GlobalFoundries has officially unveiled FDX Fusion at its facility in Dresden, Germany, introducing a brand-new generation of its fully depleted silicon-on-insulator (FD-SOI) technology explicitly tailored for the burgeoning market of physical artificial intelligence. The major strategic move is built on the premise that the next wave of high-growth semiconductor demand will diverge sharply from the massive data center GPU race that has dominated the technology landscape in recent years.
Unveiled on October 9, FDX Fusion is engineered for edge systems and autonomous machines that are required to continuously sense, decide, communicate, and act, rather than simply perform heavy computational tasks. The initial generation of the platform is projected to deliver 7-nm-class digital performance while achieving more than twice the transistor density of GlobalFoundries’ original FDX technology. Furthermore, the architecture is designed to seamlessly integrate radio frequency (RF), analog, mixed-signal capabilities, embedded nonvolatile memory, in-memory computing, and dense logic onto a single platform.
To bring this technology to market, GlobalFoundries plans to invest roughly $1 billion over the coming years toward the development of FDX Fusion. This substantial financial commitment includes an initial allocation of about $250 million dedicated to the initial capability toolset. According to company timelines, demonstrator silicon is targeted for early 2027, with an initial process design kit scheduled to follow in the middle of 2027.
The high-profile technology launch coincided with celebrations in Dresden for the topping-off of SPRINT, a major fab expansion project backed by the European Chips Act. This expansion is actively adding cleanroom, laboratory, and manufacturing capacity at the German site, which GlobalFoundries intends to utilize heavily for advanced, differentiated semiconductor solutions.
For Manfred Horstmann, senior vice president and general manager of European fabs at GlobalFoundries, the introduction of the FDX Fusion architecture highlights a stark structural difference between the silicon required for massive server farms and the chips needed when artificial intelligence is deployed directly into the physical world through cars, robots, drones, industrial equipment, and other autonomous systems.
“Data center AI is extremely power hungry,” Horstmann noted during an exclusive interview ahead of the official announcement. “Physical AI is a different class of problem—and it needs more than compute. It needs sensors, actuators, communications, low power, and cost efficiency.”
This philosophy underpins what GlobalFoundries describes as STAC—an acronym standing for sense, think, act, and communicate. Rather than optimizing its architecture primarily for maximum digital compute density, FDX Fusion is purposefully designed to pair substantial processing power with the comprehensive array of functions that a modern machine requires to interact dynamically with its surrounding environment.
From FD-SOI to FDX Fusion
To understand the engineering leap behind the new platform, it is helpful to examine the foundational mechanics of FDX, which serves as GlobalFoundries’ proprietary implementation of FD-SOI technology. In this architecture, the transistor channel sits inside an ultra-thin silicon film measuring approximately 70 angstroms, or 7 nanometers, positioned directly above a buried oxide layer. Because this silicon layer is exceptionally thin, the underlying channel remains inherently fully depleted during operation.
“You don’t have to waste energy depleting the channel,” Horstmann explained, pointing out that this unique physical characteristic provides FDX with many of the same electrostatic advantages traditionally associated with FinFET architectures, but without the manufacturing complexity of physically constructing a silicon fin. “A lot of the functionality comes from the substrate itself. So we don’t need to build a fin to get a fully depleted transistor.”
GlobalFoundries is developing the specialized strained-silicon substrate in close collaboration with CEA-Leti and Soitec through a European research initiative that Horstmann identified as the “Fast and FAMES” project. The industrial partners have already successfully established a pilot manufacturing line in France dedicated to the substrate. This specialized material utilizes a strained 70-angstrom silicon film engineered to boost electron mobility by approximately 25%. Horstmann emphasized that this ongoing collaboration builds upon the same trusted research consortium that previously partnered with GlobalFoundries to develop its successful 22FDX platform.
This new substrate forms a critical component of GlobalFoundries’ broader strategy to achieve 7-nm-class operational performance utilizing a significantly simpler manufacturing process flow compared to conventional, highly complex advanced FinFET technologies.
“For 7-nm-class performance, we don’t need EUV,” Horstmann stated, referring to extreme ultraviolet lithography systems that are traditionally required for leading-edge node manufacturing.
GlobalFoundries anticipates that the initial FDX Fusion process will require roughly 50 masks in total. In contrast, Horstmann cited that advanced 2-nm and 3-nm FinFET processes typically demand between 70 and 80 masks. By relying on the unique properties of the substrate, the company expects to streamline manufacturing complexity, dramatically reduce the number of potential points for defects, and substantially bolster both production yields and overall cost-efficiency.
Rather than turning to EUV tools, GlobalFoundries plans to leverage advanced ArF immersion lithography, tighter overlay control protocols, and ultra-high-precision photomasks supplied by Dresden-based mask manufacturer AMTC. Horstmann noted that the newest immersion systems from ASML deliver substantial improvements in overlay and critical-dimension control. Simultaneously, multi-electron-beam-written masks produced by AMTC provide roughly 30% better overlay and critical-dimension margins when compared against standard industry masks.
Physical AI Needs More Than Logic
This deliberate process strategy is critical because GlobalFoundries is not positioning FDX Fusion merely as a lower-cost alternative pathway to smaller digital transistors. Instead, the platform is purposefully engineered to blend dense logic circuitry with the diverse analog, radio frequency, memory, and power management functions that physical AI applications demand.
“You can use the bulk area for high-voltage devices, and the FD-SOI area for dense logic,” Horstmann said. “That makes integrating these different functions much simpler.”
Company leadership emphasized that FDX Fusion will successfully combine ultra-dense digital logic with high-performance RF capabilities, robust analog and mixed-signal functions, low-leakage operational profiles, embedded nonvolatile memory, and advanced in-memory computing architectures.
According to projections shared by Horstmann, the platform is expected to achieve logic densities ranging from approximately 14 million to 18 million gates per square millimeter, all while preserving the exceptional low-power consumption and integration traits that defined the preceding FDX architecture.
“You maintain the advantages of 22FDX—extremely low power and very good integration with sensors and actuators,” he observed. “Extremely low power is what you need for mobile autonomous systems and drones.”
A Different AI Opportunity for Europe
Europe faces an undeniable and uncomfortable reality within the broader landscape of artificial intelligence semiconductors: a vast majority of the established data center ecosystem and its supporting supply chains are heavily concentrated outside the continent.
“The data center AI train has left the station,” Horstmann acknowledged. “लेकिन [However], for physical AI, the world is still open.”
His core argument is that Europe does not need to compete directly in a race it has already missed. Physical AI naturally rewards an optimized combination of low-power computing, environmental sensing, secure communications, analog processing, and advanced power electronics. This unique technological blend aligns far more closely with Europe’s established, historical strengths in the semiconductor sector.
The city of Dresden itself provides a strong industrial anchor for this forward-looking regional strategy. The local ecosystem already hosts major manufacturing footprints from GlobalFoundries, Infineon, and Bosch, alongside specialized research organizations, critical supply chain partners, and the upcoming ESMC manufacturing joint venture led by Taiwan Semiconductor Manufacturing Company.
“The cluster is becoming self-propelling,” Horstmann said. As the regional industrial hub continues to expand, specialized equipment suppliers establish local service capacity, maintenance teams station themselves nearby, and independent design houses are naturally drawn to the geographic cluster.
Even the entry of ESMC into the market, despite operating as a competitor, ultimately strengthens the local ecosystem by increasing Dresden’s overall manufacturing scale and making the region a significantly more attractive destination for global equipment and material suppliers.
For Horstmann, the powerful combination of established manufacturing scale and highly differentiated proprietary technology creates a compelling window of opportunity.
“This is a big opportunity for Europe to define a new market,” he concluded, “and to be first.”
Leave a Reply