AI Power & Nuclear — Oklo, Nano Nuclear, Constellation

Forward-looking commentary for a general audience, not personalised advice. Prices as-of US close 4 Sep 2026 unless noted; research as-of 7 Sep 2026. Horizon: 2–5 years.

Lead

The market is pricing “AI needs nuclear” as if every press release is a power plant. The tension is simpler: hyperscalers will buy firm carbon-free megawatts — but only a few issuers can deliver them inside this decade on assets that already exist. Everything else is a licensing and financing option with a customer logo attached.

The setup / the cycle

IEA Energy and AI (Base Case): global data-centre electricity ~415 TWh in 2024 (~1.5% of world electricity), rising toward ~945 TWh by 2030 (~3%), with the US the largest share of the increase (IEA report series).

Near-term AI load still mostly meets gas, renewables, and existing grids. Nuclear enters as firm, clean attributes — either via restart / life-extension / uprate PPAs on operating LWRs, or via advanced reactors / SMRs whose first commercial Western fleets are widely described as late-2020s / ~2030+.

Policy backdrop (2024–2026): ADVANCE Act; EO 14300 (May 2025); NRC Part 53 finalized Mar 2026; proposed Part 57 for microreactors (2026 comment process). HALEU / fuel-chain remains a disclosed bottleneck for many non-LWR designs (Reuters, Mar 2025).

Secondary mid-2026 trackers cite hyperscaler nuclear commitments on the order of ~9.8 GW across disclosed deals — useful as a map, not as a substitute for reading whether each line is a binding PPA or an MOU.

The business in plain English

Quality scorecard

Comparability warning: OKLO and NNE are loss-making / near-zero reactor revenue — P/S and EV/Rev are not economic multiples. CEG is an operating generator; use earnings and EV/EBITDA. Prices 4 Sep 2026 close.

Oklo (OKLO)

• Price $41.27; mkt cap ~$7.7B; EV ~$5.0B; shares out 186.0M (Yahoo).
• Q2’26 revenue $1.21M (first disclosed product/services); TTM NI -$152.8M; TTM EPS -$0.94; OCF (ttm) -$116.9M; levered FCF (ttm) -$239.3M.
• Liquidity: Yahoo cash $2.47B / 10-Q cash+marketable securities $3.01B (period ended 2026-06-30); debt de minimis. 2026 cash opex use guided $120–150M; PP&E $400–500M (secondary 10-Q coverage).
• Dilution: avg shares 70M (2022–23) → 146M (2025) → 164M TTM; H1’26 ATM 23.1M shares / ~$1.85B net; new $1.0B ATM (May 2026).
• Aurora-INL: groundbreaking 2025-09-22; target completion 2028; DOE pilot path. Meta Ohio: up to 1.2 GW; Meta prepay / early funding (announced 2026-01-09); Phase 1 target ~2030, full buildout target 2034 — not disclosed as a classic operating-plant $/MWh PPA like CEG–Meta Clinton.
• Yahoo consensus 1y target avg $79.88 (range $14–$130); consensus EPS still deeply negative.

Nano Nuclear (NNE)

• Price $17.72; mkt cap ~$0.95B; EV ~$0.37B; shares out 53.7M.
• TTM revenue $214k (Yahoo); STS logistics revenue exists separately (STS CY2025 audited ~$7.1M per company — not clean reactor MWh).
• TTM NI -$33.9M; EPS -$0.69; cash ~$580–582M; debt de minimis.
• Dilution: Oct 2025 PIPE 8.49M shares / $400M gross; ATM sold 1.0M shares / ~$26.8M in quarter ended 2026-06-30; ~$373M ATM capacity remaining (10-Q summary). Short interest ~34% of float (Yahoo as of 2026-08-14).
• Licensing: NRC accepted/docketed CPA for KRONOS at University of Illinois (May 2026); company expects construction H2’27 after EA/safety eval targets in 2027.
• Commercial MW: Tillman framework (2026-08-24) non-binding, aspirational ~6 GW by 2040, with warrants/stock grant contemplated — dilutive if executed, not a PPA. Other MOUs exploratory. No disclosed binding hyperscaler PPA comparable to CEG–Microsoft/Meta.
• Yahoo consensus 1y target avg $40.83.

Constellation (CEG)

• Price $298.96; mkt cap ~$106B; EV ~$125B; shares out 354.3M.
• Q2’26 revenue $7.50B (+23% YoY); adjusted operating EPS $2.55 (+33.5% YoY); GAAP NI $513M. TTM rev $31.3B; TTM NI $3.46B; TTM EPS ~$10.2; FY26 adj. EPS guide $11.50–$12.50.
• Nuclear ops Q2: 44,160 GWh; capacity factor 93.0% (ex-Salem/STP). Cash $697M; total debt Yahoo ~$24.7B; Calpine closed 2026-01-07.
• Microsoft — Crane (TMI-1 restart): ~835–837 MW, 20-year PPA (announced 2024-09-20); restart path targeting 2027 in 2026 IR updates (remaining NRC/permit steps).
• Meta — Clinton: 1,121 MW (incl. +30 MW uprate), 20-year PPA from Jun 2027.
• Q2’26: additional 920 MW of 15–20 yr clean PPAs (start 2029–2032), incl. Walmart / Dresden uprate.
• Trailing P/E ~28–29x; forward ~21x; EV/EBITDA ~15.2x. Yahoo consensus 1y target avg $348.30. Share repurchases (~$2.2B YTD cited) vs dilutive ATMs at the developers.

The differentiated insight

1. Contract quality is the whole game. CEG’s Microsoft/Meta deals are long PPAs on restart / operating LWR megawatts. Oklo–Meta is a customer-backed development + prepay toward a 2030–34 campus. NNE’s GW figures are mostly non-binding frameworks. Same headline word (“nuclear for AI”); different cash-flow date.

2. Liquidity is not the same as runway to fleet. OKLO and NNE are cash-rich after equity raises — and that cash is the cost of FOAK CapEx and dilution, not proof of a completed power business.

3. NuScale’s SDA is the regulatory benchmark the market underweights when it treats all SMR tickers as interchangeable. Oklo’s DOE pilot construction narrative and NNE’s CPA docketing are real steps — neither is NuScale’s US460 Standard Design Approval (May 2025), and Oklo’s HALEU/recycle path adds fuel-chain risk NuScale’s LEU pitch tries to avoid.

4. Inside 2–5 years, CEG can compound contracted MWh; OKLO/NNE mostly cannot. Disclosed schedules put material Aurora / KRONOS fleet revenue outside the early part of that window.

What is priced in

Base. Hyperscalers keep signing firm clean PPAs with operators; SMR/advanced names raise equity and hit intermediate NRC milestones without commercial fleet revenue this side of 2028–30; gas/renewables still clear most incremental AI load.

Bull. Crane and Clinton execute on time; more GW-scale PPAs migrate to nuclear operators; Oklo INL hits 2028 and Meta Ohio converts prepay into a priced offtake; NNE converts CPA → construction → paying campus load; Part 53/57 actually shortens calendars; HALEU bottlenecks ease.

Bear (steelmanned). Interconnection and gas peakers absorb AI load cheaper/faster; Crane slips; SMR FOAK costs blow out; HALEU stays scarce; OKLO/NNE keep diluting into ATMs while LOIs fail to become PPAs; short interest (especially NNE) becomes a volatility amplifier, not a thesis.

What would change our mind

Breaks: Hyperscaler CapEx pause on power; Crane/Clinton schedule breaks; Oklo INL slips materially past 2028 without a clear path; NNE’s frameworks stay non-binding past CPA milestones; serial dilutive raises without offtake conversion.

Re-expresses (doesn’t kill the scarcity thesis): Share rotates from developers to operators (or the reverse) if one Oklo campus PPA is fully priced while CEG’s multiple compresses — that changes which name, not whether AI needs firm power.

Relative ranking (by durability of cash flows inside 2–5 years)

1. CEG — operating nuclear + binding hyperscaler PPAs
2. OKLO — funded FOAK + Meta-backed campus optionality (still pre-fleet)
3. NNE — CPA progress + liquidity, weakest commercial MW quality

Ranking by asymmetric upside if FOAK works can invert 2 and 3 — that is a different question than durability.

What we are watching

1. Crane (TMI-1) remaining approvals and 2027 restart path
2. Meta–Clinton start (Jun 2027) and further CEG PPA GW disclosures
3. Oklo INL construction milestones vs 2028 target; any priced $/MWh disclosure on Meta Ohio
4. NNE EA/safety eval (2027 targets) and any binding offtake (not MOUs)
5. HALEU / fuel-chain headlines for non-LWR designs
6. Dilution: OKLO/NNE ATM/PIPE usage vs CEG buybacks
7. NuScale commercial progress as the SMR regulatory yardstick

Bottom line

AI power scarcity is real; nuclear that already spins turbines is scarce in a different way than nuclear that still needs a license. Inside a 2–5 year horizon, CEG is the cash-flowing expression of firm clean megawatts under binding PPAs; OKLO is a well-funded FOAK/build-own-operate option with a Meta logo and heavy dilution; NNE is smaller, CPA-progressing, and still mostly non-binding on commercial GW. We would rather underwrite contracted operating megawatts than LOI slideware — and we would size developer names as options that can expire against NRC calendars and equity raises.

Primary sources

• IEA Energy and AI executive summary / demand chapters
• Yahoo Finance quotes & key-stats for OKLO, NNE, CEG (close 4 Sep 2026)
• Oklo IR / Meta Ohio PR (2026-01-09); Oklo Overview Fact Sheet (Jul 2026); NRC Oklo pre-application pages
• Nano Nuclear Q3 FY26 IR (2026-08-12); PIPE PR; Tillman framework (2026-08-24)
• Constellation Q2’26 results (2026-08-06); Microsoft Crane PR (2024-09-20); Meta Clinton PR (2025); Calpine close (2026-01-07)
• NRC Part 53 highlights; NuScale US460 SDA (May 2025); Reuters HALEU supply chain (Mar 2025)