R15 strike day 0 minus one — mediation collapsed, Suwon injunction added, publish-gate still held

Update

R15 strike day 0 minus one — mediation collapsed, Suwon injunction added, publish-gate still held

  • Type: refresh (subtype: continuity-annotation)
  • event_id: 2026-05-20-memory-supercycle-r15-strike-day-zero-minus-one
  • Source: R15 current-week research desk cycle — workflow/RUNS.md 2026-05-20 r15-current-week-research-desk-cycle. Inputs: published/research-notes/2026-05-20-r15-memory-strike-day-zero-minus-one.md; published/scan-result-buckets/2026-05-20-r15-current-week-research-desk-cycle.json (memory-cohort-strike-day-zero-minus-one bucket); targeted web verification (Korea Herald 5/20, CNBC 5/20, Nikkei 5/20, TechTimes 5/19, 24/7 Wall St 5/19, TrendForce 5/13 + 5/15, Tom's Hardware Q1 capex).
  • Decision context: R14 publish-gate ("cohort breaks green") was tested overnight by the 2026-05-19 NLRC mediation outcome and the cohort tape. Did the gate hold?
  • Verdict — gate STILL HELD; posture unchanged. Three meaningful new facts that reinforce, not change, the thesis:
    1. Mediation COLLAPSED 2026-05-19 evening. Union accepted NLRC proposal; Samsung initially rejected then withdrew rejection and asked for more time; union ended talks and confirmed 2026-05-21 strike start. Breakdown over 70/30 bonus split (Samsung opposing "excessive compensation for employees in loss-making divisions"). Korea Herald 5/20 + Nikkei Asia 5/20 + CNBC 5/20. Direction: supports thesis.
    2. NEW operational fact — Suwon District Court 2026-05-19 injunction locks 7,087 essential Samsung chip workers in fabs to maintain safety systems, ventilation, drainage, and in-process wafer preservation. Violations: 100M KRW/day per union, 10M KRW per union leader. Direction: marginally caps strike downside (the 3-4% global DRAM / 2-3% NAND impact estimate has a floor; the safety/preservation layer can't be struck), but does not invalidate the core impact estimate since production-output drop above the floor is unaffected. TechTimes 5/19.
    3. Cohort small-bounce starting on the leader names, but no green-break. MU $698.74 on 2026-05-19 fresh bar (+2.5% absolute vs R14's 5/18 $681.54); HY9H.F €1005 on 2026-05-20 Korea bar (+1.3% vs R14's €992). 24/7 Wall St 5/19 attributes MU's bounce to AI/HBM demand leverage + Reddit-trader strike-on positioning ("the cleanest play on imminent Samsung strike"). Direction: supports thesis but not yet a cohort-breadth signal — cohort 7d still all-red (MU -13%, STX -3%, WDC -5%, DRAM fund -8%, HY9H.F -10%).
  • TrendForce check: 2Q26 forecast (DRAM +58-63% / NAND +70-75% QoQ) UNREVISED; two strike-specific notes (2026-05-13 "Samsung Strike Seen Contained on Revenue Impact, but May Support Prices"; 2026-05-15 "Samsung Reportedly Begins Preemptive Strike Measures... Shifts Product Mix Toward HBM") lean bullish on ASPs. Demand-side primary input UNCHANGED.
  • Hyperscaler memory-throttle check: no commentary found in 2026-05-13 → 2026-05-20 window. Older Q1 commentary (Microsoft CFO $25B capex attributed to memory; Meta raising 2026 capex range citing memory) remains in record. Demand side intact.
  • Effect on thesis: HOLD — posture unchanged. Cohort = MU/SNDK/STX/WDC/DRAM/HY9H.F kept. Status active, priority high. Publish-gate "cohort breaks green" stays; small bounce on the leaders is not enough — needs cohort breadth (4 of 5 names 7d positive) and RSI ≥ 40 floor across the cohort.
  • Worker-count hygiene note: CNBC 2026-05-20 cites 47,000+ workers; R14 used 45,000 figure. Likely late membership sign-ups or different counting methodology (full-NSEU vs unionized-chip-only). Not material to 3-4% DRAM math given Samsung's 36% global share + Pyeongtaek/Hwaseong production weighting per TrendForce. Logged as data hygiene, not a thesis input change.
  • Watches (→ SCANS): (a) Strike day 0 + day 1 cohort tape (2026-05-21 / 2026-05-22) — the cleanest gate test; (b) TrendForce contract-price revision watch carried (day 1 of 5); (c) Hyperscaler capex commentary watch carried (day 1 of 7); (d) NEW Suwon-injunction floor watch — does Samsung's actual fab-output drop noticeably less than the 3-4% DRAM estimate? Monitor 5/21-5/23 via TrendForce or Samsung 8-K-equivalent disclosures.
  • Operational gate sharpening (for next refresh): R14 noted the "cohort breaks green" gate should be tightened. Proposed operational definition: 5-day cohort price ROC turning positive AND RSI flooring at ≥ 40 across at least 4 of 5 cohort names. Adopt at the next refresh that touches the README body.
  • Per-ticker (R15 tape):
    • MU → $698.74 (2026-05-19 fresh), RSI 61, -13.05% 7D, +55.82% 30D — first +5%-class bounce since 5/12; not yet confirmed by SNDK/WDC/STX
    • SNDK → carried at R14 5/18 read (not in memory.json; tracked elsewhere)
    • STX → $740.84 (still 5/18 stale), RSI 60, -2.93% 7D
    • WDC → $458.68 (still 5/18 stale), RSI 58, -5.36% 7D
    • DRAM fund → $49.77, RSI 64, -8.06% 7D — basket still red
    • HY9H.F → €1005 (2026-05-20 Korea fresh), RSI 59, -10.48% 7D (window-roll wider), +45.23% 30D — small bid on the foreign read
2 events
DateTypeEvent
Jun 17 Investigationmemory-tier bifurcation: NVIDIA halving a mid memory tier May 29 UpdateMemory & Storage Candidates

No direct external sources are attached to this read.