Samsung Electronics is expected to begin full-scale mass production of 2nm in 2025.

Samsung Electronics announced at the Samsung Foundry Forum held at COEX on the 9th that it plans to mass-produce 2nm (SF2)-based AI accelerator semiconductors from Japan's Preferred Networks (PFN) in a 2.5-dimensional (I-Cube S) advanced package in cooperation with domestic DSP company Gaonchips.

This AI accelerator semiconductor is said to be developed by Gaonchips based on Arm, and Preferred is said to be used in the data center for the project to meet the ever-increasing computing demands of large-scale language models in the future.

In particular, Samsung Electronics is reported to begin mass production using the 2nm process in 2025.

The SF2 process is known to be targeted at smartphones and high-performance computing applications.

Preferred Networks is a Japanese artificial intelligence company specializing in deep learning. It develops advanced software and hardware technologies by vertically integrating the AI value chain, from chips to supercomputers and generative AI-based models.

Samsung Electronics succeeded in mass producing foundries based on the world's first 3nm GAA (Gate-All-Around) structure in 2022, and its second-generation 3nm process is also proceeding as planned based on stable performance and yield.