Keysight Falls 11.1% in 3 Months While AI Demand Keeps Accelerating

Keysight reported fiscal Q3 2026 revenue of $1.846 billion, up 36% year over year, and non-GAAP EPS of $3.07, up 79%, while orders rose 56% to a record $2.091 billion on AI infrastructure and connectivity demand. Management guided fiscal Q4 revenue to $1.930-$1.950 billion, implying 37% growth at the midpoint, and consensus EPS estimates have increased 14.3% over four weeks. Despite these fundamentals, KEYS shares have fallen 11.1% over three months, reflecting supply constraints, China/export-control and tariff uncertainty, and a declining software/services revenue mix; the stock trades at 22.6x forward earnings.

Analysis

KEYS is becoming a higher-beta way to express AI networking and validation capex rather than a recurring-revenue compounder. The key underwriting question is whether elevated order activity converts cleanly into shipments and gross margin; if constrained components defer delivery, the market will discount backlog quality and apply a lower multiple despite continued estimate revisions. The mix shift toward hardware also raises working-capital intensity and makes earnings more exposed to hyperscaler deployment timing than the headline growth rate implies.

Competitive read-through is favorable for the semiconductor-design/test ecosystem, but KEYS is not interchangeable with CDNS. CDNS monetizes design complexity earlier in the silicon cycle through more recurring software economics, while KEYS monetizes the later physical-layer validation cycle; a pause in AI cluster construction would hit KEYS first even if chip-design activity remains resilient. Advantest (ATEYY) is a closer cyclical beneficiary, but its AI-test exposure carries more direct foundry/customer-concentration and Japan FX sensitivity.

Near term, consensus revisions can support KEYS through the next results, provided book-to-bill remains above 1x and management does not extend lead times or reduce margin assumptions. Over 6-18 months, export-control tightening is the asymmetric risk: losing China-adjacent communications demand could combine with a lower software mix to compress the earnings multiple. The contrarian point is that the recent pullback is not necessarily a valuation reset; it may be the market pricing a transition from unusually easy comparisons to execution-dependent growth.

Key Decisions for Investors

• Initiate a modest 1-3 month long KEYS only on confirmation that quarterly book-to-bill remains above 1.0x and gross-margin guidance holds; target 10-15% upside from multiple normalization, with a 7-8% stop if orders decelerate materially or lead times extend.
• Prefer a 6-12 month pair: long CDNS / short KEYS in equal beta-adjusted dollars. This captures design-tool recurring revenue versus late-cycle test-equipment cyclicality; exit if KEYS software/services mix stabilizes above 35% or CDNS shows a material AI-driven license slowdown.
• Use KEYS earnings as an event-risk watch rather than a pre-results outright short. A revenue guide at or above the high end without incremental supply-capacity commentary would likely squeeze shorts; conversely, any shipment deferral or China-policy caution is a catalyst for 10%+ downside given hardware-mix multiple risk.
• Monitor U.S. export-control announcements and Chinese AI-networking procurement indicators over the next 1-3 months. Escalation is a trigger to reduce KEYS exposure and favor domestically oriented AI infrastructure beneficiaries; absent such evidence, the policy discount remains speculative rather than thesis-defining.