GlobalFoundries Unveils FDX Fusion 7nm FD-SOI Platform for Physical AI: Design Strategy and Roadmap Analysis
Publicado originalmente como: Manfred Horstmann: GlobalFoundries Bets on FDX Fusion for Physical AI

Resumen con IA
GlobalFoundries' FDX Fusion platform gives hardware architects a new 7nm-class FD-SOI option for high-density edge AI and robotics design, bridging low-power computing with advanced RF and mixed-signal integration.
GlobalFoundries has officially introduced its FDX Fusion technology platform at its Dresden facility, establishing a strategic roadmap for fully depleted silicon-on-insulator (FD-SOI) scaling targeted at physical AI systems. Designed to handle demanding STAC (Sense, Think, Act, Communicate) workloads, the platform targets 7nm-class digital performance while packing extensive analog, mixed-signal, RF capabilities up to 0.5 THz, and embedded non-volatile memory into a unified process.
For hardware design teams, the core advantage of FDX Fusion lies in its ability to condense multi-chip architectures into monolithic or highly integrated solutions. By supporting over 2x density improvements compared to first-generation FDX alongside ultra-low leakage and high-temperature tolerance, the platform addresses major bottlenecks in edge inference, autonomous machinery, next-generation radar, and humanoid robotics where thermal management and board space are severely constrained.
Volume manufacturing for FDX Fusion is slated to kick off in 2028 at the Dresden fab, backed by the ongoing SPRINT expansion framework. Hardware and silicon engineering leads should monitor early process design kit (PDK) releases targeted for mid-2027 and demonstrator silicon evaluations to map out future platform migrations for high-reliability edge applications.
Preguntas y respuestas
FDX Fusion is a next-generation fully depleted silicon-on-insulator (FD-SOI) technology platform designed for physical AI, combining 7nm-class digital performance with dense analog, mixed-signal, embedded memory, and high-frequency RF capabilities.

