Lightmatter and GUC Partner to Produce Co-Packaged Optics (CPO) Solutions for AI Hyperscalers

About this update from Global Unichip Corp.

Collaboration combines GUC’s ASIC design leadership with Lightmatter’s industry-leading 3D CPO platform to redefine AI infrastructure scalability MOUNTAIN VIEW, Calif. & HSINCHU, Taiwan, January 27, 2026--(BUSINESS WIRE)--Lightmatter®, the leader in photonic interconnect solutions for AI, and Global Unichip Corp. (GUC), the Advanced ASIC leader and a key enabler for hyperscaler AI infrastructure, today announced a strategic partnership to bring commercial Passage™ 3D Co-Packaged Optics (CPO) solutions to market. The collaboration integrates Lightmatter’s revolutionary Passage photonic interconnect with GUC’s state-of-the-art ASIC design services and advanced packaging expertise. This joint solution is specifically engineered to overcome the critical connectivity bottlenecks currently limiting the scaling of next-generation AI and High-Performance Computing (HPC) workloads for the world’s largest data center operators. "The fundamental architecture of computers is changing. The world has hit a wall in per silicon area performance, the network is becoming the computer, and that network needs to run on light. GUC has unmatched ASIC engineering execution in this rapidly-evolving landscape. Pairing their silicon expertise with our photonic interconnects provides the industry with a concrete path to escape the energy and performance traps of legacy signaling, enabling the massive scale-up required for tomorrow’s intelligence," said Nick Harris, founder and CEO of Lightmatter. Redefining AI Interconnect with Passage The integrated solution will leverage GUC's advanced node chiplet and packaging workflows to incorporate Passage, a silicon photonics-based platform that sets a new standard for AI interconnect performance. Passage delivers unprecedented bandwidth density and power efficiency for XPU and switch chip-to-chip communications, pushing the performance boundaries of world's largest and most complex foundation models. These advantages mark a decisive evolution beyond existing solutions, which are fundamentally constrained by physical I/O limits—or ‘shoreline’—of the chip, restricting the maximum bandwidth and radix per optical engine (OE). By seamlessly extending the scale-up domain of AI clusters across multiple racks, the Passage platform substantially improves training time and token throughput for the next wave of frontier AI model...

View stock analysis, news, and events for Global Unichip Corp.