Sigurd, Winstek align optical testing push as AI demand shifts toward connectivity

CPO is not a packaging upgrade—it is a fracture in the test paradigm. Once the optical engine migrates onto the same substrate as the switch ASIC, the linear test-die, test-module, test-system sequence collapses entirely. What replaces it is a three-axis parallel architecture: wafer-level photonic integration verification, package-level thermo-mechanical coupling stress testing, and system-level signal integrity validation running concurrently. Sigurd's 2027 revenue-doubling target is a calculated bet on the architecture-transition premium window spanning 2025 to 2027. During this period, LPO, LRO, and CPO coexist across customer roadmaps, forcing OSATs to maintain dual-capability test lines simultaneously. Test service pricing inflates structurally. Winstek's alignment is not a capacity play—it is a yield insurance policy against architectural uncertainty in end-customer designs. The real competitive story lies elsewhere. Advantest and Teradyne dominate SoC test; Keysight and Anritsu own the optical test domain. CPO forces these two worlds into direct collision. The test-bridge capability Sigurd and Winstek are constructing is something neither pure-play semiconductor test houses nor pure-play optical test vendors can replicate independently. Geopolitical exposure remains acute: advanced packaging test equipment stays under US export controls. If China's AI infrastructure accelerates CPO deployment, the test equipment supply chain becomes a new chokepoint. Twelve-to-twenty-four-month outlook: NVIDIA and Broadcom will lock in CPO test specifications by mid-2026, creating a de facto industry standard. The long-tail effect points to accelerated consolidation among test service providers—smaller players lacking dual photonic-electronic stack capability will be absorbed or eliminated.