Rising baseload power demand, led by AI data centers, is colliding with a uranium mining industry that spent a decade underinvesting. We map the demand evidence, the supply gap, and four names across the fuel chain.
Spot uranium has climbed to roughly $89/lb, the highest level since early February, following five months range-bound between $84 and $87.
That is the second leg of a move that already ran further once this year — spot futures briefly cleared $100/lb in January, the highest print since February 2024, before cooling back into range. The long-term contract market, where utilities actually secure multi-year supply, has been steadier: TradeTech's Long-Term Price Indicator reached $97/lb by end-June, up $10 since December and the highest level in more than 18 years.
Three independent data sources point to the same structural imbalance.
Goldman Sachs projects global data center power demand will rise 160–175% by 2030 versus 2023 levels. The IEA's Electricity 2026 Outlook separately shows nuclear generation at a record high, with nuclear and renewables together supplying roughly half of global electricity by 2030.
Global uranium production in 2025 was approximately 173 million pounds against primary demand of roughly 204 million pounds, a gap presently bridged by drawing down finite secondary stockpiles. A survey of over 600 investors found a majority expect mined supply to cover less than 75% of future reactor requirements.
Nuclear fission carries an energy density nothing else matches: 1 kilogram of enriched uranium-235 delivers roughly the energy of 10,000 kg of oil or 14,000 kg of coal. It is the only source that is simultaneously carbon-free, always-on, and deployable at grid scale.
"The demand case is not really an AI story. It is a physics story that AI happens to be accelerating." — Analyst synthesis
Even credible skeptics do not dispute the demand math; they dispute the timeline. JPMorgan Asset Management's own research notes that nuclear projects average nine years from groundbreaking to commercial operation, and that the last US plant built came in at $30bn — double its budget, seven years late. The bear case is a construction-speed argument, not a demand argument.
First time since Feb 2024, on accelerated institutional physical accumulation.
Long-term contract price keeps climbing regardless — a split between speculative and structural buying.
TradeTech indicator reaches $97/lb as utilities lock in multi-year supply.
Microsoft's $16B, 20-year Three Mile Island restart with Constellation; Meta's 1.1GW Clinton deal; Amazon's 1.9GW Talen/Susquehanna PPA through 2042.
A fresh six-month high, against a still-unresolved supply deficit.
The thesis is not one stock — it is four positions on one supply chain, each with a distinct risk profile.
| Ticker | Company | Chain position | Thesis |
|---|---|---|---|
| CCJ | Cameco | Mining & fuel | $46.8B cap. The only name with contracted revenue at scale. Direct exposure to the scarce input itself. |
| OKLO | Oklo Inc. | Reactor build | $7.2B cap, largest of the pre-revenue SMR developers by market cap. Down ~79% from its high — high risk, long duration. |
| BWXT | BWX Technologies | Components | $14.0B cap. ~77% of Q2 2025 revenue came from Government Operations (naval propulsion, special materials, defense uranium enrichment) — insulated from civilian-cycle swings. Pays a dividend. |
| CEG | Constellation Energy | Generation | $100.4B cap, largest US nuclear fleet. Counterparty on Microsoft's $16B TMI deal and Meta's 1.1GW Clinton agreement — revenue signed, not projected. |
For illustration, not a recommendation on any individual name beyond the four featured above: the broader set of publicly traded names FMP surfaces across the chain, grouped by position. Market caps as of Aug 27, 2026.
| Ticker | Company | Mkt cap |
|---|---|---|
| Mining & fuel supply | ||
| CCJ | Cameco Corp Featured | $46.8B |
| NXE | NexGen Energy | $7.3B |
| UEC | Uranium Energy Corp | $6.5B |
| UUUU | Energy Fuels | $3.9B |
| LEU | Centrus Energy | $3.6B |
| DNN | Denison Mines | $3.3B |
| ISOU | Isoenergy Ltd | $0.8B |
| UROY | Uranium Royalty Corp | $0.6B |
| URG | Ur-Energy | $0.6B |
| Reactor design & build | ||
| OKLO | Oklo Inc Featured | $7.2B |
| SMR | NuScale Power | $2.8B |
| NNE | Nano Nuclear Energy | $1.0B |
| FISN | Deep Fission Inc | $0.4B |
| Components & services | ||
| BWXT | BWX Technologies Featured | $14.0B |
| CW | Curtiss-Wright | $22.9B |
| HII | Huntington Ingalls | $11.7B |
| Power generation | ||
| CEG | Constellation Energy Featured | $100.4B |
| VST | Vistra Corp | $47.2B |
| NRG | NRG Energy | $24.5B |
| TLN | Talen Energy | $13.9B |
| Physical uranium (fund NAV, not mkt cap) | ||
| SRUUF | Sprott Physical Uranium Trust | $6.8B NAV |
This extended list is provided for mapping the sector, not as a ranked recommendation — depth of coverage and diligence varies widely by name, and several are pre-revenue or micro-cap with materially higher risk than the four featured above.
JPMorgan's own data shows a nine-year average build time and a track record of major cost overruns. New reactor supply cannot arrive fast enough to matter this decade — if AI power demand cools or efficiency gains reduce per-query energy cost faster than expected, the demand side of this thesis softens.
A meaningful share of recent price support is financial — physical uranium trusts accumulating pounds rather than utilities burning them. Fund-driven buying can reverse on a sentiment shift, removing a real source of current demand.
Pre-revenue reactor developers (OKLO) carry execution risk on top of commodity risk. This basket's risk is not uniform — it ranges from BWXT's government-anchored stability to OKLO's binary, multi-year execution bet.
Supply — production below consumption in 2025 (173M vs. 204M lbs); multi-year mine lead times.
Demand — WNA reference case shows demand more than doubling by 2040.
Pricing — spot at a 6-month high; long-term contracts at an 18-year high.
Catalyst — hyperscalers signing 20-year nuclear PPAs, a structurally new demand source.
Key risk — construction timelines, not resource scarcity, are the binding constraint on the bull case.
Uranium was underpriced for a decade because nobody needed it built faster than mines could recover. A well-capitalized new buyer — hyperscale AI operators signing 20-year power contracts — has changed that. The construction bottleneck is real, but it argues for the existing fuel supply chain being scarce and valuable, not for the thesis being wrong.
Get real-time discussion on setups like this one — sector rotations, supply-chain mapping, and position sizing — from a community of active traders.
Join Insight Invest →Note: the January 2026 spot price spike is reported as $99–$101/lb depending on data provider and exact date; we use $100/lb as a rounded reference point. Energy-density figures in Section 03 are order-of-magnitude accurate per standard nuclear engineering references but the specific multipliers cited come from a secondary popular-science source, not a peer-reviewed one.