Part of the day: High-layer-count PCBs

Assembled by the nightly build from this catalog’s own entry for High-layer-count PCBs (4.05, Layer 4 — Servers, racks & integration). The tables below are rebuilt every night; first published 2026-09-21, refreshed 2026-09-24.

The bare printed circuit boards themselves — high-density interconnect, backdrilled, 20 to 40 copper layers — that GPU and switch trays are built on.

This is the physical board substrate beneath the baseboards and switch trays (4.03, 4.04): a laminate stack of 20 to 40 signal and power layers, built with HDI microvias and backdrilling to keep high-speed signal integrity across the board. It sits downstream of the laminate and copper-foil suppliers (4.06) and upstream of everything that gets populated onto it. AI boards push layer count and manufacturing precision well past a standard server motherboard because scale-up interconnect, high pin-count sockets and dense power delivery all need more routing layers and tighter tolerances than a general-purpose board. That raises both the capital intensity of the fabrication step and the qualification burden, since a board with dozens of layers has far more opportunities for a manufacturing defect to show up only under load. For investors, this is a capacity-constrained, capital-heavy tier of the supply chain: the fabricators that can reliably hit the highest layer counts at volume are a shorter list than the board designers who spec them, and that gap tends to widen, not narrow, with each new rack generation.

AI delta: AI boards need far more routing layers and tighter fabrication tolerances than a standard server motherboard, because dense scale-up interconnect and high-pin-count sockets have to be routed across the same board.

Who makes it

The full entry — sub-categories, news, videos and sources — lives on the High-layer-count PCBs page.