Constraint Sectors: Why Bottlenecks Equal Pricing Power
Constraint Sectors are the segments of the AI supply chain where demand structurally exceeds supply — creating pricing power that persists through cycles. Closelook identifies five key constraints: advanced packaging (CoWoS/hybrid bonding), chip testing (Advantest ATE), thermal management (liquid cooling), glass substrates, and precision manufacturing. Companies in these sectors can raise prices without losing customers because there are no alternatives. The market systematically undervalues constraint companies because they are small-cap and under-followed compared to NVIDIA or TSMC.
The Constraint Investment Thesis
In a supply chain, the bottleneck determines the throughput of the entire system. If TSMC can fabricate 100,000 wafers per month but the packaging capacity only handles 60,000, the effective output is 60,000. The packaging provider has pricing power because expanding their capacity is the only way to increase total output.
This creates an asymmetric investment opportunity: constraint companies have disproportionate pricing power relative to their size. BESI (hybrid bonding, €5B market cap) can raise prices because no one else offers comparable hybrid bonding equipment. Vertiv and Modine (liquid cooling) can raise prices because every new GPU cluster needs cooling and deployment timelines are tight.
Five Key Constraints
Advanced Packaging: CoWoS and hybrid bonding capacity. TSMC, BESI, Kulicke & Soffa. → Deep dive
Chip Testing: Advantest ATE equipment. → Deep dive
Thermal Management: Liquid cooling for AI clusters. Vertiv, Modine. → Deep dive
Glass Substrates: AGC, Corning, Intel. Next-generation packaging substrates replacing organic materials.
Precision Manufacturing: DISCO (wafer dicing), Tokyo Electron (deposition). Irreplaceable process steps.