Samsung Foundry has reportedly received an order from Japan's Preferred Networks (PFN) to make 2nm AI chips. The South Korean firm failed to attract chip clients for its 3nm semiconductor chip fabrication process, and most big chip firms went with TSMC's 3nm and 4nm processes. However, this seems like a small win for Samsung Foundry.
Contract for making 2nm AI chips for PFN is a win for Samsung Foundry
According to a report from KED Global, the Japanese firm Preferred Networks Inc. has offered a contract to Samsung Foundry to make 2nm-based AI chips. Under this deal, Samsung will make AI accelerators and other AI chips for PFN. The Japanese firm has been collaborating with TSMC since 2016, but it seems to have switched to Samsung Foundry for its next-generation AI chips. Apparently, PFN switched to Samsung, as the South Korean firm offers a more comprehensive chip manufacturing service that also includes stages of chip design, optimization, and packaging. PFN gets a newer chip fabrication technology for its chips, while Samsung Foundry now seems to have an actual client for its newer fabrication process.
PFN was started in 2016, and it makes AI and deep learning-related technologies. The Japanese firm has received large-scale investments from the likes of FANUC (Fuji Automatic Numerical Control), NTT (Nippon Telegraph and Telephone), and Toyota.
TSMC reportedly still has more clients interested in its 2nm chip fabrication process. However, gaining access to PFN's order could signal a positive sign towards Samsung Foundry's 2nm process. Earlier this week, Arm announced its partnership with Samsung Foundry for a more optimized production of next-generation Cortex-A and Cortex-X series processors. Samsung plans to complete the development of its second-generation 3nm fabrication process by mid-2024, and the first chips using that process will most likely be the Exynos 2500 that will be used in the Galaxy S25 series early next year.
