Cohu beats third quarter estimates on strong AI demand

Cohu beat Q3 analyst estimates, reporting adjusted EPS of $0.43 versus $0.35 consensus and revenue of $173.0 million versus $170.24 million, up 37% year over year. Bookings rose 126% year over year, and management raised its fiscal 2026 high-performance computing revenue outlook to approximately $120 million; Q4 revenue guidance was $178 million ±$7 million, implying 46% year-over-year growth at the midpoint. GAAP net income was $12.1 million versus a $4.1 million loss a year earlier, and shares rose 2% in after-hours trading.

Analysis

The key signal is not the beat itself but whether elevated systems bookings convert into shipments without sacrificing gross margin. If HPC devices require more demanding thermal test, Cohu could gain a larger share of test spend per device; that would support a better-quality growth profile than a one-off equipment cycle. It may also pressure competing test-equipment suppliers, including Teradyne and Advantest, if customers allocate incremental budgets toward Cohu’s thermal solutions—but the reported pipeline is not booked revenue, and customer qualification timelines matter.

Near term, the risk is a crowded interpretation of forward indicators: a sharp increase in systems bookings can be lumpy, while recurring-revenue growth is a more durable but slower signal. Over 1–3 months, watch order conversion, backlog, and gross-margin performance alongside Q4 delivery. Over 6–18 months, sustained HPC adoption could improve utilization and operating leverage; conversely, a pause in AI infrastructure spending or delayed customer programs could expose the cyclicality of systems demand. The press release does not establish customer concentration, cancellation terms, or valuation, so the result alone does not justify chasing the after-hours move.

Key Decisions for Investors

• Consider a measured long in COHU on a post-print pullback or if the initial reaction holds through regular trading; do not size the position as though the stated AI opportunity pipeline were contracted revenue. Upside depends on bookings converting into recognized sales while margins hold.
• Treat the next 1–3 months as a confirmation window: track systems bookings and backlog conversion, HPC revenue progress against management’s outlook, and gross margin. A bookings reversal, delayed shipments, or material margin deterioration would invalidate the growth-and-operating-leverage thesis.
• Avoid a direct short of Teradyne or Advantest based on this report alone. Revisit a relative-value position only if customer qualification or share-allocation evidence shows Cohu winning incremental thermal-test spend at their expense.
• Before adding materially, verify valuation versus Cohu’s normalized earnings, customer concentration, order cancellation exposure, and the timing of pipeline conversion; these are not established by the reported figures.