AI supply-chain thesis — mapping bottlenecks, focus companies, and supply-chain exposure for investors.
**Bottleneck theme:** Lithography & Fab Tools
**Focus:** $KLAC — KLA CORP
KLA sits at the top of the process-control food chain — wafer inspection, reticle inspection, e-beam metrology, and overlay/critical-dimension metrology — with roughly 50% global share of the process-control segment. It is effectively a near-monopoly at the most advanced nodes (N5 / N3 / N2 / sub-2nm) and at advanced packaging, where the smallest defect on a 12nm-pitch HBM via or a CoWoS bonding interface can scrap multi-thousand-dollar dies. Camtek ($CAMT) and Onto Innovation ($ONTO) compete in advanced packaging metrology specifically — but KLA sits as the umbrella above both, owning the front-end logic and DRAM process-control budget that dwarfs the packaging segment.
The investment case is mid-cycle compounding plus secular complexity expansion: more layers, smaller features, and more process steps all require more inspection and metrology per wafer-out. KLA has the highest gross margin in WFE (~62%+) and consistently delivers the best operating leverage. Risks: China revenue exposure under export-control headline risk, customer concentration (TSMC + Samsung + Intel + Micron drive most demand), and a multiple that already prices a lot of the advantage.
Lithography & Fab Tools — Investment-thesis bucket from bottlenecks.app: Lithography & Fab Tools
HBM — High Bandwidth Memory — 3D-stacked DRAM (HBM2E/HBM3/HBM3E/HBM4) connected via through-silicon vias, delivering 1+ TB/s of bandwidth per stack. Co-packaged with GPUs, TPUs, and custom AI accelerators for datacenter AI training/inference and HPC workloads.
