Cohu, Inc.

▲ Bull Case

• ◆HPC scales beyond guidance: Second Neon HBM design win at SK Hynix or Micron in 2026; Eclipse HPC handler at AI compute customers; HPC revenue reaches $130M+ in FY2026 and $250M+ by FY2028, transforming Cohu's growth profile and partially de-correlating it from the automotive cycle.
• ◆Margin expansion exceeds historical peak: Volume leverage + contactor mix at 60%+ recurring + HPC ASP premium drive non-GAAP gross margin to 50–51% and non-GAAP operating margin to 16–17% — above any prior cycle peak — producing FY2028 non-GAAP EPS of $2.50–3.00.
• ◆Multiple re-rating toward peers: As HPC revenue validates, COHU's EV/Revenue compresses the discount to peer median (currently ~4.3x vs. peer ~4.5x); price target reaches $75–83 at 28x FY2028 EPS, representing 35–50% upside.

▼ Bear Case

• ◆Automotive recovery stalls: OSAT automotive utilization plateaus below 80% through FY2026–FY2027 as EV production growth disappoints (Ford/GM EV pullback compounds) and Tier 1 inventory destocking extends; handler revenue underperforms by 10–15% vs. guidance, and FY2027 revenue lands at ~$555M (vs. base ~$634M).
• ◆HPC fails to scale: Neon HBM second design win is delayed past FY2026; Eclipse HPC faces Teradyne competitive response; FY2026 HPC revenue undershoots to $50–60M (vs. $80–100M guidance), invalidating Pillar 3 of the thesis tracker and triggering multiple compression.
• ◆Margin compression from competition / mix: Chroma ATE price pressure in commodity handlers + handler-heavy mix as automotive recovers + Philippine peso appreciation → non-GAAP gross margin stays at 44–45% (vs. base 47–48%); EPS reaches only $0.95 in mid-cycle and stock returns to $22–25.

“The sell-side is uniformly bullish (9 Buy / 0 Hold / 0 Sell per Step 17), but the average price target ($35.50) sits 35% below the current price ($54.95) while Jefferies' high target ($60) sits 9% above. The dispersion reflects the actual debate: the bears argue the stock has fully priced in the cycle recovery (re-rated from $17–22) and the HPC opportunity is too early to underwrite. Average targets reflect this caution. The bulls argue (Jefferies) that the HPC pivot is genuine and the test-intensity multiplier will compound over multi-year periods, justifying higher multiples. The variant view from this research (Step 16) is that the test-intensity multiplier is structurally underappreciated by all consensus models — but this is a 3–5 year story unmeasurable in the next 12 months, so it cannot drive near-term price action. The crux of the debate is the HPC revenue trajectory in FY2026 — every quarter's update on Neon HBM design wins and HPC revenue run-rate moves the multiple between bear and bull camps.”

• ◆Q2 2026 earnings (Jul–Aug 2026): Revenue $144M ±$7M; watch HPC update + GM trajectory
• ◆Neon HBM second design win (any time): At SK Hynix or Micron — validates Neon as a platform
• ◆OSAT automotive utilization crossing 85% (Q3–Q4 2026): Trigger for broad-based IDM/OSAT handler restocking
• ◆HPC revenue exceeding $100M run-rate (Q3–Q4 2026): Triggers peer multiple re-rating
• ◆NXP / Infineon capex guidance raises (Q2/Q3 2026): Direct leading indicator for handler orders
• ◆SiC/GaN handler ASP premium realization (FY2027): Step 16 variant view — premium handler share gain
• ◆India OSAT first major design win (2027): Long-term geography expansion
• ◆Potential acquirer interest (speculative): Strategic logic for Teradyne; 30–40% takeout premium

• ◆SEMI cycle reversal (Moderate probability, Catastrophic impact): -30–40% revenue decline
• ◆Automotive recovery stalls / EV slowdown (Moderate probability, High impact): -15–25% revenue decline
• ◆HPC revenue undershoots guidance (Moderate 20–25% probability, High impact): Pillar 3 invalidation
• ◆China trade restrictions escalate (Moderate probability, Moderate-High impact): 15–20% revenue at risk
• ◆Teradyne competitive response in handlers (Low probability, High impact if realized)
• ◆Margin recovery fails (Low-Moderate probability, High impact): 200–400 bps GM compression
• ◆Future M&A at cycle peak (Low probability, High impact): Capital misallocation
• ◆Customer concentration (auto cluster) (Moderate probability, High impact): Synchronous capex cuts