Stock-Market Research Dispatch — 2026-05-25

Auto-generated by daily research routine. Not a source; not ingest-pending material. Summary of what changed in the wiki today.

Top of mind

China mineral controls are a two-layer system — and one layer just became permanent infrastructure. The October 2025 export controls (the "tariff pause" REE story) are SUSPENDED until November 10, 2026 under the US-China trade deal. But the June 15, 2026 Mineral Resources Law (79 articles, 8 chapters, signed and active) is a domestic production control mechanism — legally distinct from export controls and NOT touched by the trade deal. These two instruments operate independently: the trade deal can suspend export controls while the Mineral Resources Law quietly governs what China mines and processes domestically. For the us-critical-mineral-independence thesis, this means the November 10 binary catalyst is real, but the June 15 law creates ongoing production-level leverage regardless of any future US-China deal. MP Materials ($64.46 May 24) and the rest of the non-China REE supply chain remain structurally advantaged. Near-term watch: what metals does the June 15 law actually cover? Tungsten, antimony, and silver are whitelisted under the trade deal; REEs are not confirmed exempt.

The Section 232 semiconductor thesis has a new legal leg: the Supreme Court just cleared its path. The Learning Resources v. Trump ruling (Feb 20, 2026, 6-3) struck down IEEPA tariffs — but the majority explicitly contrasted IEEPA with Section 232, writing that Section 232 "authorizes duties." The ruling legally fortifies Phase 1 (Proclamation 11002, 25% on chips) and makes Phase 2 cleaner to enact. Simultaneously, the 10% Section 122 replacement surcharge expires ~July 19, 2026. The convergence: July 1 Commerce report (Phase 2 gate) + July 19 Section 122 expiry means mid-July is a 3-week window where the entire tariff regime either re-sets or creates a gap. TSMC Arizona Fab 21's year-one profitability ($514M) confirms Phase 1 is already delivering a structural advantage for US-fab chips vs Taiwan-fab equivalents — the thesis isn't waiting for Phase 2 to be operative.

New wiki pages

Significant wiki updates

CSP capex NOT peaking — definitive (csp-capex-cycle-peak-or-sustained): Q1 2026 earnings answers the question. $650-725B combined 2026, +62-64% YoY. Azure $80B of unfulfilled demand (capacity constrained, not demand constrained). Goldman through-2027 model sees sustained multi-year level. Power grid is the binding constraint (4-7 year interconnection queues), not demand. The data-center electrical picks-and-shovels hypothesis datacenter-construction-electrical-picks-shovels is confirmed: GEV Q1 orders +71%, ETN backlog +44% YoY, VRT project backlog doubled. These three entities graduate from "watch" to confirmed evidence.

INTC re-rate thesis intact but no longer "undervalued" (calibrated): The IFS pivot is positive — 14A is now the primary external customer node (18A supports Apple M7 and internal; 14A is for iPhone A21 ~60M/yr, Tesla, and future large customers). Tesla confirmed as first 14A customer. ASML Q1 2026 orders €13.2B (2x estimates) provide picks-and-shovels confirmation of the capex commitment. BUT: INTC at $119-120 is trading 24-35% above BofA's post-deal PT of $96. The easy money from the IFS re-rate is made (+478% YTD). Remaining upside requires bull-case execution: Apple deal finalizing with volume orders, 14A securing 2+ major customers by year-end, and yield curve reaching industry-standard by 2027. Status changed from "undervalued vs base case" to "at/above base case — bull-case premium required."

TSMC Q1 2026 comprehensively strong (tsmc-capacity-shortfall-and-pricing-power): $35.9B revenue (+40.6% YoY); gross margin 66.2% (above guidance ceiling). Four consecutive price hikes confirmed at 3-5%/yr. N2 wafer >$30K, fully booked (Apple >50%). CoWoS 50+ week lead times; Nvidia ~60% of global demand. Arizona Fab 21 $514M year-one profit — the fastest-profitability-ramp in TSMC history. TSM at ~$1.88T market cap: the pricing-power thesis is well-understood; incremental upside is from Arizona geopolitical de-risking and margin compounding above guidance.

AMKR gap in CoWoS bottleneck thesis (cowos-packaging-capacity-crunch): TSMC is outsourcing CoWoS overflow to both ASE AND Amkor Technology (AMKR) — but AMKR is not in the concept's suggested-tickers. The persistent 15-20% supply-demand gap and TSMC's confirmed outsourcing to AMKR suggest Amkor is getting direct CoWoS revenue not captured in the current thesis. New hypothesis question filed.

Watchlist

Undervalued candidates (today)

Genuinely undervalued vs base case is harder to identify after today's calibration that INTC's re-rate is priced in. Candidates requiring fresh price data (403-blocked) before forming a view:

• datacenter-construction-electrical-picks-shovels — GEV, VRT, ETN. Q1 data confirms these picks-and-shovels are getting the orders (GEV +71%, ETN +44% backlog). Whether they're undervalued depends on whether the market has caught up with the Q1 beat-and-raise pace. Need price snapshots to form a view.
• amkor-cowos-tsmc-overflow-osat — AMKR. Uncaptured CoWoS exposure not yet in any suggested-ticker list. If TSMC's CoWoS outsourcing to Amkor is material, AMKR may be undervalued vs the scenario implied by TSMC's full-sold-through-2027 commentary.

New chains (hypothesis-stage)

• amkor-cowos-tsmc-overflow-osat — AMKR. Graduate: AMKR earnings confirming CoWoS revenue from TSMC contracts.
• quantum-computing-chips-grants-first-mover — IBM, GFS (investable); QBTS, RGTI (speculative). Graduate: final CHIPS award execution + first commercial customer for Anderon/QTS foundry.

Calibration logged

[[2026-05-25 10:15] INTC "undervalued" framing no longer accurate after +478% YTD. Prior belief: INTC was the clearest undervalued re-rate candidate. Updated: at ~$119-120 and trading 14-35% above BofA's post-deal PT ($96), the IFS thesis has been substantially priced in. Remaining upside requires bull-case execution (Apple orders, 14A multi-customer, industry-standard yield by 2027). Pattern: overconfidence/re-rate-timing — didn't re-baseline "undervalued" framing after the stock's +4x move.

Step 4b — Valuation snapshots

WebFetch blocked (HTTP 403) for ninth consecutive run. Only MP Materials price ($64.46, May 24) available via WebSearch snippet. All other concept pages carry stale snapshots. GEV/VRT/ETN are new entities with no baseline snapshot yet.

What I looked at

• Step 1 topics: China two-layer mineral architecture (June 15 MRL vs trade deal suspension); CSP capex cycle definitiveness (Q1 2026 answered); Intel 18A/IFS status (14A as external node, Tesla 14A confirmed, ASML €13.2B orders); TSMC capacity/pricing comprehensive update; US industrial policy delta (IEEPA ruling fortifies 232, Section 122 expiry July 19, CHIPS quantum grants $2.013B)
• Step 2 macro buckets: Energy & critical minerals (China two-layer architecture); US industrial policy & tariffs (IEEPA ruling, Section 122, quantum grants)
• Podcast episodes transcribed: 0 (all RSS feeds HTTP 403; ninth consecutive failure)
• Sources promoted: 5 (all 2026-05-25 autoresearch); sources ingested: 5
• Valuation snapshots refreshed: 0 via WebFetch (403 blocked); cowos page date updated to May 25
• New hypothesis chains opened: 2 (AMKR CoWoS overflow, quantum grants first-mover)
• New entities created: 4 (energy-fuels/UUUU, gev/GEV, vrt/VRT, etn/ETN)
• Calibration logged: 1 (INTC re-rate priced in)

Open questions for human review

1. June 15 Mineral Resources Law scope: Which specific minerals does the MRL regulate production of? Tungsten/antimony/silver are whitelisted under the trade deal — are REEs also whitelisted, or is China's MRL the route to tightening supply independently of the trade deal?
2. July 1 / July 19 convergence window: July 1 Phase 2 Commerce report + July 19 Section 122 expiry create a 3-week window where the full tariff picture could shift materially. Does this warrant accelerating position sizing in US-fab beneficiaries before July 1?
3. AMKR: look it up? AMKR is getting direct TSMC CoWoS overflow contracts that ASE was already receiving — but AMKR isn't tracked. Is Amkor worth adding as a named ticker, or is ASE sufficient coverage for the OSAT CoWoS angle?
4. INTC: hold or reduce? After +478% YTD and at/above BofA's base-case PT, the INTC position (if held) warrants a review. The thesis is intact for bull case; but base case is largely priced. What would have to be true to justify holding vs trimming?
5. Podcast feed 403 (ninth consecutive run): Network policy issue with the remote execution container. This is infrastructure, not content — likely needs to be addressed at the container configuration level rather than within the daily routine.