Drew Baglino, Tesla's former SVP of Powertrain and Energy Engineering (18 years; built the Megapack and energy storage business from zero to $10B+ revenue), Heron makes solid-state transformers that convert grid power straight to data-center-ready voltage. The claim: faster to build, smaller footprint, software-controlled, and made without the supply-constrained materials that have conventional transformers backordered 12-24 months.
| Round | Date | Amount | Lead | Key Participants |
|---|---|---|---|---|
| Series A | May 29, 2025 | $38M | Capricorn Technology Impact | Breakthrough Energy Ventures, Energy Impact Partners, Gigascale Capital, Powerhouse Ventures, Valor Equity. Angels: JB Straubel, Zach Kirkhorn. |
| Series B | Feb 18, 2026 | $140M | a16z (American Dynamism) + Breakthrough Energy Ventures | Capricorn (returning), Energy Impact Partners (returning), Valor Atreides AI Fund, Gigascale Capital (returning) |
| Credit Facility | Sep 10, 2026 | $60M | J.P. Morgan + TriplePoint Capital | Debt, not equity - the $178M equity total is unchanged. Earmarked for converting the factory site into a mass-production plant and scaling operations. Announced alongside former Tesla CFO Zach Kirkhorn (an existing angel) joining the board. |
Andreessen Horowitz - American Dynamism Fund ($1.2B fund) - a16z's pitch is national security through critical infrastructure. It invests in semiconductors, energy, defense, and AI. The bet here: the grid is a strategic weak point.
Breakthrough Energy Ventures ($3.5B+, Bill Gates) - 20-year investment horizon. Their criteria: technologies that can reduce global GHG emissions by at least 0.5%. Portfolio includes TerraPower, Fervo Energy, Form Energy, and Malta. They led both rounds alongside a16z.
Valor Equity Partners - Also led Aalo Atomics' round. Valor is assembling a portfolio across the data center power stack.
Baglino's Tesla track record is the strongest signal on this company. He joined in 2006 as an electrical/firmware engineer during the original Roadster, rose to SVP, and by the time he left in April 2024, he had built and shipped:
He was one of Tesla's three named corporate officers. He didn't leave because he failed - he left during a broader 10% workforce reduction in April 2024 and immediately started Heron.
When evaluating hardware startups, the question is "can this team actually build and ship at scale?" Baglino already did it at Tesla, in the same sector (power electronics and energy infrastructure), at a scale few people on earth have managed.
JB Straubel (Tesla co-founder) and Zach Kirkhorn (former Tesla CFO) both invested in the Series A. When your former co-founder and CFO write personal checks, it's a signal. In September 2026 Kirkhorn went further and joined Heron's board, announced alongside a $60M credit facility from J.P. Morgan and TriplePoint Capital - angel check upgraded to board seat.
A conventional transformer uses electromagnetic induction with heavy iron cores and copper windings. Same fundamental technology from the 1880s. Heron Link replaces all of it with power electronics.
| Spec | Heron Link | Conventional Transformer |
|---|---|---|
| Conversion | 34.5 kV AC - 600-800V DC (direct) | Multiple step-down stages required |
| Backup Power | Integrated SuperBBU (30-sec ride-through) | Separate UPS systems needed |
| Control | Software-defined, remote monitoring | Analog, manual inspection |
| Manufacturing | Weeks | 12-24 months |
| Key Material | Wide-bandgap semiconductors (SiC/GaN) | Grain-oriented electrical steel (shortage) |
| Scalability | Modular, infinitely stackable | Fixed capacity per unit |
| Power Flow | Bidirectional | One-way |
Heron says Heron Link does in a single unit what currently takes a large transformer, a distribution transformer, and a UPS system. Fewer components, smaller footprint, faster deployment.
AI data centers don't have a reactor problem - they have a power delivery problem. Even if you can generate 100 MW of clean power, you still need transformers to step it down and deliver it to server racks. The transformer supply chain is broken.
Heron sidesteps all of this. No electrical steel. Manufactured in weeks. Software-controlled for AI workload variability. Integrated backup power eliminates separate UPS procurement.
Heron is the picks-and-shovels play for the AI data center gold rush. Whether data centers are powered by nuclear, solar, gas, or grid - they all need transformers. And they can't get them fast enough.
The latest round is earmarked for a 40 GW annual manufacturing facility in the U.S. On August 13, 2026, Heron said it had picked the site: Heron Factory One, a $100M+ conversion of a former 286,000 sq ft distribution warehouse in Morgan Hill, CA, about 30 minutes south of San Jose and close to its Scotts Valley engineering HQ. The company claims the plant will turn out up to 10,000 Heron Link units a year - at 5 MW apiece, that's where the 40+ GW figure comes from. Site prep starts immediately; mass production is slated for late 2027. The project is partly backed by a $26.4M CalCompetes state tax credit (GO-Biz), which California says comes with 600+ jobs. To put the 40 GW claim in context:
Because Heron Link uses power electronics (not iron cores and copper windings), the manufacturing process looks more like a semiconductor fab than a steel foundry. More automatable, less material-constrained, and faster to scale.
| Phase | Date | Activity |
|---|---|---|
| Field Demos | Summer 2026 | First 10 engineering prototypes built; some delivered to early customers |
| Pilot Production | Early 2027 | Partner installations and early revenue |
| Full Production | H2 2027 | Manufacturing ramp begins |
| Scale | 2028-2029 | Production scaling toward 40 GW annual capacity |
The transformer market is enormous, boring, and completely unprepared for AI. Demand is up 274% since 2019. Lead times are measured in years. The core material - grain-oriented electrical steel - is controlled by a handful of global suppliers.
Heron doesn't need to capture the entire market. Even a small slice of a $70B+ market at higher margins (due to software control and integrated backup) represents a massive business.
Heron reports 40+ GW of early interest from data center operators, utility-scale solar developers, and energy storage companies - Baglino now says interest is "well north of" the first factory's 40 GW capacity (Latitude Media interview, Aug 24, 2026). At scale, that pipeline represents $8-12B in potential revenue. Interest is not backlog: the company has disclosed three customers by name - Intersect Power (the clean-energy developer Google acquired in 2025), data-center developer Crusoe, and (as of Sep 21, 2026) utility RWE Americas - and says it will spend the next year field-testing units at solar and data center sites to build runtime. Baglino's own framing: "To be truly bankable, we need more real runtime in the field."
The RWE deal (Sep 21, 2026) is Heron's first utility-scale pilot: Heron Link will replace low-voltage inverters and 60 Hz transformers at an RWE battery storage facility co-located with the Pyron Wind Farm in Scurry County, Texas, with installation targeted for April 2027. The agreement also outlines a path for RWE to reserve at least 500 MW annually of capacity from Heron's US factories - subject to definitive purchase agreements, so it's a reservation framework, not a purchase order. Still, it's the first named utility pilot and the first concrete capacity number attached to a customer.
On pricing, Baglino says he doesn't expect customers to pay a premium over conventional gear, leaning on trade tailwinds - the FCC ban on new foreign inverter models and tariffs up to 55% on imported power-conversion components - and claims Heron Link will be 5-6% more valuable than alternatives over a project's lifespan. Company math, not independently verified.
These companies dominate the traditional transformer market but are locked into 100-year-old electromagnetic technology. They have massive installed bases, long customer relationships, and manufacturing at scale - but they're slow to innovate and constrained by the same electrical steel shortages affecting the entire industry.
Heron's edge: different technology, different supply chain, faster manufacturing, software control. The risk is that incumbents develop competing solid-state products - likely by 2028-2029 - which gives Heron a 12-18 month first-mover window.
No other solid-state transformer startup has this combination of funding ($178M), team (Baglino), and investor quality (a16z + BEV). But it's not alone anymore: Ars Technica (Aug 24, 2026) counts Amperesand and DG Matrix alongside Heron as US solid-state transformer startups that have collectively raised $280M+ over the past year. Neither has a Baglino-scale operator or a factory in the ground, but the "only credible player" window is narrowing.
ARC Clean Technology develops the ARC-100, a 100 MWe sodium-cooled fast small modular reactor (SMR) derived from Argonne National Laboratory's EBR-II, which ran for 30 years. It is the most regulatory-advanced SMR in Canada - the only one to both complete a CNSC Phase 2 design review and have a License to Prepare Site application under review.
Founded in 2006, ARC's pitch is "proven, not novel": it builds on a reactor design with real operating history instead of a clean-sheet concept, which it sells as lower technology risk to utilities and data-center customers.
| Legal Name | ARC Clean Technology Inc. |
| HQ | Saint John, New Brunswick, Canada (US office: Washington, DC) |
| Founded | 2006 |
| Reactor | ARC-100 - 100 MWe sodium-cooled fast reactor (pool-type) |
| Lead Site | Point Lepreau, NB (with NB Power) |
| Stage | Private; late-stage development, pre-construction |
| One-line thesis | The proven EBR-II design path to commercial SMRs - regulatory leader in Canada |
| Round | Date | Amount | Notes |
|---|---|---|---|
| Provincial Support | 2018-2021 | ~$30M (CAD) | New Brunswick government ($20M in 2021 + $10M prior) |
| Series A | Apr 2022 | $30M | Per company announcement; plus Government of Canada & North Shore Mi'kmaq Tribal Council grants |
| Series B | Dec 16, 2025 | Undisclosed | CVC arm of a major (undisclosed) energy company, VC funds, strategic investors, family offices, existing investors. Round amount not disclosed. |
20+ years in international business and finance. Former partner at an Asian VC firm; prior roles at Cerberus Capital Management (Japan) and Deutsche Bank (New York). Stanford (International Relations).
34+ years in commercial nuclear operations. Former President/Chief Nuclear Officer of PSEG Nuclear and Site VP of Exelon's Peach Bottom station. Senior reactor operator background - rare operational depth for an SMR developer.
Career spanning nuclear design, construction and strategic planning. Former leader of CH2M Hill's Nuclear Business Group and former Chairman of the ITER Council. PhD Nuclear Engineering, Kansas State.
ARC's senior advisors include Dr. John Sackett (former Associate Director, ANL; director of the EBR-II division and originator of the 1986 "inherently safe" passive-safety tests), Dr. Leon Walters (former director of ANL's Fuels & Materials Division; metal-fuel pioneer), and Ronald King (30+ years on EBR-II mechanical/safety systems). This is the original EBR-II team - the deepest sodium-fast-reactor heritage of any startup.
Co-founders Donald M. Wolf (Chairman) and Scott L. Campbell (former senior US DOE official) round out the board.
The ARC-100 is a pool-type, sodium-cooled fast reactor running on metallic uranium-alloy fuel at atmospheric pressure - a design lineage traceable directly to EBR-II, which ran 1964-1994 and demonstrated walk-away passive safety in landmark 1986 tests.
| Spec | Detail |
|---|---|
| Electrical Output | 100 MWe |
| Thermal Output | 286 MWt |
| Reactor Type | Sodium-cooled fast neutron, pool-type |
| Coolant | Liquid sodium (355 °C inlet / 510 °C outlet) |
| Fuel | Metallic uranium alloy |
| Pressure | Atmospheric (no high-pressure accident path) |
| Fuel Cycle | 20-year refueling interval |
| Plant Life | 60 years; footprint < one city block |
| Partner | Date | Focus |
|---|---|---|
| NB Power | 2023 | Point Lepreau demonstration site; filed License to Prepare Site application |
| Deep Atomic | Jun 2025 | MOU - Swiss-American nuclear-for-data-centers; site ID, feasibility, hyperscale/edge customer engagement |
| Nucleon Energy / NuARC JV | May / Oct 2025 | Combined heat & power + electric-only sites in Alberta & Texas; multi-unit deployment framework |
| Korea Hydro & Nuclear Power (KHNP) | 2023 / Aug 2025 | 2023 MOU; Aug 2025 North American JV building on the May 2024 trilateral with NB Power |
| IC Nuclear (Türkiye) | Apr 7, 2026 | Non-binding term sheet for potential ARC-100 deployment in Türkiye |
| Canadian Nuclear Isotope Council | 2025 | Medical-isotope production using ARC-100 neutron flux |
| Milestone | Date | Status |
|---|---|---|
| CNSC Phase 2 Vendor Design Review | Jul 2025 | Complete |
| License to Prepare Site (Point Lepreau) | 2026 (expected) | Under review |
| Construction license application | 2027 | Planned |
| First commercial unit (Point Lepreau) | 2029-2031 | Projected |
| US NRC pre-application expansion | 2027-2028 | Planned |
ARC competes in the same sodium-fast-reactor lane as TerraPower's Natrium but at smaller scale (100 MWe vs 345 MWe) and with far less capital. What sets it apart is regulatory progress in Canada and the EBR-II operating history - a "de-risked, proven" story aimed at utilities wary of first-of-a-kind designs.
| Company | Technology | Capacity | Status |
|---|---|---|---|
| ARC (ARC-100) | Sodium fast reactor | 100 MWe | CNSC Phase 2 complete; LTPS under review |
| TerraPower (Natrium) | Sodium + molten-salt storage | 345 MWe | Under construction (Wyoming), $2B+ DOE |
| X-Energy (Xe-100) | HTGR, TRISO fuel | 80 MWe | DOE ARDP demo; Amazon partnership |
| Kairos (KP-FHR) | Fluoride salt, TRISO | 50-140 MWe | NRC permits; Google/TVA PPA |
| Aalo (Aalo Pod) | Sodium, LEU | 50 MWe | Achieved zero-power criticality Jul 2026; commercial deployment targeted 2028 |
Note: refreshed June 21, 2026 to incorporate 2026 developments (Türkiye/IC Nuclear term sheet, NRC Part 53 reaffirmation). The Series B (closed Dec 16, 2025) amount remains undisclosed and is not published here.
Fervo Energy makes 24/7 carbon-free baseload power by drilling deep into hot rock and creating artificial underground reservoirs. They took horizontal drilling and hydraulic stimulation from the oil and gas industry and applied it to geothermal energy - tapping heat sources that were never accessible before.
Founded 2017 by Tim Latimer and Jack Norbeck. Cape Station, their 500 MW flagship project in Utah, started delivering first power to the grid on Sep 24, 2026. Google is already buying their electricity for data centers. Fervo went public in May 2026 (Nasdaq: FRVO) - the first geothermal IPO in over a decade - at a valuation over $10B.
| Round | Date | Amount | Lead | Key Participants |
|---|---|---|---|---|
| Seed | 2018-2020 | ~$4.5M | Prelude Ventures | Breakthrough Energy Ventures, Congruent Ventures |
| Series A | Aug 2021 | $28M | DCVC | Breakthrough Energy Ventures (returning) |
| Series B | 2022 | ~$138M | Mixed | Liberty Energy (Chris Wright's company) |
| Series C | Mar 2023 | $221M | Devon Energy | One of largest US oil/gas companies |
| Series D | Mar 2024 | $244M | Devon Energy | Plus $25M DOE EGS Demonstration Grant |
| Series E | Dec 2025 | $462M | B Capital | Google, CalSTRS, CPP Investments, Capricorn, JB Straubel, Mitsubishi, Mitsui |
| Project Financing | Jun 2025 | $206M | Mixed | Loans + preferred equity for Cape Station |
| Cape Station Phase 1 debt | Mar 19, 2026 | $421M | Non-recourse (oversubscribed) | Project debt for Cape Station Phase 1 |
| IPO (Nasdaq: FRVO) | May 2026 | ~$2.2B | J.P. Morgan, BofA, RBC, Barclays | 70M shares at $27 (upsized base); gross proceeds ~$2.2B per Q2 2026 reporting; debuted May 13, rose ~33%; valuation over $10B |
B Capital (Lead, Series E) - Global multi-stage investment firm founded by Eduardo Saverin (Facebook co-founder) and Raj Ganguly. The Series E set up Fervo's May 2026 IPO on Nasdaq (FRVO).
Google - Investor and customer. Google joined the Series E and buys Fervo's electricity through Nevada PPAs for its data centers. Fits Google's 24/7 carbon-free energy goal for 2030.
Devon Energy (NYSE: DVN) - $28B oil and gas major that led Series C and D. It knows drilling and treats geothermal as an extension of it. A fossil-fuel incumbent on the cap table.
JB Straubel (Tesla co-founder) - Also invested in Heron Power. Building a portfolio across the clean energy infrastructure stack.
Liberty Energy - Company of Chris Wright, currently serving as US Energy Secretary under Trump. Political winds favor geothermal with Wright in the cabinet.
Former drilling engineer at BHP in the Permian and Eagle Ford basins. Left oil and gas in 2015 for climate, got a Stanford MBA, and founded Fervo in 2017. Named to Forbes 30 Under 30 Energy. Understands both the drilling technology and the business model needed to compete with fossil fuels at scale.
PhD in Energy Resources Engineering from Stanford, postdoc in earthquake hazard analysis. Principal Geothermal Reservoir Engineer. The technical brain behind Fervo's horizontal drilling approach to enhanced geothermal. Deep expertise in subsurface engineering and reservoir modeling.
Promoted to COO on June 10, 2026. Leads operations as Fervo scales Cape Station and integrates the business as a newly public company.
Traditional geothermal requires rare geology - natural hot water near the surface. Enhanced geothermal creates artificial reservoirs by drilling deep into hot dry rock and fracturing it. Fervo's approach is applying horizontal drilling and hydraulic stimulation from oil and gas to reach heat that wasn't reachable before.
| Spec | Detail |
|---|---|
| Drilling Depth | 8,500 ft at Cape Station (deeper projects up to 12,000+ ft) |
| Rock Temperature | 450°F+ at target depth |
| Well Configuration | Horizontal well pairs - injection and production wells |
| Stimulation | Hydraulic fracturing to create flow paths through hot rock |
| Capacity Factor | 90%+ (24/7 baseload power, like nuclear) |
| Technology Readiness | TRL-9: proven in real-world operational environment |
Fervo tested the technology at Project Red in Nevada. A 30-day well test delivered 63 liters/sec flow at 191°C. First company to drill horizontal well pairs for commercial enhanced geothermal. Project Red went operational in November 2023 and feeds carbon-free electricity to Google data centers via the Nevada grid.
As of April 14, 2026, Fervo reported Project Red had logged 614 operating days at ~2.1 MW gross / ~1.4 MW net, 175°C, with 98.4% uptime - which points to long-run reservoir stability.
Drilling times went from ~1 month per well to mid-teens days - a 70% year-over-year reduction. Roughly half the cost of wells is in drilling time. Faster drilling means lower cost electricity and competitive economics vs. fossil fuels.
The "record well" claim now has published numbers. Sawtooth 7 (announced Jul 8, 2026) reached 19,448 ft measured depth with a 7,500-ft lateral in 21 days - a company pace record and, Fervo says, a 143% drilling-rate improvement over its first Cape Station well. Follow-up coverage (Yahoo Finance, Jul 21; Stock Titan, Aug 12) repeats the "deepest and hottest" framing - depth and day-count are now on the record, but no temperature figure has been disclosed, so treat "hottest" as a company claim.
Cost figures are now on the record too (Latimer to CNBC, Sep 5, 2026): the first Project Red wells took 70 days to 11,220 ft; Cape Station Phase I averaged 21 days and 14,483 ft at ~$7,000 per kW; Phase II is targeted at $5,500 per kW and Latimer's long-term goal is $3,000 per kW. Those are company targets for Phase II and beyond - Phase I's $7,000/kW is the only realized number.
Fervo signed a Turboden agreement for up to 1.75 GW of ORC turbines (Apr 9, 2026) and an ABB equipment deal (Jan 28, 2026). Its well-productivity roadmap runs from EGS 1.0 (~2 MW/well) toward an EGS 3.0 target of ~16 MW/well. On Jun 23, 2026 PNNL, Fervo and NVIDIA announced a partnership to build an AI-powered Enhanced Geothermal System digital twin, targeting a deployable platform by 2029 - useful R&D, but well outside the Cape Station delivery window.
Cape Station in Beaver County, Utah is the world's largest next-generation geothermal development. Fully permitted, fully funded, and under construction.
| Phase | Capacity | Timeline | Status |
|---|---|---|---|
| Phase I | 100 MW | First power Sep 24, 2026 / ~100 MW by early 2027 | First power to grid achieved (Sep 24, 2026); GeoBlocks 1 & 2 mechanically complete (Aug 2026) |
| Phase II | 400 MW | 2028 | Construction planned |
| Total | 500 MW | - | - |
CNBC's on-site feature (Sep 5, 2026) reports Cape Station is on pace to send first power to the grid in October 2026 - which would make it the first commercial enhanced geothermal system in the US. Per CNBC: the first power plant is 33 MW, Phase II plants are standardized at 50 MW with a claimed 30%+ cut in construction time, and the Google deal brings Fervo's total contracted power to ~1 GW (Google, SCE, Shell Energy, NV Energy), with Latimer claiming advanced negotiations with more buyers and floating behind-the-meter data-center deals. CNBC also puts numbers on the early land grab: ~600,000 acres of geothermal rights across the West, bought at ~$4/acre before the AI boom - the same land now goes for $400+/acre. Company statements to a visiting journalist, but consistent with the filings.
First power landed ahead of that schedule: Fervo sent first electricity from Cape Station to the Utah grid on Sep 24, 2026 (Yahoo Finance, Seeking Alpha, Benzinga, Futu/Moomoo), which the company bills as the first next-generation geothermal power on the US grid. FRVO jumped on the news. One caveat on the framing: this is initial power from the first plant, not the full Phase I 100 MW - the ramp to ~100 MW by early 2027 is still ahead, and sustained output at scale is the part that actually validates the model.
Fervo has the deepest commercial relationship with Big Tech of any enhanced geothermal company. Google isn't just an investor - they're a long-term customer with signed power purchase agreements.
| Milestone | Date | Details |
|---|---|---|
| 115 MW PPA Signed | 2021 | NV Energy to supply Google data centers in Nevada |
| Project Red Operational | Nov 2023 | Carbon-free electricity flowing to Nevada grid for Google |
| Google Series E Investment | Dec 2025 | Google participated directly in $462M round |
| 396 MW Utah PPA | Sep 2026 | Largest geothermal power deal yet (WSJ, Sep 1); Fervo says the project comes online in 2028 - FRVO jumped ~24-28% on the news; follow-up coverage reports a path to ~1 GW (XenoSpectrum, Sep 3 - treat as reported); CNBC (Sep 5) puts Fervo's total contracted power at ~1 GW across Google, SCE, Shell Energy and NV Energy |
Google committed to 24/7 carbon-free energy across all operations by 2030. Solar and wind can't do it on their own because they're intermittent. Enhanced geothermal runs at 90%+ capacity factor - the kind of baseload Google's data centers need. The customer is also an investor, which is its own signal.
Google also holds a right-of-first-refusal on Fervo's output.
Rhodium Group analysis shows enhanced geothermal could provide nearly two-thirds of new data center demand through 2030. 80 GW of total data center peak demand is expected by 2030. Geothermal installed base is growing from 17.35 GW to 27.5 GW (9.65% CAGR).
Fervo's head start with Google could let it take meaningful share if other hyperscalers chase the same carbon-free strategy.
Fervo is the furthest along of any enhanced geothermal startup. They have commercial operations, signed PPAs, and are building the world's largest next-generation geothermal project.
| Advantage | Why It Matters |
|---|---|
| Commercial operations | Project Red delivering electricity since Nov 2023 |
| Signed PPAs with Google | 115 MW Nevada contract plus a 396 MW Utah deal (Sep 2026) validates the commercial model |
| 500 MW Cape Station | Largest next-gen geothermal project globally |
| Oil & gas expertise | Proven drilling techniques, faster execution |
| 90%+ capacity factor | Baseload power vs. intermittent renewables |
Sage Geosystems (partnered with Ormat), Eavor (closed-loop), GreenFire Energy, and Zanskar are developing enhanced geothermal, but none have reached Fervo's commercial scale.
Faster to deploy (years not decades), no radioactive waste, no NRC licensing. Enhanced geothermal delivers clean baseload power with less regulatory complexity.
Currently limited to western US where hot rock is accessible at reasonable depths. Expansion east requires deeper drilling and higher costs. This is the main constraint on national scale deployment.
AI workloads are driving the largest single-category increase in electricity demand since industrialization. The IEA projects global data-center electricity use roughly doubling to ~945 TWh by 2030; US utilities and grid operators are forecasting multi-hundred-terawatt-hour additions concentrated in a handful of regions. The bottleneck has shifted from compute to power: generation, interconnection, and the equipment that delivers it.
Advanced nuclear / SMRs (Kairos, Aalo, ARC, plus TerraPower/X-Energy/Oklo) promise dense, dispatchable, carbon-free power but face NRC/CNSC licensing timelines and first-of-a-kind build risk - most commercial units land 2028-2031+. Enhanced geothermal (Fervo) delivers nuclear-like capacity factors with no fission licensing, but is geographically limited to the hot-rock West. Cutting across both: grid hardware (Heron) - whatever generates the electrons, they still need transformers data centers can't currently buy fast enough.
| Company | Partner | Deal | Scale |
|---|---|---|---|
| Kairos Power + TVA | PPA for Hermes 2 reactor, data centers in AL/TN | 500 MW by 2035 | |
| Amazon | X-Energy + Energy NW | SMRs in Washington State; Dominion MOU in VA | Multi-GW |
| Microsoft | Constellation Energy | Restart Three Mile Island Unit 1 for data center power | 837 MW |
| Meta | Oklo + TerraPower | 1.2 GW nuclear campus in Pike County, OH | 1.2 GW |
| Switch | Oklo | 12 GW Aurora deployment over 20 years | 12 GW |
Total announced Big Tech nuclear commitments: $10B+ across 2026-2035 deployment timelines.
| Company | Tech Type | Stage | Total Raised | Valuation | Timeline to Deploy | Key Partner |
|---|---|---|---|---|---|---|
| Aalo Atomics | 50 MWe modular nuclear (LEU) | Pre-IPO | ~$133M | Undisclosed | Jul 2026 demo / 2028 commercial | Valor / Microsoft |
| Kairos Power | Molten-salt reactor (KP-FHR) | Pre-IPO | $100M+ est. + $303M DOE | Undisclosed | 2027 demo / 2030 commercial | Google / TVA / DOE |
| Heron Power | Solid-state transformers | Pre-IPO | ~$178M | Undisclosed | Mid-2026 demos / 2027 production | a16z / BEV |
| ARC Clean Tech | 100 MWe sodium fast reactor | Late dev. | ~$26M | Undisclosed | 2029-2031 first unit | NB Power / Deep Atomic |
| Fervo Energy | Enhanced geothermal (EGS) | Public (Nasdaq: FRVO) | ~$2.2B IPO + $1.3B+ private | >$10B | First power 2026 / ~100 MW early 2027 | Google / Devon |
| TerraPower | Natrium sodium (345 MWe) | Construction | $1B+ + $2B DOE | ~$8B+ | Late 2020s | Bill Gates |
| X-Energy | HTGR Xe-100 (80 MWe) | Late dev. | $700M+ | ~$2B | 2028+ | Amazon |
| Oklo | Aurora fast reactor | Public (NYSE: OKLO) | Public | Public mkt | 2027+ | Altman / Switch |
Green = profiled in this dashboard; gray = context. Valuations are private and largely undisclosed; figures shown are reported targets/estimates only. "Total Raised" mixes equity and, for Fervo, project financing. As of ~mid-2026.
The regulatory environment is the single biggest gating factor for advanced nuclear - and, as of 2026, it is unusually favorable. Federal policy under Energy Secretary Chris Wright (whose former company, Liberty Energy, backed Fervo) is explicitly pro-nuclear and pro-geothermal.
All five companies profiled here are pre-IPO or late-stage private - early enough to still carry risk, far enough along to have real technology, real investors, and real milestones. Each targets a different part of the data-center power problem:
| Company | Solves | Thesis |
|---|---|---|
| Kairos Power | Where the electricity comes from | Lowest-risk nuclear play. Google PPA, NRC permits, DOE backing. Furthest-advanced reactor in the U.S. |
| Aalo Atomics | Right-sized power for data centers | 50 MW sweet spot, LEU fuel advantage, behind-the-meter model. Higher risk, higher reward than Kairos. |
| Heron Power | How electricity gets delivered | Source-agnostic infrastructure play. Every data center needs transformers regardless of generation method. Fastest to revenue. |
| ARC Clean Tech | The proven-design nuclear option | EBR-II heritage, Canada's regulatory leader. Lowest technology risk but smallest balance sheet; longest timeline. |
| Fervo Energy | Clean baseload power - today | Already delivering electricity to Google. 500 MW under construction. Now public (Nasdaq: FRVO) at >$10B. Lowest regulatory risk of the five. |