A pre-revenue bet that the AI build-out runs out of electrons before Oklo runs out of runway. The Aurora fast-reactor as a call option on the power bottleneck.
Oklo is the purest listed proxy for the thesis that electricity, not silicon, is the binding constraint on the AI build-out. Its Aurora sodium-cooled fast-reactor is sold as power under 20-40 year PPAs, not as hardware - an independent-power-producer model wrapped around a first-of-a-kind reactor. The stock is a compound option on three sequential unlocks: (i) an NRC operating licence after the historic 2022 COLA denial, (ii) conversion of a ~14 GW non-binding pipeline into binding offtake, and (iii) commercial heat at Idaho National Lab by late-2027/early-2028. At $52.62 the market prices roughly half of that succeeding. We rate the name a HOLD / Speculative Accumulate, conviction 3/5: the structural tailwind is real and the probability-weighted value ($76) sits ~44% above spot, but a 30% left tail to $22, ~20%-of-float short interest, ~95% implied vol, and relentless ATM dilution demand small sizing and a preference to add into weakness toward the $45 trough.
Few stocks map the full euphoria-to-capitulation cycle as cleanly as Oklo. From the post-SPAC trough of $5.35 (09 Sep 2024) to the intraday all-time high of $193.84 (15 Oct 2025) is a +3,524% round trip - then a 76.5% collapse to $45.58 (30 Mar 2026). The chart below reconstructs the journey; the acts below decode the catalysts.
Opens ~$15, dumps 54% day-one on AltC redemptions. Drifts to the $5.35 ATL. No design, no NRC filing, no revenue.
Microsoft-Constellation (Three Mile Island) proves the trade. Oklo ~4×'s. Kerrisdale "Fission Impossible" short bounces off Trump's win.
Switch 12 GW deal, Atomic Alchemy, Trump EO 14302 (4× nuclear), NRC readiness "no gaps." Pipeline 0.7 → 14 GW.
September +51.6%, +500% YTD. INL groundbreaking, DOE pilot pick. Retail/WSB max. ATH 15 Oct at ~$20B cap.
DeepSeek-R1 efficiency fears, $1.5B ATM overhang, momentum unwind. Trough $45.58. NRC PDC approval sparks partial recovery.
The current-window read is a textbook good-news-that-won't-move-the-tape setup. The bullish catalysts are stacking - the 01 Jul DOE Final Safety Analysis approval, the 18 Jun Centrus/LEU HALEU fuel LOI (which "directly kills the market's biggest overhang"), the Eielson AFB date-certain, the Creative Engineers sodium-systems tuck-in - yet the stock printed $52.33 below all three EMAs, RSI 40, MACD bearish on the DOE news. Reddit sentiment holds bullish (62-66, spiking 72 on 25 Feb) with a persistent WSB "OKLO to $250 by July 4" narrative tied to America's 250th birthday, while the broader retail base has rotated out. The bear voice is quieter but pointed: valuation and decade-out commercialisation.
There is no revenue to decompose - so the honest analogue to a "revenue split" is the pipeline mix and the future unit economics of the IPP model. Below: what Oklo is building, who has signed (and how firmly), and the margin structure a mature Aurora fleet would carry.
Reactor: liquid-sodium-cooled, metal-fuelled fast reactor (EBR-II heritage), heat-pipe transport into a Siemens SST-600 steam/sCO₂ power block. Three tiers: 15 MWe (original), 50-75 MWe (current commercial build), scaling to 100 MWe+ for campuses.
Model: Oklo builds, owns and operates the plant and sells electricity + heat under 20-40 year PPAs. It is an independent power producer, not a reactor OEM - recurring, contracted, inflation-linked cash flows once online. First unit: Aurora-INL, 75 MWe, groundbreaking 22 Sep 2025 (Kiewit constructor), commercial heat targeted late-2027/early-2028.
Fuel edge: runs on HALEU and recycled spent fuel; the newcleo Oak Ridge center ($1.68B) and Atomic Alchemy (VIPR isotope reactor) extend the model into a closed fuel cycle and a near-term isotope revenue line.
Nuclear IPPs carry very high gross margins once built: fuel is a small share of cost and the asset runs at ~90%+ capacity factor for decades. We model a mature Oklo unit at ~60-65% gross, ~40-45% EBITDA and ~30-35% operating margin at scale - but only after the 2026-2030 capex wave. Today every margin line is undefined (revenue = $0); FCF does not inflect positive until the first ~2-4 units are operating and the fuel center is contributing.
The one unambiguous strength. Q1'26 liquidity is ~$2.54B (cash + marketable securities) after a fully-drawn $1.5B ATM and a fresh $1.0B facility. Against ~$80-100M operating burn and $350-450M capital deployment guided for 2026, that is multi-year runway - Oklo can fund Aurora-INL to completion without a financing cliff, though not without further dilution. Debt is negligible (<$2M); net cash ≈ $2.5B.
Oklo reports as a single pre-revenue entity, but the equity is really three stacked options: the Aurora power business, the Atomic Alchemy isotope line, and the fuel-recycling franchise. A risk-adjusted SOTP lands the enterprise near where it trades - the market is paying roughly fair value for the parts, with little conglomerate discount and a lot of execution faith.
| Segment | What it is | Basis | Risk-adj EV | Status |
|---|---|---|---|---|
| Aurora IPP core | Power under 20-40yr PPAs | DCF of credible 2-4 GW (INL, Meta, Equinix) | $5.0-7.0B | Pre-rev |
| Atomic Alchemy | Radioisotopes (VIPR) | Near-term rev optionality '26-'27 | $0.5-1.0B | Launching |
| Fuel recycling | newcleo Oak Ridge center | Pure optionality / moat | $0.5-1.5B | Early |
| Net cash | Cash + securities − debt | Balance sheet | $2.5B | Held |
| SOTP enterprise | Sum of parts | - | $8.5-12B | vs ~$9.1B cap |
Reverse-engineering the ~$6.6B EV: discounting at ~15% (a high-beta, ~95%-IV, pre-revenue name) over ~9 years implies the market expects a future enterprise value near $23B by 2035. At a 15× EBITDA exit that is ~$1.5B of terminal EBITDA - roughly 3.5-5 GW of operating capacity at 40-45% margins, i.e. the market is underwriting ~35-50% of the current LOI pipeline being converted, built and run profitably. Aggressive, but not absurd if AI-power scarcity persists and the NRC clears the design.
| Company | Mkt cap | Revenue | Reactor | NRC status | vs OKLO |
|---|---|---|---|---|---|
| OkloOKLO | $9.1B | $0 | Na fast, 75 MWe | PDC approved | - |
| NuScale PowerSMR | $3.3B | $0.6M | LWR SMR 77 MWe | Design certified | Ahead on licence |
| NANO NuclearNNE | $1.2B | $0 | Microreactor | CPA accepted | Earlier stage |
| BWX TechnologiesBWXT | $17.3B | $3.75B | Naval + SMR | Commercial | Revenue + backlog |
| Centrus EnergyLEU | $3.35B | $0.5B | HALEU enrichment | n/a supplier | Oklo's fuel supplier |
| CamecoCCJ | $20B+ | $2B+ | Uranium miner | n/a | Upstream fuel |
| GE VernovaGEV | $80B | large | BWRX-300 | Pre-licensing | Utility-scale |
Private peers frame the licensing ladder: Kairos (Hermes construction permit in hand, first corporate SMR PPA with Google), TerraPower (Natrium CP under active NRC review, Gates-backed) and X-energy (Amazon/Nucor) are all further along on either licence or offtake specifics. Oklo's edge is pipeline scale and vertical integration; its deficit is that it is behind on the single gating item - a licence.
Toggle the cases. For a pre-revenue optionality name, the "target" is a probability-weighted anchor, not a point forecast - the distribution is bimodal, skewed by a fat regulatory left tail.
The 2022 COLA denial was the only one in NRC history. PDC approval is a topical report, not a licence; full COLA review could take 2-4 years. The single gating binary for the whole thesis.
+76% shares since end-'24; $1.0B ATM still open. Raising equity at depressed prices on a $0-revenue balance sheet compounds against holders.
An AI-efficiency narrative (DeepSeek-style) or a data-centre capex pause removes the power-scarcity bid that underwrites the multiple. Already happened once in Jan-2026.
First-of-a-kind fast reactor with no cost benchmark. Nuclear FOAK projects are endemic to overruns and delay; Kiewit is capable but the design is novel.
HALEU is <1% of global enrichment; Russia ban tightening to 2028. Centrus LOI (Jun-'26) is non-binding; first-core fuel not yet contractually locked.
Switch's 12 GW is a framework; most of the 14 GW carries no penalty for walking away. Conversion to binding PPAs is unproven at scale.
Every insider transaction in 12 months is a sale/sell-to-cover; CEO sold ~$22M via 10b5-1. Altman stepped down as chair (Apr-'25). No open-market buys.
~20% of float short (most-shorted utility), ~95% implied vol, a 2× leveraged ETF (OKLL) exists. Multi-sigma daily moves in both directions; squeeze and flush risk both live.
0.5-1.0% of an equity sleeve as a speculative, uncorrelated call option. Not a core holding. Scale toward 1.5% only on a confirmed NRC operating licence (conviction → 4). Treat as venture-style: size to survive the -58% bear.
Accumulate patiently. The risk-reward inflects near the $45 trough - below $45 this becomes a conviction-4 buy (skew turns clearly favourable). DCA in thirds; avoid chasing green days into resistance ($77 prior recovery high). Support $49.56 / $44.88.
Trim into $110-130 (Street bull cluster) and again toward the $155 bull case. Hard stop / thesis-break: a materially negative NRC ruling, Switch or Meta walking, or a confirmed structural cut to hyperscaler power demand. Re-underwrite below $40.
The honest tension: probability-weighted value ($76) sits 44% above spot, which on its own reads as a Buy. We stop at HOLD / Speculative Accumulate because the expected value is carried by a 25% bull tail while a 30% bear tail loses more than half your capital - a distribution that rewards small size and patient entry over conviction. If the AI build-out is genuinely power-constrained this decade, a licensed, contracted 14 GW reactor pipeline is worth multiples of today's cap. The variable is not the prize; it is the path.
The July-4-2026 criticality goal under the DOE pilot program is the nearest date-certain proof point. Hitting it validates the accelerated timeline; slipping it re-opens the "always two years away" bear.
Watch cash burn vs the $80-100M opex guide, ATM #2 drawdown, and any LOI-to-binding conversion. Isotope (Atomic Alchemy) first-revenue commentary is the swing factor.
The full safety-analysis submission and NRC acceptance is the thesis's spine. Acceptance de-risks; an RFI-heavy pushback or denial is the -58% path.
Any of Switch (12 GW), Meta (1.2 GW) or Equinix converting a framework into firm, penalty-bearing offtake would re-rate the pipeline from "optionality" to "backlog."
Late-2027/early-2028 first heat is the secular-thesis validation milestone - the moment Oklo stops being a story and starts being an operating IPP with revenue.
Hyperscaler capex trajectory and any efficiency-driven demand scare (the DeepSeek risk) plus the discount-rate regime for long-duration, cash-negative equities.