Investors Who Have Been Following AI Chips and Memory Chips May Be Forgetting About the Vital Industry That Enables Both of Them

AI and memory-chip demand is driving strong semiconductor-equipment spending, with Applied Materials reporting 25% year-over-year fiscal Q3 sales growth and targeting $10.25B of fiscal Q4 revenue, implying more than 12% sequential growth. Lam Research's revenue rose 15% sequentially and its $8.1B Q3 midpoint forecast implies 21% sequential growth, while Cohu delivered 38% year-over-year Q2 growth and lifted its AI-compute pipeline to roughly $850M. The article argues that faster equipment replacement cycles for increasingly advanced AI chips support continued growth and margin expansion across equipment suppliers.

Analysis

The investable distinction is not “AI equipment” broadly, but exposure to leading-edge logic versus memory-capex and backend test. AMAT and LRCX monetize incremental process complexity across deposition/etch steps, while KLAC is the cleaner yield-management beneficiary as advanced packaging and HBM raise defect costs. If memory suppliers sustain capacity additions, the equipment cycle can broaden beyond NVDA-led logic spending; if they merely replace legacy capacity, the apparent multi-year backlog visibility will prove less durable.

The article’s growth framing should be discounted until reconciled with company filings: the cited quarterly revenue target appears implausibly large for AMAT and may conflate an annualized or long-term metric. That matters because equipment stocks already price a strong wafer-fab-equipment recovery; incremental upside now requires upward revisions to 2027 spending, not simply healthy order commentary. Export-control tightening on China remains the most immediate downside catalyst for AMAT/LRCX, while a pause in HBM pricing or hyperscaler capex guidance would likely compress multiples before reported tool revenue slows.

COHU is a different risk profile: a return to profitability can create high operating leverage, but it is also more exposed to backend-test timing, customer concentration, and weaker automotive/industrial utilization than the large front-end tool vendors. The contrarian view is that AI-driven test demand may accrue more to TER and Advantest than COHU unless COHU converts its stated pipeline into booked orders and demonstrates sustainable gross-margin expansion. Over 6-18 months, chiplet architectures increase test-content intensity, but that is a structural option rather than evidence of near-term earnings power.

Key Decisions for Investors

• Maintain/establish a 1-3 month relative-value long KLAC versus short LRCX in equal dollar size if the spread narrows toward recent lows: KLAC has better protection if customer capex shifts from wafer starts toward yield and advanced packaging. Exit if LRCX guides system revenue materially above consensus or KLAC order growth decelerates for two consecutive quarters.
• Do not chase AMAT on promotional growth claims. Set an alert around its next earnings release for disclosed China revenue mix, backlog conversion, and fiscal-2027 WFE assumptions; initiate only if management raises normalized WFE outlook without a larger China concentration. A China-restriction announcement is the principal near-term stop condition.
• Treat COHU as a small, catalyst-driven watch position rather than a core AI holding: consider a starter long only after quarterly bookings and gross margin confirm conversion of the compute pipeline, with a 6-12 month target of sustained GAAP profitability. Falsify on renewed losses, sub-$170M quarterly revenue trajectory, or margin reversal; TER is the preferable liquid proxy for immediate AI-test exposure.
• For a broader cycle expression over 3-6 months, favor long SOXX/short a basket of lower-quality semiconductor-capital suppliers only after memory producers lift 2027 capex plans. Without verified capex revisions, the sector signal is momentum-driven and offers insufficient asymmetric entry.