View All

Please refer to the English Version as our Official Version.Return

France(Français) Germany(Deutsch) Italy(Italia) Russian(русский) Poland(polski) Netherlands(Nederland) Spain(español) Turkey(Türk dili) Israel(עִבְרִית) Denmark(Dansk) Switzerland(Deutsch) United Kingdom(English)
Japan(日本語) Korea(한국의) Thailand(ภาษาไทย) Malaysia(Melayu) Singapore(Melayu) Philippines(Pilipino)
Africa, India and Middle East
North America
United States(English) Canada(English) Mexico(español)
on July 10th

Samsung wins its first 2nm order to produce AI accelerator chips for Japanese companies

Samsung Electronics announced on July 9th (Tuesday) that it has received an order from Japanese artificial intelligence (AI) company Preferred Networks (PFN), which will use Samsung's 2nm foundry process and advanced chip packaging services to manufacture chips for AI accelerators.

This is the first cutting-edge 2nm chip foundry order announced by Samsung, but the size of the order has not been disclosed.

Samsung has always hoped to take the lead in TSMC's mass production of 2nm technology, defeat the other party at a faster pace, and gain a competitive advantage in the new generation of process nodes.

Samsung stated in a statement that these chips will be manufactured using the Interposer Cube S (I-Cube S) process using Ring Gate (GAA) and 2.5D packaging technology to enhance interconnect speed and reduce size.

Samsung stated that South Korea's Gaonchips Co designed these chips.

It is reported that Preferred Networks was established in 2014, mainly engaged in AI deep learning development, and has attracted significant investment from major companies in various fields, including Toyota, NTT, and Fanuc. It is understood that the reason for cooperating with Samsung is because Samsung has both memory and OEM services, strong comprehensive capabilities and technological accumulation, and can provide a complete solution from high bandwidth memory (HBM) design to production and advanced 2.5D packaging.

Junichiro Makino, Vice President and Chief Technology Officer of Computing Architecture at Preferred Networks, stated in a statement that these chips will be used to manufacture high-performance computing hardware for Preferred Networks to be applied in generative AI technologies such as large-scale language models.
Shopping cart (0 Items)
It is empty.
Compare List (0 Items)
It is empty.

Your feedback matters! At Allelco, we value the user experience and strive to improve it constantly.
Please share your comments with us via our feedback form, and we'll respond promptly.
Thank you for choosing Allelco.

Drag or click to upload file
Upload File
types: .xls, .xlsx, .doc, .docx, .jpg, .png and .pdf.
Max file size: 10MB