Lightmatter, a photonic computing specialist, has joined NVIDIA's NVLink Fusion ecosystem to deliver high-performance optical connectivity for AI infrastructure. The company will provide Co-Packaged Optics and Near-Packaged Optics products compatible with NVIDIA's optical and SerDes technologies. Lightmatter's bi-directional optical link architecture will create a unified platform for semi-custom AI factories whilst reducing fibre and connector requirements by 50%. The technology enables semi-custom XPUs to connect with NVIDIA switch silicon through Lightmatter's products, creating seamless high-bandwidth connectivity for chips from various suppliers. The partnership combines Lightmatter's Passage CPO solutions with NVIDIA's NVLink Fusion architecture. Analysts view this collaboration as a critical milestone in the maturation of co-packaged optics, providing hyperscalers with a blueprint to overcome traditional I/O bottlenecks and scale AI clusters.

Lightmatter has unveiled Guide DR, a high-density laser network interface card that quadruples rack density compared to conventional external laser small form factor pluggables. The liquid-cooled LNIC relocates the light source from the faceplate into the chassis, solving scaling bottlenecks whilst leveraging existing cooling infrastructure. Built to OCP NIC 3.0 dimensions, a single Guide DR module can support up to 51.2 Tbps of aggregate bandwidth. Four modules in a 1 RU switch tray enable 204.8 Tbps of switching bandwidth, eliminating the need for larger 4 RU chassis with liquid-cooled lasers. The module supplies 200 mW of optical power per fibre across up to 64 fibres, driving 256 lanes at 200G each, with zero front-panel footprint. Guide DR will begin sampling in Q4 2026.

Lightmatter, a photonic computing company, has appointed Roy Kim as Vice President of Product to lead deployment of its photonic interconnect platform. Kim brings over 15 years of AI infrastructure experience from Google, AMD and NVIDIA. At Google since 2021, Kim served as Director of AI Infrastructure Product Management. He previously led Data Center GPU Product Management at AMD and spent eight years at NVIDIA in AI and high-performance computing roles. At Lightmatter, Kim will oversee product roadmap, management, marketing and strategy. The appointment follows significant product launches, including the Guide light engine in January and the Passage L20 optical engine in March. Lightmatter's technology aims to eliminate data bottlenecks in AI data centres through photonic interconnects.