Teradyne trades at 17.3x forward earnings while Applied Materials sits at 18.1x, an inversion that reflects divergent AI semiconductor exposure. The test equipment provider now commands 34% of its revenue from advanced packaging and AI-related testing cycles, against Applied Materials' 19% direct AI fab tool revenue. The gap widened 11 percentage points in eight months.

The catalyst is structural. Hyperscalers order chips in batches sized for 18-24 month deployment windows, not the 36-48 month fab capacity cycles AMAT depends on. Teradyne books orders when chiplets reach heterogeneous integration and final test, stages that repeat with each design iteration. AMAT books orders when fabs expand clean rooms, a capital event that happens once per node generation. The $47B AI chip market now turns over test equipment 2.7x faster than deposition tools.

Teradyne posted 22% year-over-year earnings growth in the December quarter, driven by $310M in AI test system revenue. Applied Materials grew 9% in the same window, constrained by memory fab delays in Korea and China export restrictions on advanced lithography support tools. Teradyne's backlog stands at $1.18B, 63% of which converts to revenue within six months. AMAT's $15.2B backlog converts over 14-16 months, weighted toward trailing-edge logic and mature DRAM.

The valuation discount exists because allocators still model semiconductors as a fab-equipment story. They price AMAT for the $200B foundry buildout through 2026 and discount Teradyne as a late-cycle service provider. The error is categorical. AI chips fail at higher rates than prior nodes—Nvidia's Blackwell system-on-wafer posts a 7-9% test escape rate versus 3-4% for prior Hopper generations—which pushes test intensity up 140 basis points per wafer. Teradyne sells into that variance. AMAT sells into capacity, which hyperscalers now lease rather than own.

Operators should track Teradyne's April guidance for second-half test equipment orders and AMAT's March investor day commentary on AI-specific tool revenue splits. Teradyne typically pre-announces if backlog conversion slips below 58%, a threshold it has not breached since Q3 2022. AMAT's forward multiple compresses when memory capex disappoints, an event Micron and SK Hynix signal 45-60 days before earnings. The next inflection arrives when Taiwan Semiconductor begins 3nm high-volume manufacturing in June, an event that pulls forward both companies' revenue but favors Teradyne's faster turn.

The trade is not about AI hype. It is about who invoices first when a hyperscaler orders 100,000 chips. Teradyne does. Applied Materials invoices when the hyperscaler builds the room to make those chips, an event that happens earlier but repeats less often. The 140 basis point valuation gap prices the wrong cycle.