India, Oct. 22 -- Nvidia plans to expand its supply chain for optical connectivity used in artificial intelligence (AI) data centres, with manufacturing partner Siliconware Precision Industries (SPIL) targeting a 15-fold increase in optical-engine capacity by mid-2027. Announced at Nvidia's GTC event in Berlin on 21 October 2026, the plan reflects growing demand for faster, higher-bandwidth connections between processors and networking equipment in large-scale AI infrastructure.

The expansion is aimed at supporting co-packaged optics (CPO), a technology that brings optical connectivity closer to switching silicon to help address the bandwidth, power and scaling constraints of conventional data-centre interconnects. Nvidia and SPIL are al...