Arm — AI supply-chain exposure

Arm sits inside the silicon rather than beside it. It designs neither chips nor systems; it licenses the instruction-set architecture and CPU core designs that others build into their own parts, collecting a royalty on each. Its designs appear across the AI stack — as the host and control cores inside custom accelerators and ASICs, and increasingly as the server CPUs that orchestrate data center compute.

The structural hook is embedment. Because Arm's IP is licensed rather than sold as a box, it captures value without owning a fab or carrying inventory, and its architecture becomes more entrenched as custom silicon and Arm-based server processors proliferate through the data center. That royalty model, layered under a widening share of AI hardware, is the moat. The model weights it highly because it sits beneath much of the compute layer rather than competing within any single slice of it.