Theta Research
  • Home
  • Graph
  • CompaniesPro
  • Pulse
  • Shipped
  • Engine

For serious investors who think in systems.

AboutAbout
ToolsOption LabDashboard
FeedbackRoadmap
© 2026 theta.md
← Back to Events
COMPLETEDTier 1macro-event

Samsung 18-Day Chip Strike (May 21–June 7) — Global Memory Supply Disruption Risk

2026-05-21 → 2026-06-07·11 companies·10 predictions·Brier: 0.158
↓Samsungdirect↑MUcompetitive↑SKHynixcompetitive↑SNDKcompetitive?NVDAsupply-chain↓LINsupply-chain?KLACsupply-chain?AMATsupply-chain—ASMLsupply-chain?LRCXsupply-chain?QCOMsupply-chain

Scenario Comparison

Bull CaseS4
20%

Full 18-day strike with 30-40K worker participation. DRAM output reduced 3-4%, NAND 2-3%. 36-day total production blackout (strike + 2-3 week recovery). Samsung OP impact KRW 5-8T.

↑ 6↓ 411 companies
−Samsungstrong
+MUstrong
+SKHynixstrong
+SNDKmoderate
+7 more
Base CaseS2
25%

Strike begins May 21 but settles within 5 days as government pressure and management concessions align. Participation 15-20K workers (below 30K threshold). <1% DRAM output impact.

↑ 2↓ 14 companies
−Samsungmild
+MUmild
+SKHynixmild
~LINmild
Bear CaseS1
20%

Samsung and union reach settlement before May 21 — compromise on 10-12% of operating profit as bonus with partial cap removal. Strike averted.

↑ 1↓ 34 companies
+Samsungmild
−MUmild
−SKHynixmild
−SNDKmild

All Scenarios

5

Positioning Suggestions

→ MU/SKHynix have asymmetric upside from Samsung disruption — consider small tactical adds if strike begins and lasts >5 days. MU especially benefits from NVL72 allocation narrative shift.

→ Samsung: avoid buying into the pre-strike uncertainty. If a settlement is reached, the relief rally may create a sell opportunity given base-case PW return is already -10%. If strike proceeds, deeper pullback toward KRW 150K improves risk-reward.

→ SNDK: current $635 already reflects pricing tailwinds. Strike is incremental positive but not sufficient alone to shift from hold to buy. Monitor enterprise SSD contract repricing.

→ LIN: take-or-pay contracts insulate downside. Any Pyeongtaek-driven pullback is a buying opportunity — Linde's semiconductor moat is infrastructure-physical, not demand-cyclical.

→ Equipment names (KLAC/AMAT/LRCX): post-strike capex acceleration thesis is medium-term bullish. Samsung's triple-source memory capex convergence (per KLAC/AMAT catalystWatch) validates equipment demand. Accumulate on any sympathy sell-off.

→ NVDA: HBM supply risk is manageable given SKHynix 70% allocation dominance. Any NVDA sell-off on Samsung strike headline is likely overdone — focus on SKHynix delivery capacity.

Predictions

10
✗

Samsung chip workers strike begins on or after May 21, 2026 (NLRC talks fail to produce pre-strike settlement)

70%P1

If confirmed → MU/SKHynix tactical long; Samsung tactical short or reduce. Strike-driven pricing tailwind materializes.

✗

More than 25,000 workers participate on the first day of the strike

45%P2

If <25K → limited impact, buy Samsung dip. If >25K → full disruption scenario in play, hold memory competitor longs.

✗

Samsung strike lasts the full 18 days (May 21 through June 7) without settlement

20%P3

If full duration → extend MU/SKHynix positions; Samsung likely tests KRW 170K support. NVDA HBM supply narrative becomes material.

✗

Korean government intervenes with binding mediation or arbitration order within the first 7 days of the strike

40%P4

If binding mediation → quick resolution, take profits on memory competitor positions. Samsung recovery trade.

✗

DRAM contract price increases exceed 5% QoQ beyond pre-strike trajectory, attributable to Samsung supply disruption

35%P5

If confirmed → validates extended memory cycle thesis. MU/SKHynix Q2-Q3 earnings estimates need upward revision.

✗

Samsung concedes to uncapping performance bonuses at >12% of operating profit (union's core demand)

15%P6

If conceded → structurally higher Samsung labor costs (KRW 5-6T annual). Negative for Samsung long-term margins but removes recurring strike risk.

✗

NVIDIA publicly or through industry channels confirms reallocation of HBM4 orders away from Samsung during the strike window

15%P7

If confirmed → structural thesis change for Samsung HBM competitive position. SKHynix allocation above 75% would trigger Samsung bear-case reassessment.

✗

MU and SKHynix stocks outperform Samsung by more than 5% during the May 21–June 7 strike window

55%P8

Market pricing of competitive dynamics. If divergence exceeds 10%, consider rebalancing — pricing in too much permanent shift.

○

Samsung fab utilization returns to pre-strike levels within 3 weeks of strike resolution

60%P9

If recovery is faster → Samsung dip-buy signal. If slower than 3 weeks → HBM delivery delays compound, extending competitor benefit window.

✗

Total production blackout period (strike + recovery) exceeds 30 calendar days

25%P10

If >30 days → Q2 2026 Samsung earnings significantly impaired. Equipment capex pull-forward thesis (KLAC/AMAT) strengthens for H2 2026.

Key Questions

  1. Will the NLRC-mediated talks produce a deal before the May 21 strike deadline, given Day 1 ended without agreement?
  2. What is the actual participation floor — does SECU withdrawal reduce turnout below 20K (limiting disruption) or does the remaining union's 93% authorization signal higher commitment?
  3. How quickly can Samsung's Pyeongtaek/Hwaseong fabs resume full production post-strike? TrendForce estimates 2-3 weeks recovery — is this optimistic given HBM4 cleanroom requirements?
  4. Will NVIDIA proactively shift HBM4 allocation away from Samsung toward SKHynix/MU as a precautionary supply chain de-risk?
  5. Does the strike settle the bonus dispute permanently (higher labor costs structurally) or create a recurring annual negotiation risk?
  6. How does this interact with Samsung's Pyeongtaek P5 fab completion timeline (H2 2026) and HBM capacity expansion (+50%)?

Monitoring Checklist

0/10 checked

Post-Event Results

Brier Score0.158
correct0/9
Trade Actions8

Actual Outcomes

  • • Strike AVERTED on May 20, 2026 — hours before the scheduled May 21 walkout. Scenario 1 (Pre-strike settlement, p=0.2) actualized.
  • • Government-mediated tentative wage agreement reached after post-NLRC talks: Special Management Performance Bonus funded by 10.5% of operating profit (union wanted 15%); housing loan program up to KRW 500M; average wage increase 6.2% (base 4.1% + performance 2.1%).
  • • Bonus partially funded by company stock over 10+ years, contingent on chip division exceeding OP targets: 200 trillion won for 2026-2028, 100 trillion won for 2029-2035.
  • • Memory division employees can receive up to KRW 600M; non-memory (System LSI/Foundry) ~KRW 210M; DX division likely no OPI given weak performance.
  • • Union ratification vote runs May 22-27, 2026; as of May 22, NSEU turnout 69.15% and yes vote >66%.
  • • Samsung shares: surged 8.5% to record high on May 21 on news; gave back ~2.3% in Seoul on May 22 as the rally faded.
  • • MU: +4% pre-market May 20 (pre-deal, on strike fears), +2.4% overnight May 21, then DOWN in pre-market May 22 as the deal cooled the memory rally.
  • • SNDK: +2.5% May 20, also down May 22 pre-market. Analyst PT raises May 20 capitalized on the rally regardless (Citi $2,025 from $1,300; Melius $2,350 from $1,500).
  • • DRAM ETF: -1.5% May 22 pre-market. Memory peers gave back gains as Samsung supply normalized.
  • • Samsung outperformed MU/SKHynix by ~6% on May 21 — opposite direction of prediction #8.
  • • No DRAM/HBM/NAND production disruption occurred. No NVIDIA HBM4 reallocation announced. No Korean government binding arbitration triggered (NLRC mediation succeeded pre-strike).
  • • JPMorgan analyst Jay Kwon prior estimate: 'if union demands met in full, 2026 OP faces 7-12% downside from increased labor costs alone.' Actual deal at 10.5% of OP (vs 15% demanded) is materially less impact than full-demands scenario.
  • • Net: strike was the low-probability outcome's mirror — Scenario 1 (settlement) won despite model assigning only 0.2 to it.

Market Reaction

Samsung+8.5%MU+2.4%SNDK+2.5%

Thesis Updates Needed

Samsung·Strike risk transitions from 'active/watching' to RESOLVED (no production disruption). New permanent labor cost structure: 6.2% wage hike + 10.5% OP-linked Special Management Performance Bonus + KRW 500M housing loan program. JPMorgan framing: actual deal is materially less than full-demand 7-12% OP downside scenario, but bonus formula now structurally inflates Samsung labor cost base for 2026-2035 horizon. Memory division employees up to KRW 600M individual bonus is unprecedented and creates internal division equity concerns (DX division likely no OPI). NEW CATALYST: ratification vote May 22-27 — if rejected, strike risk re-emerges. Watch turnover/yes-vote trajectory (currently >66% as of 5/22).
risksmonitoring.watchTriggersmonitoring.catalystWatchscenarioAnalysis.scenarios

Trade Recommendations

→ Samsung: relief rally to record high already played (+8.5% on 5/21); positioning should be NEUTRAL not chase. Watch ratification vote 5/22-5/27 — rejection re-introduces strike tail risk. Long-term labor cost reset is fundamentally bearish for FY2026-2028 OP trajectory (10.5% Special Bonus is structural drag) but mitigated by FY2025 record OP base.

→ MU: strike-driven tactical long thesis VOID. Today's MU thesis Hold→Sell rebuild (5/22 06:38) holds regardless — valuation/cycle-peak concerns dominate. No add or trim signal from this event.

→ SKHynix: no allocation windfall from Samsung disruption. SKHynix today's rebuild (5/22 10:39) already explicitly notes 'Samsung 18-day strike suspended May 20.' Thesis intact at Hold.

→ SNDK: NAND tactical pricing tailwind dissipates. Today's SNDK rebuild (5/22 17:35) post-Q3 print already captures the analyst PT rally context. Hold positioning.

→ LIN: NO Pyeongtaek utilization drop. Q2 2026 Linde Asia APAC segment unaffected. Take-or-pay revenue intact. No thesis change.

→ Equipment names (KLAC/AMAT/LRCX): post-strike capex acceleration thesis VOID (no strike = no post-strike spending bump). Equipment demand reverts to organic schedule.

→ QCOM: no Samsung foundry disruption → Snapdragon production on schedule. Thesis unchanged.

→ Methodology lesson: model assigned Scenario 1 (pre-strike settlement) only 0.2 probability. In retrospect, given (a) Samsung's record FY2025 OP cushion enabling generous bonuses, (b) Korean government's standing mediation infrastructure (NLRC), (c) SECU withdrawal weakening union front, the prior was too bearish on settlement. For future labor-action events, model should anchor settlement-probability higher when company financials are strong AND government mediation has structural authority.