GlobalFoundries Sets 2028 Production for FD-SOI Platform Targeting Physical AI
GlobalFoundries announced its FDX Fusion FD-SOI platform, designed for Physical AI workloads, with manufacturing planned for 2028 at its Dresden, Germany facility. The first generation is expected to deliver 7nm-class digital performance, RF capabilities up to 0.5 THz, in-memory computing, embedded non-volatile memory, and more than double the density of first-generation FDX. The platform aims to support edge and Physical AI applications while reducing system complexity, latency, and power consumption.
GlobalFoundries says its FDX Fusion platform is aimed at STAC workloads—sensing, thinking, acting and communicating—and will be produced in Dresden starting in 2028. The first version is planned to combine 7nm-class digital logic with RF performance reaching 0.5 THz, analog/mixed-signal functions, low leakage, in-memory computing, embedded nonvolatile memory and dense logic.
It is expected to more than double the density of the company’s first FDX generation. Target uses include edge inference, robotics, industrial automation, autonomous systems, sensor fusion and connected devices. GF is working with early customers and offers design support through MIPS, ARC and RISC-V resources.
If delivered, this platform could affect automakers, robot makers, industrial equipment suppliers, device manufacturers, and consumers. By putting more AI processing at the edge, it may enable faster responses, lower data transmission needs, and more energy-efficient connected products. Workers in factories and logistics could see new automation capabilities, while privacy and reliability questions may grow as more sensing and inference move into physical environments.