Stock-Market Research Dispatch — 2026-05-30
Auto-generated by daily research routine. Not a source; not ingest-pending material. Summary of what changed in the wiki today.
Top of mind
CEG's nuclear fleet just received a $2.3B/year incremental capacity payment boost — and the stock is still -30% from peak. PJM's 2026/2027 capacity auction cleared at $329.17/MW-day, up from $28.92/MW-day (+10×). Data centers drove 63% of the increase, totaling $9.3B in cost recovery across the footprint. At ~21 GW of nuclear capacity in PJM, Constellation captures ~$2.3B incremental annual capacity revenue compared to prior auction pricing. The May 2026 guidance miss and TMI 2031 interconnection delay risk have pushed CEG -30% from its 52w high — but the capacity auction repricing (a different positive) was confirmed by today's research and is not yet widely attributed to CEG's forward earnings picture. The pjm-capacity-prices-to-nuclear-premium mechanism is now canonical.
MU has two dated, independent forcing functions converging in Q3 2026. SK Hynix's 6-month helium inventory buffer closes June–July 2026 (South Ras Laffan not restarting before end of summer); SK Hynix already cut HBM4 volumes 20-30% on yield issues; 80%+ of Nvidia's HBM supply is exposed. Simultaneously, the Section 232 Phase 2 Commerce report is due July 1 and is expected to target Korean-origin memory (Lutnick: "100% tariff or build in America"). Micron is the only US-domiciled HBM producer. Neither forcing function is priced in at current levels. The helium-cliff-to-hbm-supply-crunch mechanism is the cleanest causal chain opened today.
CHIPS 48D December 31, 2026 construction deadline creates a hard H2 2026 fab groundbreaking rush. No congressional extension has been introduced despite SIA + 17 trade groups pressing (May 12 letter). $640B in announced US fab investment is anchored to this deadline. A 35% ITC on a $10B fab = $3.5B incentive to start construction before year-end. PWR and FIX are the primary construction beneficiaries. New hypothesis pages filed; graduation to active concept pending further evidence.
Strongest-conviction buys
From step 4c ranking. Research output, not advice.
| # | Ticker | Mechanism | Conviction | Valuation vs base | Fundamentals (one line) | Next catalyst |
|---|---|---|---|---|---|---|
| 1 | MU | helium-cliff-to-hbm-supply-crunch + S232 | medium-high | Marginal (52w high) — new catalysts additive | 2026 HBM 100% committed; US-domicile premium; CHIPS Act backstop | SK Hynix helium inventory cliff June–July 2026 |
| 2 | CEG | pjm-capacity-prices-to-nuclear-premium | high | Yes — -30% from peak; PJM 10× not yet in forward estimates | ~$2.3B incremental annual capacity revenue from PJM $329/MW-day clearing | Next hyperscaler PPA announcement; Q2 2026 earnings capacity commentary |
| 3 | CCJ | ai-capex-to-power-and-materials-cascade step 5a | high | Yes — slightly below $116.93 entry; base case +15–20% | Q1 EPS $0.34 beat by 30%; uranium LT price 14-yr high; Sprott purchasing resumed | Meta 7.8 GW PPA volume timeline; uranium spot vs $85.20/lb |
Watchlist
Active theses with current valuation context — prices via Twelve Data, 2026-05-29.
| Ticker | Thesis | Conviction | Price | Undervalued vs base? | Next catalyst |
|---|---|---|---|---|---|
| MU | hbm-supply-bottleneck | medium-high | $971 | Marginal (near 52w high) | Helium cliff June–July; S232 July 1 |
| CEG | nuclear-baseload-for-ai-data-centers | high | $287.75 | Yes | PJM capacity revenue recognition; PPA pipeline |
| CCJ | nuclear-baseload-for-ai-data-centers | high | $112.70 | Yes (marginal) | Meta PPA volumes; uranium LT trend |
| TSM | tsmc-capacity-shortfall-and-pricing-power | high | $418.45 | No (near 52w high) | Q2 2026 earnings (July) |
| ASML | picks-and-shovels-leading-edge-fab-buildout | high | $1,612.76 | Marginal | 48D H2 2026 fab demand; KLAC/AMAT Q3 earnings |
| SCCO | copper-supercycle-ai-data-centers | medium-high | $191.30 | Marginal | FCX Q2 earnings; copper LME trajectory |
| INTC | us-fab-capacity-bottleneck | medium | $114.68 | No (flagged for exit — sector-excess −11.6%) | VLSI symposium June 14–18 |
Undervalued candidates (today)
- CEG (nuclear-baseload-for-ai-data-centers): at $287.75 vs 52w high $412.70. PJM $329/MW-day × ~21 GW nuclear capacity = ~$2.3B incremental annual capacity revenue. TMI 2031 delay and Calpine integration risk are real headwinds, but the capacity repricing is a separate positive not embedded in the -30% selloff narrative. Catalyst: explicit PJM capacity revenue commentary in Q2 2026 earnings.
- CCJ (nuclear-baseload-for-ai-data-centers): $112.70 vs $116.93 entry (−3.6%). Uranium spot $85.20/lb; Sprott Physical Uranium Trust resumed; Meta 7.8 GW PPA volume = near-term fuel demand uplift. Base case +15–20% upside is intact.
New chains to investigate (hypothesis-stage)
- skhynix-helium-section232-micron-hbm-window — MU. Graduate to active when: SK Hynix announces explicit production reduction citing helium, OR July 1 Commerce report recommends Korean HBM tariff.
- chips-48d-deadline-rush-construction-pwr-fix — PWR, FIX. Graduate to active when: H2 2026 fab groundbreaking announcements confirmed, OR PWR/FIX raise guidance citing 48D-accelerated construction starts. Hard calendar event (Dec 31, 2026) makes this time-sensitive to watch.
- domestic-critical-metals-dod-contract-pipeline — USAR, CRML, MP, UUUU. Graduate to active when: USAR Serra Verde acquisition completion confirmed Q2-Q3 2026.
- nee-dominion-virginia-scc-block-risk — NEE, D. Bear case on NEE if Virginia SCC blocks; monitor Virginia AG or consumer advocacy opposition filings.
New theses (now active)
None today — four hypotheses created (see above); none reached active graduation bar.
Updated theses (existing active)
- nuclear-baseload-for-ai-data-centers — PJM capacity auction 10× confirmed ($28.92→$329.17/MW-day; new mechanism pjm-capacity-prices-to-nuclear-premium); UEC Burke Hollow ISR commenced production (world's newest ISR mine, May 2026); uranium spot $85.20/lb + Sprott purchases resumed.
- us-critical-mineral-independence — China April 2025 active HREE controls now confirmed (Dy 41%, Tb 49%, Y 42% global output; MIIT enforcement up to 5×); USAR Serra Verde acquisition expected Q2-Q3 2026 (production-stage integration); CRML $354M Pentagon contracts added; new mechanism china-ree-controls-to-us-producer-stack.
- picks-and-shovels-leading-edge-fab-buildout — CHIPS 48D analysis added; PWR (below Stifel PT $784 at $711.73) and FIX (4M sq ft modular capacity) added as 48D-deadline construction beneficiaries.
- hbm-supply-bottleneck — new dual-catalyst context (helium cliff + Section 232) integrated; re-baseline recommended at $971 before treating +26.7% prior gain as ongoing edge.
- ferc-large-load-to-dc-gridscale-construction — new mechanism created; FERC Chair Swett committed to end-of-June 2026 ruling on RM26-4-000 (large-load Section 403 streamlining).
Contradicted / weakened
None today.
Open questions worth a human's eye
- CEG position sizing. At -30% from 52w peak with ~$2.3B incremental PJM capacity revenue now confirmed, does the position deserve to be larger? The TMI 2031 delay narrative may have over-punished a stock where the underlying structural thesis (nuclear baseload for AI) is strengthening, not weakening.
- INTC exit discipline. Sector-excess −11.6% confirms the "easy money made" call from CALIBRATION 2026-05-24. VLSI June 14–18 is the last near-term catalyst. If 18A-P yield data disappoints, the exit case becomes unambiguous.
- PWR/FIX hypothesis timing. The CHIPS 48D deadline is December 31 — 7 months away. H2 2026 fab groundbreakings would show in Q3/Q4 order intake. Worth watching: any SIA/fab operator announcement of "construction commencement to lock in 48D credit" is the hypothesis confirmation signal.
- Weekly mark confirmed via Twelve Data (first successful price refresh since 2026-05-18, 12 days). All 22 priced positions confirmed at 2026-05-29 closes. Four still unpriced: CBRS (entry ≈$290.69 unconfirmed), SMNEY (OTC 404), Samsung/SK Hynix KRX listings.
What I looked at
- Sources ingested today: 10 (from autoresearch batch spanning China REE controls, US industrial policy/CHIPS, FERC large-load interconnection, helium/uranium/copper energy, regulatory/antitrust, and 4 podcast transcripts)
- Macro buckets researched: Bucket 2 (Energy/critical minerals), Bucket 3 (US industrial policy/tariffs), Bucket 4 (China/geopolitics), Bucket 5 (Regulatory/antitrust)
- Mechanisms extracted: 4 new (china-ree-controls-to-us-producer-stack, helium-cliff-to-hbm-supply-crunch, pjm-capacity-prices-to-nuclear-premium, ferc-large-load-to-dc-gridscale-construction)
- Hypothesis pages created: 4 (skhynix-helium-section232-micron-hbm-window, chips-48d-deadline-rush-construction-pwr-fix, nee-dominion-virginia-scc-block-risk, domestic-critical-metals-dod-contract-pipeline)
- Valuation snapshots refreshed: 8 (first successful refresh via Twelve Data API; 12-day stale period resolved)
- Weekly paper ledger mark: confirmed 2026-05-30; marks unchanged from 2026-05-29 (markets closed Saturday); no falsification triggers fired
- Podcast episodes transcribed: 4 (via podcast-ingest; sources ingested as part of batch)