Core Viewpoint - AMD has signed a memorandum of understanding (MOU) with Samsung Electronics to enhance collaboration in next-generation AI memory and computing technologies, specifically focusing on high-bandwidth memory (HBM4) for its Instinct MI series GPUs [1][2]. Group 1: Partnership and Collaboration - AMD has designated Samsung Electronics as the preferred supplier for HBM4 memory, which will be integrated into the Instinct MI455X GPU accelerator, set to be supplied to cloud service providers and large-scale data centers starting in Q3 [1]. - The partnership aims to leverage Samsung's advanced memory technologies to enhance AMD's competitiveness in the AI sector, with a focus on performance, reliability, and energy efficiency [1][2]. - Both companies have agreed to discuss future collaboration in the foundry sector, indicating potential expansion of their partnership beyond memory supply [2]. Group 2: AI Infrastructure Development - AMD plans to deploy the Helios AI architecture in the second half of this year, utilizing the sixth-generation EPYC processors and customized GPUs based on the Instinct MI450, along with Samsung's high-performance DDR5 memory and HBM4 [2]. - Samsung's DS division head emphasized the shared goals of both companies in advancing AI computing, highlighting the importance of this agreement in broadening their collaborative efforts [2]. - AMD's CEO noted the critical nature of industry collaboration for implementing next-generation AI infrastructure, expressing enthusiasm for integrating Samsung's memory technology with AMD's processing platforms [2]. Group 3: Broader Industry Engagement - Prior to the meeting with Samsung, AMD's CEO discussed plans to strengthen AI infrastructure with Naver, which has decided to utilize AMD's EPYC processors and Instinct MI series GPUs for infrastructure diversification [3].

Core Viewpoint - Samsung Electronics and AMD have strengthened their partnership with a new agreement focused on artificial intelligence infrastructure, responding to the increasing global demand for high-performance data center systems [1][2]. Group 1: Partnership Details - The memorandum of understanding was signed at Samsung's semiconductor campus in Pyeongtaek, South Korea, during a visit by AMD CEO Lisa Su and Samsung Electronics Vice Chairman Young Hyun Jun [2]. - The agreement emphasizes closer coordination between memory and computing technologies, addressing bottlenecks in AI computing, particularly in memory speed, power efficiency, and system integration [2][3]. Group 2: Technological Advancements - Samsung will supply its next-generation high-bandwidth memory, HBM4, for AMD's upcoming Instinct MI455X AI accelerator and develop DDR5 memory for AMD's sixth-generation EPYC processors, codenamed Venice [3][6]. - Samsung's HBM4 is designed to achieve speeds of up to 13 gigabits per second and deliver bandwidth of up to 3.3 terabytes per second, built on a sixth-generation 10-nanometre-class DRAM process [4]. Group 3: Market Position and Strategy - Samsung currently holds a 22% share of the global HBM market, trailing SK Hynix, which leads with 57% [6][9]. - The partnership is seen as a strategic move to strengthen Samsung's position in the competitive AI semiconductor market, particularly in securing long-term supply chains for advanced memory [9]. Group 4: Broader Industry Context - The collaboration aligns with a trend of increased partnerships among chipmakers to enhance the efficiency of AI systems across data centers [12]. - The urgency for such partnerships is driven by significant AI investments from major technology firms, with AMD recently securing a multi-year deal worth up to $60 billion to supply AI chips to Meta Platforms [11].

Behind every impactful AI experience is strong collaboration plus compute. Across open hardware and software, AMD and OpenAI are working together on Instinct MI455X and Helios for training and inference at massive scale, today and into the future. Hear Greg Brockman’s perspective from #CES2026. #CES #AMD ...