Samsung Electro-Mechanics and Nvidia Expand AI Substrate Production

Samsung Electro-Mechanics said on the 21st that it has formed a strategic partnership with Nvidia and will jointly invest to expand production capacity for artificial intelligence (AI) semiconductor substrates. Starting this year, Samsung Electro-Mechanics plans to supply Nvidia for multiple years with package substrates for next-generation AI semiconductors and high-speed networks.

Samsung Electro-Mechanics will expand Nvidia-dedicated production lines at its key domestic facilities in Sejong and Busan. The products will include substrates for Nvidia’s next-generation computing chipsets and package substrates for network switches, which enable high-speed data transfer between chips inside data centers. Supply is scheduled to ramp up this year and continue for several years.

Samsung Electro-Mechanics expects its Nvidia supply volume in the package substrate business to grow by double digits or more annually over several years. It expects Nvidia-related orders to account for a substantial share of its total annual package substrate production capacity by 2028. Nvidia’s strategy is to secure supplies of advanced package substrates early to ease bottlenecks in the global AI infrastructure supply chain.

Samsung Electro-Mechanics said yield on large, multilayer AI package substrates depends on fine circuit fabrication that prevents signal loss and flatness technology that limits warpage caused by heat. It also said a demanding process that embeds multilayer ceramic capacitors (MLCCs) and silicon capacitors directly inside substrates is needed to control power noise and improve efficiency during high-speed computing. Samsung Electro-Mechanics CEO Jang Deok-hyun said the company aims to combine its FC-BGA manufacturing capabilities with its high-reliability MLCC technology and serve as a total-solution provider supporting the timely delivery of Nvidia AI platforms.

Samsung Electro-Mechanics said on the 21st that it has formed a strategic partnership with Nvidia and will jointly invest to expand production capacity for artificial intelligence (AI) semiconductor substrates. Starting this year, Samsung Electro-Mechanics plans to supply Nvidia for multiple years with package substrates for next-generation AI semiconductors and high-speed networks.

Samsung Electro-Mechanics will expand Nvidia-dedicated production lines at its key domestic facilities in Sejong and Busan. The products will include substrates for Nvidia’s next-generation computing chipsets and package substrates for network switches, which enable high-speed data transfer between chips inside data centers. Supply is scheduled to ramp up this year and continue for several years.

Samsung Electro-Mechanics expects its Nvidia supply volume in the package substrate business to grow by double digits or more annually over several years. It expects Nvidia-related orders to account for a substantial share of its total annual package substrate production capacity by 2028. Nvidia’s strategy is to secure supplies of advanced package substrates early to ease bottlenecks in the global AI infrastructure supply chain.

Samsung Electro-Mechanics said yield on large, multilayer AI package substrates depends on fine circuit fabrication that prevents signal loss and flatness technology that limits warpage caused by heat. It also said a demanding process that embeds multilayer ceramic capacitors (MLCCs) and silicon capacitors directly inside substrates is needed to control power noise and improve efficiency during high-speed computing. Samsung Electro-Mechanics CEO Jang Deok-hyun said the company aims to combine its FC-BGA manufacturing capabilities with its high-reliability MLCC technology and serve as a total-solution provider supporting the timely delivery of Nvidia AI platforms.