Fabrinet is already “through the
qualification gate” for demanding optical programs, so incremental datacenter/telecom demand tends to flow to proven builders. The upside is non-linear when capacity comes online into a tight supply environment:
utilization stays high, mix shifts to harder builds/tests, and automation/AI process control keeps
yield stable. We assume some
multiple compression (manufacturing cyclicality + customer bargaining power), but not a full collapse given scarcity of scaled optics manufacturing execution.