Source Report

Aurora Powerhouse Technical Design

Oklo's Aurora powerhouse evolved from an initial 1.5 MWe microreactor concept—rejected by the NRC in 2022 due to insufficient data—into a scalable 15-75 MWe sodium-cooled fast fission reactor by leveraging the proven Experimental Breeder Reactor-II (EBR-II) heritage, which operated for 30 years with inherent safety during severe transients like loss-of-flow without scram. This design uses heat pipes to transfer heat from a compact, pool-type metal-fueled core to a supercritical CO2 or steam power conversion system (with Siemens providing turbines for the first unit), enabling factory prefabrication, atmospheric pressure operation, and passive cooling via natural forces for walk-away safety—no active pumps or water needed. Scalability to 75 MWe (first deployment at Idaho National Lab, INL) matches data center "halls" without new risks, using high-assay low-enriched uranium (HALEU) metallic fuel (initially from 5 metric tons of recycled EBR-II waste), recyclable for 10+ year intervals, breaching >95% more energy from spent fuel than light-water reactors.[1][2]
- Ground broken Sep 2025 at INL for 75 MWe Aurora-INL; DOE Reactor Pilot Program selection accelerates to potential 2026 criticality via authorization pathway (bypassing full NRC COL initially), targeting commercial ops late 2027/early 2028.[4][5]
- Principal Design Criteria topical report under NRC review; fuel fabrication facility design DOE-approved Oct 2024 using EBR-II HALEU.[6]
For competitors entering advanced nuclear, Oklo's EBR-II data moat (400+ reactor-years globally for similar tech) shortens licensing to 24-36 months vs. 44+ for novel designs, but HALEU supply constraints (Russia sanctions, limited U.S. production) demand fuel recycling scale-up by early 2030s.

PPA Business Model Structure

Oklo rejects traditional nuclear's "license-to-utilities" model—in which developers like Westinghouse offload lifecycle risks, leading to Vogtle-style overruns ($30B+ for 2.2 GWe)—for a build-own-operate (BOO) utility-like approach: customers sign non-binding LOIs/MOUs framing sites/pricing, escalating to binding 20-year PPAs (renewable +20 years) where Oklo funds/owns everything, delivering power/heat at fixed rates without customer capex. Revenue accrues via recurring sales ($105/MWh illustrative), capturing ops efficiencies and fuel recycling margins; e.g., upfront ROFR payments ($25k from Equinix) discount future PPAs. This aligns incentives for small-scale (15-75 MWe) co-location at data centers/factories, minimizing transmission and enabling phased redundancy (>99% uptime).[8]
- Step-by-step: MOU → Scoping/LOI (pricing, MW, milestones) → Site eval/engineering → PPA → Oklo builds/ops; e.g., Switch's 12 GW Master Agreement (Dec 2024) frameworks site-specific PPAs through 2044.[9]
- No grid sales at INL initially (DOE rules), but model suits off-grid/defense; profitability from year 1 via low opex ($2.4M fixed + $5/MWh variable illustrative).[10]
Entrants must match this customer de-risking—non-binding pipeline converts via prepayments (e.g., Equinix $25M)—but face execution hurdles like Oklo's pre-revenue status ($52.8M 2024 losses).

Announced Customer LOIs and Pipeline

Oklo's ~14 GW pipeline (up 93% YoY to Aug 2024, now 14 GW per Q1 2025 update) targets data centers (600 MW+), oil/gas, defense, hyperscalers—validated by AI demand (U.S. power growth 31% 2022-2030)—with non-binding LOIs converting to PPAs; largest is Switch's 12 GW Master Power Agreement (one of history's biggest corporate clean PPAs), plus Meta's 1.2 GW Ohio campus (prepay-funded, online ~2030). Publicly disclosed: Equinix (500 MW pre-agreement, $25M prepay), Prometheus/Wyoming Hyperscale (100 MW data center), Diamondback Energy (50 MW Permian oil/gas, 20-yr PPA LOI). Sectors: Data centers (e.g., two majors +750 MW Nov 2024), energy (TVA eval), defense (Eielson AFB tentative), industrial.[11][12][13]
- Total disclosed MW under LOI/Master Agmt: ~14 GW (Switch 12 GW dominant); conversions ongoing H2 2024-H1 2025.
- Recent: RPower phased gas-nuclear for data centers; Liberty Energy partnership.[14]
Competitors lack Oklo's hyperscaler focus (e.g., Meta prepays de-risk capex), but pipeline fragility (all non-binding) means entrants need big-tech anchors to fund first-of-a-kind (FOAK) builds.

Unit Economics and Revenue Mechanics

Illustrative NOAK economics (2023 presentation, unverified post-scale): 15 MWe Aurora capex $24M ($1.6k/kWe incl. $33M initial fuel at $7k/kg HALEU), refuel $17M/10 yrs; revenue $13M/yr ($105/MWh, 92% CF, 121k MWh); opex $3M/yr → 76% cash margin, 2.5x capex recovery over 40 yrs, LCOE ~$40/MWh (with 30% ITC). Fuel recycling cuts costs 80% long-term (used fuel >95% energy left), enabling profitability year 1 vs. traditional nuclear losses. Cash: $1.2B (Q3 2025, post-$540M ATM), burn $65-80M/yr.[10][5]
- Vertically integrates fuel (Tennessee recycler $1.68B, plutonium bridge); radioisotopes via Atomic Alchemy (2026 rev).
Skeptics note HALEU reality ($35k+/kg vs. $7k assumed) inflates LCOE; entrants validate via pilots, as Oklo's FOAK ~$34M signals scaling risks.

PPA Model vs. Traditional Nuclear Economics

Oklo's BOO-PPA flips traditional nuclear's developer-utility split—where capex overruns (Vogtle $30B/GWe, LCOE $150+/MWh) stem from misaligned incentives—by internalizing full lifecycle (design-build-op-recycle), targeting $4-5k/kWe NOAK vs. $6-10k+ for large LWRs/SMRs, with distributed 15-75 MWe suiting off-grid (no $B transmission). PPAs lock recurring rev (20-40 yrs) at competitive rates, capturing recycling uplift; traditional regulated PPAs burden ratepayers with overruns, delaying via custom licensing (10+ yrs).[8][10]
- Traditional: $10k+/kWe capex, 60-100 month build, LCOE $90-175/MWh; SMRs like NuScale $4.8k/kWe but FOAK higher.[15]
- Oklo edge: 18-month install, 90%+ CF, but pre-revenue (2027 rev start).
New entrants prioritize BOO for data centers (e.g., Meta's prepay), as regulated utilities resist SMR scale economics absent subsidies. Confidence: High on model (SEC filings), medium on economics (illustrative, HALEU volatility); further SECs needed for 2026 unit data.

---

Recent Findings Supplement (March 2026)

Aurora Powerhouse Technical Specifications

Oklo's Aurora Powerhouse has evolved from its original 1.5 MWe microreactor design (NRC-denied in 2022) to a scalable sodium-cooled, liquid metal–cooled, metal-fueled fast fission reactor platform offering flexible output of 15–75 MWe per unit (expandable to 100 MWe+ via chaining), building on Experimental Breeder Reactor-II heritage for 10–20 year fuel cycles using high-density metallic uranium or recycled fuel; groundbreaking on the first 75 MWe Aurora-INL unit at Idaho National Laboratory occurred September 22, 2025, under DOE Reactor Pilot Program, targeting commercial operations late 2027/early 2028.[1][2]
- DOE approved preliminary safety analysis for Aurora Fuel Fabrication Facility (A3F) at INL in December 2025, enabling assembly of EBR-II repurposed fuel (5 MT secured, only company with fuel for first commercial unit); November 2025 binding contract with Siemens Energy for power conversion (SST-600 turbine, SGen-100A generator).[3][4]
- NRC accepted Principal Design Criteria topical report September 30, 2025 (15 days vs. 30–60 norm), with draft evaluation due early 2026 under accelerated timeline per ADVANCE Act/EO directives.[5]

Implications for competitors/entrants: Aurora's factory-fabricated modularity and fuel recycling (e.g., 20 MT plutonium for 1.8 GW/25 units) slash capex to ~$3,000–4,000/kW vs. traditional nuclear's billions, but HALEU bottlenecks persist—new March 9, 2026 Oklo-Centrus JV for Ohio deconversion/enrichment co-locates supply next to 1.2 GW campus, de-risking fuel for fast reactors but requiring regulatory coordination.[6]

Power Purchase Agreement Business Model Structure

Oklo pursues a build-own-operate (BOO) utility model, developing/owning Aurora plants and selling electricity/heat via long-term (20-year) PPAs directly to customers (behind-the-meter or grid-additive), funded by private capital/prepayments; this generates recurring revenue while streamlining regulation vs. selling designs, with vertical integration into fuel recycling/radioisotopes for cost edges (e.g., recycled UNF unlocks 200 GW capacity).[7]
- Q3 2025 presentation reaffirms BOO with direct sales; no revenue yet (pre-commercial), but cash burn on track at ~$1.2B runway.[7]
- Meta agreement (Jan 9, 2026) introduces prepayment mechanism: Meta funds Phase 1 (fuel procurement, pre-construction 2026) for 1.2 GW Ohio campus (multiple Auroras, first power ~2030, full by 2034), adding grid power without ratepayer costs.[8]

Implications for competitors/entrants: BOO shifts capex risk to Oklo (hundreds of millions/unit vs. traditional GW-scale GW-scale billions), enabling data center "prepayment certainty" absent in utility models, but demands hyperscaler credit (e.g., Meta) for financing—new entrants need similar off-grid demand or DOE pilots to compete.

Announced Customer Letters of Intent and Pipeline

Oklo's pipeline totals ~18 GW in non-binding LOIs/MPAs (Q3 2025), dominated by data centers (4,750 MW recent adds), with oil/gas/defense; no binding PPAs beyond frameworks, but Meta elevates to funded path.[9]
- Data centers: Switch (12 GW MPA by 2044), Equinix (500 MW +$25M prepay), Prometheus Hyperscale (100 MW), three unnamed (4,750 MW total adds); Meta (1.2 GW funded).
- Oil/gas: Diamondback Energy (50 MW LOI).
- Utility/defense: TVA (power/fuel eval), Eielson AFB (DoD microreactor pilot), ~14 GW total pre-Meta per older filings.[10]

Implications for competitors/entrants: 18 GW validates demand (AI/data centers 70%+), but non-binding nature (convert via milestones) favors Oklo's early DOE fuel/site edges; rivals must secure hyperscaler prepays to match pipeline momentum.

PPA Model vs. Traditional Nuclear Economics

Oklo's BOO/direct PPA model targets LCOE-competitive output via modularity/recycling (est. $3,000–4,000/kW build, low opex from 10–20yr cycles), vs. traditional plants' $10,000+/kW overruns/2yr refuels; factory build (3–5yrs) + waste-fueled abundance beats gigawatt-scale delays, with prepays de-risking vs. utility balance sheets.[11]
- No 2025+ public LCOE (prior claims $64–73/MWh vs. legacy nuclear); fuel recycling (96k MT UNF) + plutonium/HALEU cuts costs 50%+ long-term.[7]

Implications for competitors/entrants: Oklo's scale economics unlock data center premiums (off-grid, 24/7), but pre-revenue status demands flawless execution; traditional players can't match without SMR pivots, while fuel JVs (Centrus) widen moat—estimated, high confidence on mechanics, medium on LCOE (no fresh data).