The full SMR and nuclear investment universe — $OKLO and $SMR developers, $BWXT and $UUUU fuel-cycle picks-and-shovels, and $CEG, $VST, $TLN fleet operators monetizing $MSFT, $AMZN, and $META AI-data-center PPAs — plus how the basket compares to URA, URNM, and NLR.
More Macroplane supply-chain guides · Baskets · Investment theses · Macro trends
SMR stocks are publicly traded companies building, fueling, or operating small modular reactors — sub-300 MW nuclear reactors that are factory-built, shipped to site, and designed to power AI data centers, industrial loads, and grid-stressed regions far faster than gigawatt-scale conventional plants. The investable universe spans three layers: pure-play developers like $OKLO and $SMR, component and HALEU fuel suppliers like $BWXT and $UUUU, and the nuclear fleet operators ($CEG, $VST, $TLN) signing multi-gigawatt power purchase agreements with hyperscalers right now.
This guide walks through every public name worth tracking, where each one sits in the nuclear value chain, and why the trade has gone from forgotten utility play to one of the highest-conviction AI-adjacent themes of 2026.
👉 See live prices and performance: Power & Grid basket on Macroplane — tracking the nuclear, grid, and power-infrastructure names that benefit as AI compute demand collides with a constrained electricity supply.
These are the names that matter, grouped by where they sit in the supply chain. Read the Nuclear Renaissance trend page for the broader macro frame.
$OKLO — Oklo Inc. The most-watched pre-revenue name in the space. Oklo is building the Aurora — a 15-75 MW liquid-metal-cooled fast reactor designed to run on recycled fuel, with the first commercial unit targeted for Idaho National Laboratory. The pitch is a fundamentally different architecture: smaller blocks, longer refueling cycles, and a build-own-operate model where Oklo sells power under long-dated PPAs rather than selling reactors to utilities. Sam Altman chaired the company through its SPAC merger, which makes it the de facto AI-adjacent SMR proxy. Risk is binary — the NRC combined license application is still under review.
$SMR — NuScale Power Corp The only SMR developer with an NRC-approved design (Standard Design Approval for the 50 MW VOYGR module was issued in 2023, with the upsized 77 MW variant under review). NuScale's reactor is a light-water pressurized SMR — closer to existing reactor physics, which is the regulatory advantage. The company is now focused on first deployments outside the US (Romania, Poland) after the UAMPS Carbon Free Power Project was cancelled in 2023. The stock trades as the most de-risked SMR design story; execution and balance sheet are the open questions.
$BWXT — BWX Technologies, Inc. The single most under-appreciated nuclear infrastructure name. BWXT is the prime contractor that builds nuclear reactors for every US Navy submarine and aircraft carrier — giving it the only operating heavy-forging and nuclear-grade fabrication footprint at scale in North America. As SMRs commercialize, BWXT supplies pressure vessels, reactor components, and TRISO fuel through its Lynchburg, Virginia facility. It also owns the only NRC-licensed commercial HALEU downblending capability in the US. The thesis is "picks and shovels for every SMR design" — BWXT wins regardless of which developer takes share.
$UUUU — Energy Fuels Inc Primarily known as a US uranium producer at its White Mesa, Utah mill. Energy Fuels is one of the few Western producers positioned to feed the HALEU supply chain that every advanced SMR design ($OKLO, X-energy, Kairos, TerraPower) requires. As US Department of Energy HALEU procurement contracts ramp toward multi-billion-dollar programs, $UUUU sits in the small group of Western suppliers eligible to bid.
$CCJ — Cameco Corp The Western world's largest pure-play uranium producer. Cigar Lake and McArthur River in Saskatchewan provide the lowest-cost, highest-grade production outside Kazakhstan, and Cameco's 49% stake in Westinghouse (alongside Brookfield) makes it a direct beneficiary of the reactor-services backlog. Cameco is the default "if uranium prices go up, this stock goes up" exposure — and spot has re-rated meaningfully on contracted reactor restarts plus utility long-term contracting at multi-decade highs.
$UEC — Uranium Energy Corp The largest pure-play US uranium producer, with in-situ recovery operations in Texas and Wyoming, plus the Roughrider deposit in Saskatchewan from the Uranium One acquisition. UEC's pitch is exposure specifically to US-sourced uranium at a moment when the Russian import ban and DoE strategic stockpiling converge on domestic-supply premium pricing.
$CEG — Constellation Energy Corp The largest US nuclear fleet operator, with roughly 21 GW of nuclear capacity. The 2024 announcement that Constellation would restart Three Mile Island Unit 1 under a 20-year PPA with $MSFT for AI data center power was the single event that re-rated the entire nuclear complex. The deal sets the template: a hyperscaler pays a meaningful capacity premium for clean 24/7 baseload, and the operator commits to long-dated PPA terms previously unheard of in deregulated power markets. CEG is the bellwether — when AI-power headlines hit, this is the first stock to move.
$VST — Vistra Corp. Combined nuclear (Comanche Peak), gas, and renewables fleet, plus the 2023 Energy Harbor acquisition that added the Perry, Beaver Valley, and Davis-Besse nuclear plants. Vistra has been signing data-center PPAs across its fleet and is one of the few names with the geographic footprint (Texas + PJM) and asset mix to capture data-center load growth on both nuclear and gas peaker sides. The $AMZN data-center capacity announcement and follow-on hyperscaler activity have repriced VST as much an AI-power play as a Texas merchant generator.
$TLN — Talen Energy Corp Owner of the Susquehanna nuclear plant in Pennsylvania, which signed the most aggressive data-center PPA in the space — the $AMZN AWS deal to co-locate a data center directly adjacent to the plant and draw up to 960 MW of nuclear baseload. Regulatory pushback at FERC on behind-the-meter co-location structure injected uncertainty in 2024-2025, but Talen remains the cleanest single-asset exposure to AI baseload demand, with the smallest float of the three operators.
Three forces are driving the trade right now.
$MSFT, $AMZN, $GOOGL, and $META have collectively committed tens of gigawatts of new data-center load this decade, and their net-zero commitments preclude buying merchant gas-fired power. Wind and solar with batteries do not deliver the 24/7 firm-load profile AI training clusters require. Nuclear is the only operating, dispatchable, zero-carbon source at scale — which is why the Three Mile Island restart ($CEG-$MSFT), Susquehanna co-location ($TLN-$AMZN), and Vistra PPAs ($VST) all priced inside an 18-month window. The structural picture: every gigawatt of incremental US data center load is, by elimination, a bid for existing nuclear capacity or for first-of-a-kind SMR deployment.
The NRC issued the Standard Design Approval for the NuScale 50 MW module in 2023 — the first SMR design approval in US history. Oklo's combined license application is now under active review. The 2024 ADVANCE Act lowered NRC fees, set hard timelines for advanced reactor reviews, and explicitly authorized faster licensing for sites that have already hosted reactors. The Department of Energy committed $900M to support SMR deployment, plus multi-billion-dollar HALEU procurement programs. Regulatory friction is not gone, but the slope of the curve has changed.
Every advanced SMR design under serious commercial development — Oklo's Aurora, X-energy's Xe-100, Kairos's KP-FHR, TerraPower's Natrium — uses HALEU (uranium enriched between 5% and 20% U-235, above conventional LWR fuel). HALEU does not exist at commercial scale in the West today; the only producing facility was Russia's TENEX, which is now sanctioned and import-restricted. That makes $BWXT (downblending), Centrus Energy (cascading), $UUUU (front-end), and $CCJ (raw uranium supply) into critical-path infrastructure. The DoE HALEU Availability Program is committing real dollars; the supply chain is being rebuilt from zero.
Most retail investors think "SMR = developer." That's wrong, and the framing matters for picking stocks.
The nuclear value chain has five stages, and each stage has very different economics, time-to-revenue, and competitive structure.
| Stage | What happens | Time to revenue | Public companies |
|---|---|---|---|
| 1. Raw uranium mining | Pull U3O8 ore out of the ground or recover via ISR | Producing today | $CCJ, $UEC, $UUUU, Kazatomprom |
| 2. Conversion & enrichment | Convert U3O8 to UF6, then enrich to LEU or HALEU | 2-5 yrs to scale | Centrus, Urenco (private), $CCJ (via Westinghouse) |
| 3. Fuel fabrication | Assemble enriched uranium into fuel pellets, rods, or TRISO particles | 3-5 yrs for HALEU | $BWXT (TRISO + HALEU downblend), Westinghouse ($CCJ-owned), Framatome |
| 4. Reactor design + components | Engineer SMR, fabricate pressure vessels, supply nuclear-grade forgings | 5-10 yrs to first power | $OKLO, $SMR, $BWXT (components), GE Hitachi (private), Holtec (private) |
| 5. Power generation + PPAs | Operate the reactor, sell electricity under long-dated contracts | Producing today | $CEG, $VST, $TLN, Duke, Southern Co, Dominion |
Time-to-revenue matters more than is usually appreciated. Stage 5 names ($CEG, $VST, $TLN) are already monetizing AI-data-center demand on existing reactors that took 10-15 years to build and are now paid off — earnings show up in current quarters. Stage 1 names ($CCJ, $UEC, $UUUU) benefit from spot uranium repricing now. Stage 3 ($BWXT) is a multi-year ramp tied to HALEU program execution. Stage 4 pure-plays ($OKLO, $SMR) carry the highest long-dated upside and the highest binary regulatory and execution risk.
The cleanest framing: if you want exposure to "AI needs nuclear" you own the operators ($CEG, $VST, $TLN). If you want exposure to "SMR will eventually deploy" you own the picks-and-shovels ($BWXT, $UUUU, $CCJ). If you want optionality on first-of-a-kind technology success, you own the developers ($OKLO, $SMR).
The three major nuclear-themed ETFs all approach the trade differently. Here's how to think about them versus building the basket yourself.
| URA (Global X Uranium) | URNM (Sprott Uranium Miners) | NLR (VanEck Uranium+Nuclear) | Macroplane Power & Grid basket | |
|---|---|---|---|---|
| Focus | Uranium miners + some fuel cycle | Pure-play uranium miners | Mixed: utilities, miners, services | AI-power thesis: SMR + operators + fuel + grid |
| Top holdings | $CCJ, Kazatomprom, NexGen | $CCJ, Kazatomprom, Paladin | $CEG, $VST, Public Service, $CCJ | $CEG, $VST, $TLN, $OKLO, $SMR, $BWXT, $CCJ |
| SMR developer exposure | Minimal | None | Minimal | Direct: $OKLO + $SMR |
| AI-PPA fleet exposure | None | None | Significant ($CEG, $VST) | Designed around it |
| Expense ratio | 0.69% | 0.85% | 0.61% | Free to track on Macroplane |
URA and URNM are uranium-price plays. They will move with U3O8 spot prices and Kazakh production headlines, but they have almost no direct AI-data-center linkage. NLR is closer to the right thesis exposure because it includes operators, but it's still light on SMR developers and component suppliers. If your specific thesis is "AI compute is bidding for nuclear-grade clean baseload, and the supply chain rebuild has multi-decade momentum," none of the off-the-shelf ETFs cleanly express it.
👉 Build it your way: clone the Power & Grid basket on Macroplane, swap constituents across developers / fleet / fuel-chain, set weightings — see live performance instantly.
SMRs are one leg of the Nuclear Renaissance macro trend and a critical input to the broader AI-power-build story. The full trade also includes existing nuclear restarts (Palisades, Three Mile Island), grid transmission build-out, gas peaker capacity, behind-the-meter battery storage, and the electrical infrastructure names ($ETN, $GEV, $VRT) supplying the data centers themselves. The SMR-specific story is the highest-conviction, longest-dated leg — power that doesn't exist yet, for compute that does.
If you're building an AI-power portfolio rather than a pure nuclear portfolio, pair the names above with electrical equipment ($ETN, $GEV, $PWR), data center cooling and infrastructure ($VRT, Modine), and natural gas peakers. The Power & Grid basket maintains the broader watchlist.
The credible universe splits into three groups. Pure-play SMR developers: $OKLO and $SMR. Nuclear component and fuel-cycle picks-and-shovels: $BWXT and $UUUU. Fleet operators monetizing AI-data-center PPAs today: $CEG, $VST, and $TLN. There is no single "best" pick — SMR deployment timelines and regulatory milestones are inherently lumpy and binary, so diversifying across developer optionality, fuel-chain infrastructure, and revenue-producing operators is the standard approach.
No commercial SMR is currently operating in the US. NuScale's design has NRC approval but first deployments are not yet under construction. Oklo's combined license application is still under review. First US commercial SMR power generation is most likely 2028-2030 depending on developer. That's why the trade today is split: existing operators ($CEG, $VST, $TLN) capture AI-data-center PPA value now on their installed fleets, while pure-play developers ($OKLO, $SMR) trade on long-dated technology and regulatory outcomes.
AI training and inference clusters require 24/7 firm baseload. Wind and solar with batteries do not currently deliver that load profile at cost-competitive levels at the multi-gigawatt scale hyperscalers need. Nuclear is the only operating, dispatchable, zero-carbon source — which matches $MSFT, $GOOGL, $AMZN, and $META net-zero commitments. Every available megawatt of US nuclear capacity is effectively spoken for, and incremental data-center growth is forcing the conversation about new build, restart, and SMR deployment.
HALEU is high-assay low-enriched uranium — enriched between 5% and 20% U-235 (conventional reactor fuel is under 5%). Every advanced SMR design under serious development uses HALEU because it enables smaller cores, longer refueling cycles, and higher fuel efficiency. The problem: HALEU is not currently produced at commercial scale outside Russia, which is now sanctioned. The DoE is funding the rebuild of a domestic supply chain — $BWXT (downblending), Centrus Energy (cascading), $UUUU, and $CCJ are critical-path infrastructure. The fuel constraint is, today, the bigger near-term execution risk than reactor design.
The Power & Grid basket on Macroplane shows daily and longer-term performance for each constituent, plus links to per-company supply-chain maps, news, and customer relationships. The Nuclear Renaissance trend page widens the lens to the broader policy, deployment, and fuel-cycle picture.
The credible universe splits into three groups. Pure-play SMR developers: $OKLO and $SMR. Nuclear component and fuel-cycle picks-and-shovels: $BWXT and $UUUU. Fleet operators monetizing AI-data-center PPAs today: $CEG, $VST, and $TLN. There is no single "best" pick — SMR deployment timelines and regulatory milestones are inherently lumpy and binary, so diversifying across developer optionality, fuel-chain infrastructure, and revenue-producing operators is the standard approach.
No commercial SMR is currently operating in the US. NuScale's design has NRC approval but first deployments are not yet under construction. Oklo's combined license application is still under review. First US commercial SMR power generation is most likely 2028-2030 depending on developer. That's why the trade today is split: existing operators ($CEG, $VST, $TLN) capture AI-data-center PPA value now on their installed fleets, while pure-play developers ($OKLO, $SMR) trade on long-dated technology and regulatory outcomes.
AI training and inference clusters require 24/7 firm baseload. Wind and solar with batteries do not currently deliver that load profile at cost-competitive levels at the multi-gigawatt scale hyperscalers need. Nuclear is the only operating, dispatchable, zero-carbon source — which matches $MSFT, $GOOGL, $AMZN, and $META net-zero commitments. Every available megawatt of US nuclear capacity is effectively spoken for, and incremental data-center growth is forcing the conversation about new build, restart, and SMR deployment.
HALEU is high-assay low-enriched uranium — enriched between 5% and 20% U-235 (conventional reactor fuel is under 5%). Every advanced SMR design under serious development uses HALEU because it enables smaller cores, longer refueling cycles, and higher fuel efficiency. The problem: HALEU is not currently produced at commercial scale outside Russia, which is now sanctioned. The DoE is funding the rebuild of a domestic supply chain — $BWXT (downblending), Centrus Energy (cascading), $UUUU, and $CCJ are critical-path infrastructure. The fuel constraint is, today, the bigger near-term execution risk than reactor design.
The Power & Grid basket on Macroplane shows daily and longer-term performance for each constituent, plus links to per-company supply-chain maps, news, and customer relationships. The Nuclear Renaissance trend page widens the lens to the broader policy, deployment, and fuel-cycle picture.