Initiated 2026-10-03 · Price $35.87 (as of 2026-10-02, stockanalysis.com) · Mkt cap $6.67bn (186.02mm shares) · Utilities / advanced nuclear fission developer (pre-revenue) · Model: none — pre-revenue; no module fits

Rating: Sell — Conviction: Low

Even after a ~69% fall over the past year, the market still values Oklo at roughly $3.7bn above its $3.0bn of cash and securities, for a fleet that has no binding power contract, no NRC license for commercial units, and a first 75 MWe unit targeted for 2028. At the one third-party build-cost estimate available, a unit only roughly breaks even at the top of Oklo's own price range, and every step to scale is funded by issuing new stock, so existing holders bear the gap between the story and the cash flows. The rating is directional and low-conviction because a binding PPA or a licensing win could re-rate the name sharply.

Conviction tests (3a-v-c): T1 fail — not runnable, no model (pre-revenue) · T2 fail — not runnable, no model · T3 pass — every balance-sheet and operating fact from the 10-K/10-Q; unit-cost and pricing inputs are third-party/company-presentation figures, flagged · T4 fail — conversion of the non-binding Switch 12 GW agreement, NRC COLA filing/acceptance, and Aurora-INL schedule news can each move value >25%

Business overview

Oklo designs sodium-cooled fast reactors (Aurora, 75 MWe, metal HALEU fuel derived from the EBR-II design) and intends to build, own and operate them, selling electricity and heat under PPAs rather than licensing designs. Side businesses are radioisotopes (Atomic Alchemy, the Groves isotope test reactor in Texas, which reached criticality 2026-08-06), fuel fabrication and a planned Tennessee fuel-recycling center (~$1.68bn roadmap). H1 2026 revenue was $1.2mm, almost all from two machining/engineering shops bought in June 2026; operating loss was $124.2mm. Value depends on three variables: (1) cost and schedule of the first Aurora units, (2) the price and bindingness of power contracts, and (3) HALEU fuel availability, which the 10-K says is "not currently available at scale."

Competition

The competition that matters is for the same hyperscaler and utility contracts. TerraPower's Natrium received the first NRC construction permit for a commercial non-light-water reactor (2026-03-04), began nuclear construction 2026-04-23, and signed Meta for up to eight 345 MW plants (Jan 2026) — the same customer that signed Oklo's 1.2 GW Ohio prepayment agreement. TVA received an NRC construction permit for a GE Vernova Hitachi BWRX-300 at Clinch River (2026-09-29), and X-energy (listed as XE) has a Dow construction application under NRC review that has cleared its environmental assessment with a finding of no significant impact. Oklo's first unit runs under DOE authorization, not an NRC license; its commercial combined license application has not been refiled since the 2022 denial. Pressure would show up first as price and contract terms: hyperscalers can choose vendors with licenses already in hand. The closest listed comparable, X-energy, carries a similar EV ($4.24bn vs Oklo's $4.21bn on the same source) with $150mm TTM revenue and a further-advanced NRC docket.

Bull case

  1. Pipeline converts — the non-binding Switch 12 GW Master Power Agreement and Meta's 1.2 GW Ohio campus (150 MW ~2030, full by 2034) become binding PPAs at firm-power prices. Plays out if data-center demand for firm clean power keeps outrunning grid supply. Model: not in the value range — no model
  2. Regulatory speed — DOE approvals came fast in 2026 (NSDA in March, PDSA 2026-06-11) and the NRC approved Aurora's Principal Design Criteria topical report on an accelerated schedule; a streamlined NRC could cut years off commercial licensing. Plays out if NRC timelines set by the 2025 executive orders hold. Model: not in the value range — no model
  3. Funded runway — $3,006mm of cash and securities at 2026-06-30 against 2026 guidance of $120–150mm operating cash use plus $400–500mm PP&E covers several years at the current pace. Plays out if spend stays near guidance. Model: not in the value range — no model

Bear case

  1. Unit economics are thin — BloombergNEF's estimate of $350–400mm per 75 MW powerhouse against Oklo's own $40–90/MWh price range gives ~$24–53mm a year of gross revenue per unit (est., 90% capacity factor); assuming fuel and O&M take ~40% of revenue over a 40-year life (est.), that is roughly breakeven at a ~8–10% cost of capital at the top of the range, and value-destroying at the bottom. Plays out if first-of-a-kind costs land at or above the estimate. Model: not in the value range — no model
  2. Serial dilution — shares rose from 173.6mm (2026-03-13) to 186.0mm (2026-08-04); H1 ATM sales raised $1.85bn net, a $1bn ATM launched in May was exhausted within four months, and a new $1bn program started 2026-09-11. Scaling to 1.2 GW at the BNEF estimate (~$5.6–6.4bn) needs more. Plays out if prepayments and project debt don't arrive before equity is spent. Model: not in the value range — no model
  3. Licensing and fuel gate — no commercial NRC license, no refiled COLA, and HALEU supply only via a Centrus letter of intent and a 5 t EBR-II allocation. Plays out if the NRC docket slips or HALEU enrichment doesn't scale. Model: not in the value range — no model

Valuation & balance sheet

Metric (definition) Current Own history (range or 5y avg) Peers Source, as-of
Price / book (stockanalysis) 2.03x n/a (unverified) SMR 1.54x · NNE 1.36x · XE 2.28x stockanalysis.com, 2026-10-02
EV (stockanalysis; cash excludes non-current securities) $4.21bn 52-week price change −69% SMR $2.26bn · NNE $0.26bn · XE $4.24bn stockanalysis.com, 2026-10-02
EV incl. all marketable securities (own calc.) ~$3.67bn ($6.67bn − $3,006mm liquidity + ~$1mm debt) — — 10-Q 2026-06-30; price 2026-10-02
Liquidity ÷ 2026 guided cash use (op + PP&E) ~4.6–5.8 years — — 10-Q; Q2 call guidance

Model-implied value range: none — Oklo has $1.2mm of H1 revenue and negative EBITDA, so no sector module fits; a five-year DCF would be entirely assumption. The rating rests instead on what the ~$3.7bn of non-cash value requires: at the BNEF cost estimate, that sum is roughly the build cost of ~10 Aurora units (~0.7 GW) whose economics are near breakeven at $90/MWh, before the further equity needed to build them. The market price sits on the bull side of what contracted, licensed evidence supports.

Balance sheet: not meaningful (net cash ~$3.0bn; long-term debt $0.7mm; $25.0mm right-of-first-refusal liability). The share count since 2026-08-04 is n/a (unverified): the 2026 ATM was exhausted and a new one launched after that date, so the 186.0mm figure understates current shares.

Model note: no model — pre-revenue, no module fits. The per-unit economics in Bear 1 are my estimate (est.) from a third-party build cost (BNEF via The Motley Fool, 2026-04-30) and a 2024 company price range, not from filings.

Scenario stress test

Reasoned from the bull/bear drivers above. No model, so no dollar column.

Scenario Effect Mechanism Magnitude Model Δ value vs Base ($/sh)
S1 Fast equity crash − 18% short interest, high-beta momentum holder base; ATM issuance becomes more dilutive at lower prices High —
S2 Slow bear / recession − Data-center capex cuts slow PPA conversion; equity funding window narrows High —
S3 Rapid rate shock − Long-duration, zero-cash-flow equity; higher discount rate on 2030s cash flows and dearer project debt High —
S4 Slow rate grind − Same channel as S3, more gradual; raises the hurdle on already-thin unit economics Med —
S5 Soft-landing cuts + Lower discount rate and open equity window; hyperscaler capex continues Med —
S6 Recession-driven cuts ± Lower rates help duration, but demand and funding for data centers weaken Low —
S7a Credit liquidity shock − Funding risk: project debt and equity markets close at once Med —
S7b Slow default cycle − Weaker counterparties; harder project finance for first-of-a-kind plants Low —
S8 Stagflation − Construction and fuel cost inflation plus higher rates hit first-of-a-kind capex Med —
S9a Dollar spike 0 no material effect, not modeled Low —
S9b Dollar slide 0 no material effect, not modeled Low —
S10 Melt-up + Speculative, high-short-interest name; also cheaper equity funding High —
S11 Energy supply shock + Energy-security bid for nuclear; firm-power prices rise Med —
S12 Mega-cap/AI derating − Thesis rests on AI data-center power demand; hyperscaler capex cuts hit the pipeline and the sentiment High —

Currently active/on watch per the playbook: S3 partially active; S8, S10, S11 on watch.

What would change the call

Upgrades if: the Switch or Meta agreements become binding PPAs with disclosed pricing at or above ~$90/MWh; the NRC dockets a commercial COLA; Aurora-INL cost disclosure comes in well below ~$350mm per unit; or the price falls to where non-cash value is a small multiple of contracted economics. Downgrades if: n/a below Sell; conviction rises if Aurora-INL slips past 2028, HALEU supply stays unsecured, or issuance continues without binding contracts.

Watch items

Sources