COHU · Cohu, Inc.

Current Thesis

Cohu has re-rated from a cyclical handset/auto test laggard into an AI-compute test picks-&-shovels name, and the narrative is still being discovered by sell-side rather than crowded by retail. The leg being bought: the AI xPU test-handler ramp (Eclipse + Active Thermal Control) plus HBM inspection (Neon) inflecting in H2 2026, with order velocity already confirming it (orders +57% YoY on the 2026-04-30 Q1 print). Theme status: ACCELERATING FY26 revenue-growth guide raised to +20-25%, AI-compute pipeline lifted to ~$750M, HPC FY26 outlook bumped to $80-100M. The constraint on a fresh entry: the tape is stretched at 52-week highs (~$55-57, +214.7% over 12 months as of 2026-06-04) with no hard catalyst inside the next 30 days the binary that validates or breaks the H2 ramp is the Q2 print (~2026-07-30, after close), eight weeks out. This is HIGH-conviction-but-extended, not a SUPREME fat-pitch at this price.

Bull Case

• Q1 FY26 (2026-04-30): revenue $125.1M vs consensus ~$122.1M; orders +57% YoY; gross margin 46.5% beat-and-raise with high-quality recurring/consumables mix (~60% of revenue).
• Q2 FY26 guide $144M ±$7M = +15% QoQ, ~+34% YoY sequential acceleration, not a one-quarter pop, driven explicitly by HPC/AI demand.
• FY26 revenue-growth guide raised to +20-25% on the Q1 call; AI-compute pipeline lifted to ~$750M ($650M test handlers + $100M HBM inspection).
• Eclipse AI xPU test handler: ~$650M SAM, $60-80M FY26 revenue, ~$30M follow-on orders booked; handles 3kW power dissipation with a 10kW roadmap into a GPU market compounding ~23%.
• Neon HBM inspection/metrology: ~$100M SAM, ~$20M FY26 revenue, +80% YoY direct HBM build-out leverage.
• Customer pipeline depth (Q1 call): ~12 AI/HPC customers 5 in qualification, 7 in early engagement; HPC FY26 outlook raised to $80-100M; new DiamondX GaN power orders (~$5M) add data-center-power exposure.
• Sell-side cluster upgrade (orthogonal confirmation): Needham $33→$54, Jefferies $55→$60, TD Cowen $60 (2026-05-01), plus Evercore ISI / B. Riley clustered $53-60. StocksToTrade (2026-05-27) framed it as AI orders and targets "jumping" narrative still mid-discovery, not peak-retail.

Bear Case

• Late on the chart: ~$55-57 sits at the top of the 52-week range ($16.95–$58.47) after a +214.7% YoY run a fresh buy here is the top of the range with the next catalyst eight weeks out.
• Margin tailwind is behind it: GM guided down to ~44% for Q2 from 46.5% as systems mix rises and HPC scales; the easy-margin portion of the Q1 beat does not repeat.
• The re-rate is a bet on an unshipped ramp: the multiple prices an H2-2026 HPC production ramp; any 2026-07-30 order-stall, qual slippage, or pushout snaps it.
• Small-cap semicap beta: ~$2.5B mcap moves violently with the back-end cluster (TER/ONTO/FORM/KLAC); a sector de-rate drags it regardless of company-specific news.
• Lumpy, concentrated orders: handler POs are chunky one large AI customer slipping an order is a double-digit revenue swing quarter-to-quarter.
• Stale-screener risk cuts both ways: blended consensus PT still reads ~$42 on some feeds (a lagging average that drags stale pre-print targets); the live, post-print active cluster is $53-60. Anchoring to $42 understates the re-rate; anchoring to $60 assumes the ramp already happened.

Setup & Price Structure

• Price: ~$54.5 (2026-06-01) → ~$55.75 with an intraday $53.50-$57.42 band (2026-06-04), pressing the 52-week high; full range $16.95–$58.47.
• Trend: cleanly ACCELERATING the late-April post-Q1 gap (~$33 into the $50s) held and extended; price sits above all major moving averages.
• Estimated structure (small-cap, verify live): the late-April breakout base around $48-52 (gap/consolidation shelf) is the first constructive re-entry zone on any pullback; the 20-week EMA roughly ~$44 is the line that defines the momentum leg a weekly close below it would end the trend read.
• Beginner-trap read: stretched well above the breakout base and at the top of the range, but NOT at peak-retail saturation (coverage still framed as "under-the-radar"/AI-orders-jumping in late May). No earnings inside 3 trading days. The trap here is not averaging-down it is paying the top tick of the range without a catalyst to pull it forward; the disciplined add zone is a pullback toward $48-52, not a chase of $57.

Catalyst Calendar (next 30 days)

• No hard binary inside the 2026-06-07 → 2026-07-07 window this is the defining feature of the setup right now: an accelerating name treading at highs with the next proof-point still ~8 weeks out.
• Semicap-sector read-through (mid/late June): peer prints and conference commentary (TER, ONTO, FORM, KLAC, AMAT) drive cluster beta; any HBM/AI back-end capex confirmation from a peer is a positive tell, a guide-down is the warning.
• Investor-conference circuit (June, est.): small-cap semicap names often present at June tech/growth conferences any new Eclipse/xPU design-win disclosure or follow-on order color would be an out-of-cycle accelerant; verify the IR calendar.
• Q2 FY26 print (~2026-07-30, after close — OUTSIDE the 30-day window but the key binary): validates or breaks the H2 HPC ramp against the $144M ±$7M guide.

What Would Change Our Mind

• Trend break: a weekly close below the 20-week EMA (~$44) ends the momentum leg that is the mechanical line, not the cost of any prior buyer.
• Q2 disappointment (~2026-07-30): revenue below the $144M ±$7M guide, gross margin guided under ~43%, or FY26 revenue-growth guide cut below +20% any one breaks the acceleration story.
• Ramp stall: no new xPU handler design wins and no Eclipse follow-on order cadence on the Q2 call silence where the bulls expect $30M-plus follow-ons is the first crack.
• Theme flip: semicap cluster rolling over (TER/ONTO/FORM/KLAC losing their own breakout structures) flips the theme toward SATURATED and would force a trim regardless of COHU-specific news.
• Saturation tell: mainstream/CNBC-headline "AI test play" coverage arriving alongside a stretched +50%-above-MA tape with no new catalyst = late-stage; fade strength rather than chase it.

Correlation Notes

• Trades as a back-end semicap cluster member, not a standalone AI name. Primary correlations: TER (Teradyne, direct ATE peer), ONTO (Onto Innovation, inspection/metrology), FORM (FormFactor, probe), KLAC (KLA, inspection), AMAT (sector beta).
• Second-order driver: the HBM/GPU build-out read-through from NVDA/AMD AI-accelerator demand and from memory (MU, SK Hynix, Samsung) HBM capacity adds. Neon's +80% HBM-inspection growth is a derivative of that capex.
• Beta character: ~$2.5B small-cap with sharp moves on semicap-theme flows; expect it to over-shoot the cluster in both directions. Treat sector-wide de-rates as the dominant exogenous risk over idiosyncratic COHU news between now and the late-July print.

[Note: stale blended-consensus PT (~$42) still appears on some screeners it lags the live post-Q1 active-target cluster of $53-60 and should be ignored.]