Semiconductor cycles are reasonably predictable in shape but difficult to time precisely, and the difficulty comes from the fact that no single indicator is sufficient by itself. The major leading indicators — book-to-bill (covered in sc1_l11), channel inventory days, supplier capex-to-revenue ratios (covered in sc1_l5), and end-market unit volumes — each capture a different aspect of the cycle, each leads semiconductor revenue inflections by a different amount of time, and each misses critical aspects of demand or supply that the others capture. The 2024-2025 environment is a particularly clear illustration: AI compute cycle leading-edge demand has been durable while traditional smartphone / PC / industrial end markets have been weak; memory cycle has recovered while standard DRAM oversupply remains a risk; semicap book-to-bill has been elevated on AI / leading-edge build-out while China-restricted revenue has compressed elsewhere.

Reading any single indicator in isolation produces persistent analytical errors; the discipline is triangulation across multiple indicators with explicit attention to what each one misses.