KYOTO, JAPAN — IBM introduced a new chip architecture called "nanostack" at the 2025 IEEE Symposium on VLSI Technology and Circuits in Kyoto, Japan. The architecture vertically stacks transistors in a staggered layout, allowing for nearly 100 billion transistors on a chip the size of a human fingernail.
IBM describes the technology as the "world's first sub-1 nanometer chip technology" for AI data centers and designates it as the 0.7-nanometer node, also known as the 7 angstrom node. The basic unit comprises two transistors stacked and bonded together. Each transistor is composed of three nanosheets, each 5 nanometers thick, equivalent to about 15 rows of silicon atoms. A distance of approximately 9 nanometers separates each nanosheet.
The nanostack architecture provides nearly twice the transistor density of IBM's previous generation of chip technology. IBM projects the architecture could offer 50 percent higher computing performance and 70 percent greater energy efficiency compared to its previous 2-nanometer node chips. Jay Gambetta, director of IBM Research and an IBM Fellow, stated, "It's not just an incremental step, it's a meaningful leap forward."
IBM researchers demonstrated a 40 percent improvement in scaling for static random-access memory (SRAM) using the nanostack architecture. SRAM bit cells are memory storage units consisting of six transistors. The improvement is achieved through a staggered-channel design that reduces overall cell height by 40 percent. Gambetta said, "This achievement of 40 percent will eventually industrialize itself in AI workflows, which require higher bandwidth and high efficiency."
Huiming Bu, vice president of IBM Semiconductors Global R&D and IBM Research, commented on the architecture. Bu stated, "Nanosheet has become the foundation of the next generation of transistor scaling." Bu noted that nanosheet technology is currently adopted by leading foundries for most 3-nanometer chips and all 2-nanometer chips. Bu said, "It will replace nanosheet as today's mainstream in leading foundries, whether it's CPUs or GPUs." Bu further predicted, "Within a decade, this will become another mainstream that we have invented and helped industry to transform."
IBM does not manufacture commercial chips for AI data centers or consumer devices. The company has a partnership with semiconductor firm Rapidus in Japan for the mass manufacturing of its previous 2-nanometer node chips. IBM also has a collaboration with Samsung in South Korea to commercialize related chip technology.
forum Comments (0)
No comments yet. Be the first to comment.