GlobalFoundries Bets on FD-SOI, Plans to Start FDX Fusion Production in Dresden in 2028

GlobalFoundries Bets on FD-SOI, Targeting Robotics and Automotive Sectors Source: Ban Dao Zong Heng Published: 2026-10-30 14:08 AI Data Center Generated Poster The development of data centers and physical AI will bring unprecedented transformation to the entire semiconductor industry.

GlobalFoundries is betting on specialty semiconductor technologies to tap AI market opportunities beyond advanced processors, targeting data center interconnect, power management, automotive electronics, and humanoid robotics to capture demand arising from AI's move into physical applications. The U.S. wafer foundry is extending fully depleted silicon-on-insulator (FD-SOI) technology into the physical AI domain, with plans to put its next-generation FDX Fusion process platform into production at its Dresden, Germany fab in 2028.

This strategy contrasts with the advanced logic process race led by TSMC and Samsung Electronics. AI investment has driven demand for 3nm and 2nm processes, CoWoS advanced packaging, and high-bandwidth memory (HBM), but GlobalFoundries is focused on semiconductor technologies that support the interconnect, power delivery, and control of increasingly complex AI systems. GlobalFoundries believes compute power is no longer the sole bottleneck in AI development. As AI infrastructure continues to expand and intelligent machines enter the physical world, the importance of data interconnect and power delivery systems is steadily rising. GlobalFoundries CEO Tim Breen said the development of data centers and physical AI will bring unprecedented transformation to the entire semiconductor industry.

Interconnect and Power Delivery Become Core Directions

Although the applications discussed by GlobalFoundries executives cover data centers, in-vehicle systems, humanoid robots, and more, the company's strategy revolves around two core needs: interconnect and power delivery.

GlobalFoundries set this course as early as 2018, when it indefinitely suspended 7nm FinFET process development and redirected R&D resources toward FD-SOI, RF, power semiconductors, and other specialty processes. After exiting the leading-edge logic race, GlobalFoundries no longer competes directly with TSMC and Samsung at the 3nm and 2nm nodes, and instead focuses on technologies beyond AI compute, including high-speed data transmission, power management, sensing, control, and low-power computing.

These capabilities are becoming increasingly important as AI training clusters expand from single-rack to multi-rack deployments. The challenge is no longer limited to a single chip; the problem of connecting chips has come to the fore: data traffic between GPUs is surging, transmission distances are lengthening, and rack power density continues to rise, placing higher demands on interconnect solutions and power delivery systems.

This shift creates new opportunities for specialty processes such as silicon photonics, silicon germanium (SiGe), BCD, and FD-SOI. Silicon photonics and SiGe are mainly used for high-speed data interconnect, while BCD processes meet power management needs. These processes, together with advanced logic technologies, jointly support the expansion of AI infrastructure.

GlobalFoundries builds its competitiveness on specialty wafer manufacturing, and changes in market demand allow it to seek new opportunities beyond the most advanced logic nodes. The company continues to expand its footprint in photonics, RF, power devices, automotive chips, and IP businesses. GlobalFoundries says that after acquiring Singapore-based silicon photonics foundry Advanced Micro Foundry (AMF), it has become the world's largest silicon photonics foundry by revenue.

GlobalFoundries is also expanding its physical AI business through its RISC-V solutions and IP subsidiary MIPS. In September 2026, MIPS launched three compute reference platforms for physical AI: Acies, Actus, and Aegis.

FD-SOI Targets Physical AI Scenarios

GlobalFoundries has proposed a STAC framework, with the four letters standing for Sense, Think, Act, and Communicate. The four functions correspond to the hardware requirements of end devices such as vehicles and robots.

GlobalFoundries is using FD-SOI technology to adapt to such applications, with integrating multiple functions on a stable semiconductor platform as a key design focus. It recently unveiled the next-generation FDX Fusion process platform in Dresden, positioning physical AI as a core target market covering automotive electronics, robotics, industrial automation, and edge AI devices.

GlobalFoundries believes the competitive landscape in these markets is completely different from the data center GPU market, leaving room for specialty foundries with differentiated manufacturing technologies. Rather than following TSMC and Samsung in racing toward more advanced logic nodes, GlobalFoundries is pushing FD-SOI into more scenarios that require sensing, control, interconnect, and energy-efficient computing.

Dresden Fab to Handle FDX Fusion Mass Production

GlobalFoundries plans to put FDX Fusion into production at its Dresden fab in 2028, consolidating the site's position as a strategic hub for differentiated semiconductor technologies in Europe. The region has a mature semiconductor industry cluster, bringing together Infineon, Bosch, the European Semiconductor Manufacturing Company (ESMC) led by TSMC, and a large number of semiconductor equipment and materials suppliers.

The concentration of semiconductor supply chain players and proximity to customers will help GlobalFoundries advance the commercialization of the FDX Fusion platform. For GlobalFoundries, the core of this strategy is not about competing for the smallest transistor node, but about expanding FD-SOI applications in automotive, robotics, and other physical AI scenarios. It hopes to integrate sensing, control, communication, and computing capabilities onto a stable process platform to consolidate its position in edge AI wafer manufacturing.

This article is reproduced from media reports or online platforms and represents the author's personal stance or views. Our republication is solely for sharing and does not mean we endorse or agree with it. If the source attribution is incorrect or your lawful rights and interests are infringed, please contact customer service promptly, and we, as a neutral platform service provider, will promptly correct, delete, or handle it in accordance with the law. Comments: No user comments yet. Latest News More Related Recommendations

Samsung bets on Exynos 2700 to capture a higher market share 2026-10-30 15:41

NVIDIA GPUs are no longer the only choice 2026-10-30 15:40

Wolfspeed commercializes transparent silicon carbide technology 2026-10-30 15:11