Talks about Samsung Electronics’ intentions to take technological revenge in the transition to 3nm standards have been going on for a long time, and now the publication The Korea Economic Daily reports that preparations for starting production of second-generation 3nm chips are being made in cooperation with at least five customers, including NVIDIA, Qualcomm, IBM and Baidu. Deliveries of profile products in bulk quantities will start in 2024.
The Korean contractor will reportedly produce 3 nm GPUs for NVIDIA. As is well known, Samsung produces graphics solutions from the Ampere family in 8 nm technology for NVIDIA, but their successors of the Ada Lovelace generation have switched to the 4 nm process technology from the competitor TSMC. IBM wants to entrust Samsung with the production of central processing units for servers in 3nm technology, Qualcomm traditionally focuses on orders in the mobile processor segment, and the Chinese giant Baidu is interested in the production of accelerators for cloud-class artificial intelligence systems in 3nm technology. technology .
As you know, as part of 3nm technology, Samsung expects to outperform competitors like TSMC and Intel as it transitions to using Surrounding Gate Transistors (GAA). In June, the company already started mass-producing 3nm products for Chinese customers, but it was a first-generation technology that produces relatively simple solutions to speed up mining. Samsung intends to manufacture structurally more complex components using second-generation 3 nm technology, which should be mastered by 2024. The company plans to start producing 2nm products by mid-decade and to become a leading contract chip manufacturer by the end of the decade.