A Concentrated Leading Edge

Semiconductor manufacturing divides into two broad camps: integrated device manufacturers that design and fabricate their own chips, and pure-play foundries that manufacture for external customers. At the leading edge, the number of players capable of producing the most advanced logic nodes has narrowed to a very small group, reflecting the staggering cost and complexity of each new generation. Building a leading-edge fab now runs into tens of billions, and only a few organisations can sustain that pace.

The Layers of the Landscape

Below the bleeding edge, the picture broadens considerably. Many foundries operate profitable, high-volume businesses on mature and specialised nodes, serving automotive, industrial, power and analogue markets that do not need the newest process. This layered structure means the industry’s health depends on far more than the race to the smallest transistor.

• A very small set of firms at the leading logic edge.
• Memory manufacturing concentrated among a few specialised producers.
• A broad base of foundries on mature and specialty nodes.

Geography and Resilience

Leading-edge capacity is geographically concentrated, which has prompted governments worldwide to fund new fabs in pursuit of supply-chain resilience. New facilities take years to build and staff, so shifts in the geographic balance unfold slowly. The equipment supply chain, most visibly ASML’s near-monopoly on EUV, adds another layer of concentration that shapes who can build what.

Understanding this landscape, its concentration, its economics and its geography, gives context to every technical decision made in the fab, and providing that context is part of how Avecas frames its VLSI education.