Samsung Electronics and Advanced Micro Devices have signed a memorandum of understanding to deepen their collaboration on next-generation artificial intelligence (AI) memory and computing technologies, the companies said on Wednesday.
The agreement will see Samsung supply its latest high-bandwidth memory (HBM4) chips for use in AMD’s upcoming AI accelerators, including the Instinct MI455X graphics processing units, while also working jointly on advanced DDR5 memory solutions for AMD’s next-generation EPYC server processors.
The signing ceremony took place at Samsung’s semiconductor complex in Pyeongtaek, attended by AMD Chief Executive Lisa Su and Samsung Vice Chairman Young Hyun Jun.
Samsung said the partnership reflects a growing need for tighter integration across the AI computing stack, as demand rises for faster memory and improved power efficiency in data center workloads.
Under the deal, Samsung will align as a primary supplier of HBM4 for AMD’s next-generation AI chips, while also supporting the development of DRAM solutions for sixth-generation EPYC processors, codenamed “Venice.” These components are expected to power large-scale AI systems built on AMD’s Helios rack architecture.
HBM4, Samsung’s latest memory technology, is designed using a sixth-generation 10-nanometer-class DRAM process and a 4nm logic base die. The company said the chips can reach speeds of up to 13 gigabits per second and bandwidth of up to 3.3 terabytes per second, aimed at accelerating AI model training and inference.
AMD said its MI455X GPUs, paired with EPYC CPUs and high-performance memory, will form the backbone of next-generation AI infrastructure, where performance gains increasingly depend on system-level integration.
The companies also said they would explore potential foundry collaborations, with Samsung potentially manufacturing future AMD semiconductor products.
The partnership builds on nearly two decades of collaboration between the two firms, including Samsung’s role as a key supplier of HBM3E memory for AMD’s current Instinct AI accelerators.
The move underscores intensifying competition among chipmakers to secure critical components for AI systems, as demand for advanced computing infrastructure continues to surge globally.