The Signal — Hyperscalers (Google, Amazon, Meta) are now actively securing uranium and nuclear fuel as NVIDIA CEO signals AI will need 1000x current energy. A 2.3B lb uranium deficit by 2045 meets inelastic demand: the squeeze is structural, not cyclical.
Consensus: Bullish | Conviction: High
What's Moving
- $LEU, $UEC — Primary uranium plays; hyperscalers cited as new market entrants on earnings calls. Supply tightness confirmed. (via @derekquick1, @unomasreactor)
- $SMR — Only NRC-approved small modular reactor builder; $94M DOE deployment awards signal regulatory tailwind. License moat + AI power demand = multiplier. (via @derekquick1)
- $BWXT, $CEG — Supply chain beneficiaries (reactor manufacturing, fuel fabrication, operators). CEG owns 21 reactors—largest in US. (via @unomasreactor)
- Baseload energy bottleneck — Phase 3 of AI infrastructure race: chips repriced (Phase 1), data centers repriced (Phase 2), fuel/nuclear still undervalued. (via @derekquick1)
Blind Spot — Everyone's focused on uranium scarcity, but conversion and enrichment capacity are equally constrained and less discussed. A 2.3B lb deficit doesn't matter if the conversion and fuel fabrication pipeline can't scale fast enough. Also: national defense angle (microreactors for grid resilience) is adding demand orthogonal to AI, but equity markets haven't priced dual-use tailwinds. Finally, regulatory risk is real—NRC licensing bottlenecks could throttle deployment despite political will.
One Actionable Idea — Long uranium exposure ($LEU, $UEC) + track DOE NELP funding announcements and NRC licensing timelines; short-term volatility likely, but multi-year structural supply deficit makes this a conviction hold into 2028–2030.
Sources: @derekquick1 (uranium squeeze, $SMR licensing), @unomasreactor (hyperscaler demand, supply chain), @govnuclear (regulatory momentum)