medium convictionactive · updated 2026-09-25T00:00:00.000Z
AI DC load growth → PJM capacity auction 10× → nuclear baseload structural premium → CEG/CCJ/UEC benefit
AI data center load is forcing grid capacity prices to clear at extraordinary premiums: PJM capacity cleared at $329.17/MW-day for 2026/27 vs $28.92/MW-day for 2024/25 — a 10× increase, with data centers representing 63% of the load growth. Nuclear plants — the only dispatchable, 24/7 carbon-free baseload — are the primary beneficiaries. Uranium producers (CCJ, UEC) benefit from the derived demand; nuclear operators (CEG) benefit from capacity payment premiums and hyperscaler PPA demand.
The chain
1
AI data center power demand is creating acute grid capacity shortfalls. PJM capacity cleared at $329.17/MW-day for delivery year 2026/27, up from $28.92/MW-day for 2024/25 — a 10× increase. Data centers represent 63% of the load growth driving the increase. Grid operators are passing $9.3B in cost recovery back to load-serving entities.
From 2026-05-30-autoresearch-energy-critical-minerals-uranium-helium-copper-nuclear: "PJM capacity prices 10x: $28.92/MW-day (2024/25) → $329.17/MW-day (2026/27); data centers 63% of increase; $9.3B cost recovery"
robin-millican in 2026-06-30-podcast-columbia-energy-exchange-doug-arent-and-robin-millican-on-what-s-really: "PJM is a special situation where they have a capacity market that's... bit dysfunctional. They also are supply constrained. It's really hard for new projects to come online." Plus the cost side: "an estimated $4.3 billion in transmission projects were approved in 2024 that were associated with data center load. And there are another 2.5 billion under consideration." — Independent (CGEP) confirmation that PJM specifically is the supply-constrained region where DC load is inflationary. **Regional caveat**: the same CGEP work finds load growth is *not* a national price driver and can lower prices in supply-rich regions (see electricity-price-drivers-decomposition) — this chain's forcing function is PJM-specific by construction, and generalizing it nationally would overstate it. Arent adds the market-design point that PJM's anticipated load "will need new generation... the anticipation has marginal costs greater than the average cost today, and therefore it will be inflationary in that region."
From 2026-09-19-autoresearch-pjm-bra-rm26-4-nuclear: "The 2028/29 BRA (14 July 2026) cleared at the next collar cap of $325/MW-day UCAP — 2.5% below the prior cap, still at the ceiling, and 6,831 MW short of the reliability requirement"
2
Nuclear plants are the primary capacity auction beneficiary: they are dispatchable (always on), carbon-free (meets hyperscaler CFE mandates), and have long operational lifespans at near-zero variable cost once constructed. Constellation Energy (CEG) is the largest US nuclear operator. Uranium spot (CCJ primary product) rose to $85.20/lb with long-term prices at $90/lb — highest since 2008. Sprott Physical Uranium resumed purchasing.
From 2026-05-30-autoresearch-energy-critical-minerals-uranium-helium-copper-nuclear: "uranium spot to $85.20/lb (May 28, 2-month low); CCJ 'forward demand never been bigger'; Sprott resumed purchasing"
From 2026-05-20-autoresearch-nuclear-uranium-meta-ppa-ccj-ceg-smr-may-2026: "Constellation Energy TMI restart; Microsoft 20-year PPA with CEG for nuclear power; nuclear capacity factor 92% vs 35% for wind/solar"
From 2026-09-19-autoresearch-pjm-bra-rm26-4-nuclear: "the 2028/29 8-K cleared 15,700 MW of nuclear at $325 versus 3,175 MW fossil/other"
3
New uranium supply responds: Uranium Energy Corp (UEC) commenced production at Burke Hollow (Texas ISR mine, world's newest ISR mine as of May 2026 — first new US ISR in over a decade). CCJ CEO states 'forward demand never been bigger.' Uranium supply response is constrained — new mines take years to develop — which sustains the pricing premium and benefits existing low-cost producers (CCJ, UEC ISR operations).
From 2026-05-30-autoresearch-energy-critical-minerals-uranium-helium-copper-nuclear: "UEC Burke Hollow production commenced May 2026 (world's newest ISR mine); dual ISR platform (Burke Hollow TX + Christensen Ranch WY)"
From 2026-05-30-autoresearch-energy-critical-minerals-uranium-helium-copper-nuclear: "uranium spot $85.20/lb; CCJ 'forward demand never been bigger'; LT $90/lb highest since 2008"
What would falsify this
- Step 1: PJM capacity prices collapse in 2027/28 auction — indicates AI DC load growth didn't materialize into physical grid demand. **Checked 2026-09-19 — NOT fired:** 2027/28 printed $333.44 and 2028/29 printed $325, both at the collar. From 2026-09-19-autoresearch-pjm-bra-rm26-4-nuclear.
- Step 2: Hyperscalers shift to behind-the-meter nuclear SMRs at scale — bypasses grid capacity markets, removing the CEG/uranium premium
- Step 3: Major new uranium supply (Kazakhstan, Australia) comes online unexpectedly fast, pushing spot below $65/lb — compresses CCJ/UEC margins
Contradictions / tensions
- PJM capacity prices are forward-looking (2026/27 delivery year). Actual cleared capacity may not reflect physical consumption if DC construction is delayed.
- SMR (small modular reactor) deployment could add new dispatchable nuclear capacity faster than expected — pressure on existing uranium plant operators.
- Uranium spot at $85/lb is below the ~$90+/lb needed to incentivize significant new greenfield mine development — but ISR at $30-40/lb cash cost is highly profitable at current spot.
Implications
- Primary: Cameco (CCJ, ~$112.70 as of May 30) — largest publicly traded uranium producer; captures uranium price appreciation from AI-driven nuclear demand. Analyst consensus: 'forward demand never been bigger.'
- Secondary: Uranium Energy Corp (UEC, ~$13.68, analyst target ~$18) — newly operational Burke Hollow ISR mine; low-cost ISR production at exactly the moment uranium spot cleared $85/lb.
- Tertiary: Constellation Energy (CEG) — US nuclear operator; capacity payment windfall from PJM 10× clearing; Microsoft PPA validates nuclear-as-AI-power thesis.
- Higher-order: FERC large-load interconnection ruling (end of June 2026) would further accelerate AI DC power demand absorption into the grid — compounding the PJM capacity premium for dispatchable nuclear.
- **RM26-4** (DOE-directed large-load interconnection rulemaking) is the **end-June catalyst**: FERC's April-16 "Order Regarding Intent to Act" commits to act by end of June 2026, but the *form* (final rule vs **NOPR**) is **genuinely undetermined** in every analysis fetched — a NOPR is the natural next step procedurally, but "quick, efficient, legally durable" language + DOE acceleration leave a substantive order possible. A real binary swing, not "likely a NOPR."
- **EL25-20** (separate PJM co-location proceeding, FERC Dec-2025 order) is *further along and distinct*: directs PJM to stand up **two new front-of-meter transmission services** (Firm / Non-Firm Contract Demand) while narrowing behind-the-meter (BTMG) rules to "smaller retail loads." This is the cleaner, regulator-blessed path.
- **CEG insulation:** CEO Joe Dominguez has pivoted to front-of-the-meter — *"we do not need to have load co-located or behind-the-meter… to achieve compelling pricing"* (Microsoft/TMI restart the proof point). So the BTM block that killed the rejected AWS–Talen structure is **largely not CEG's risk**. Conviction held (0.68 per 06-09); the dislocation case intact, the BTM-block fear mostly mispriced onto CEG from the AWS–Talen precedent.
Companies
Concepts
Nuclear as the only viable AI data center baseloadAI capex sprint → power-supply gap → grid-component + materials bottleneck → nuclear/copper/transformer beneficiary cascadeS-Curve Evaluation Lens
Open questions