 | | 0 |
| Samsung Electronics Chairman Lee Jae-yong and AMD CEO Lisa Su hold a dinner meeting in Seoul on March 18 to discuss AI and semiconductor cooperation. |
Samsung Electronics has agreed to prioritize supplying next-generation HBM4 memory to Advanced Micro Devices (AMD), while also exploring broader cooperation including potential foundry production.
Samsung Chairman Lee Jae-yong and AMD CEO Lisa Su held a dinner meeting in Seoul, signaling an expansion of strategic collaboration in artificial intelligence and semiconductor technologies.
The meeting followed Samsung’s announcement that it will supply HBM4—high-bandwidth memory critical for AI computing—to AMD. The two companies are expected to deepen cooperation across AMD’s next-generation AI roadmap.
AMD plans to adopt Samsung’s HBM4 for its upcoming AI accelerator, the Instinct MI455X GPU, used in data center environments. The partnership also includes collaboration on high-performance DDR5 memory solutions to enhance AMD’s EPYC server CPUs and rack-scale AI platforms.
Samsung said the companies also agreed to discuss foundry cooperation, raising the possibility that AMD could outsource production of future chips to Samsung rather than relying solely on other manufacturers.
Such a move could significantly boost Samsung’s system semiconductor and foundry business, where it has been seeking to expand market share. Industry observers say a potential foundry deal with AMD could improve utilization rates and accelerate the division’s path to profitability, with expectations of a turnaround as early as 2027.
The partnership could also impact the competitive landscape in the GPU market, currently led by Nvidia, with AMD seeking to narrow the gap through stronger AI hardware capabilities supported by advanced memory solutions.
“Samsung and AMD share a common goal of advancing AI computing,” said Vice Chairman Jun Young-hyun, adding that Samsung’s integrated capabilities—from HBM4 to advanced foundry and packaging—position it as a key partner in AMD’s AI strategy.
Su emphasized the importance of industry-wide collaboration, saying she was “very pleased” to combine AMD’s GPU and CPU technologies with Samsung’s advanced memory leadership.
Separately, Su also met with Naver CEO Choi Soo-yeon to sign an agreement aimed at expanding the AI ecosystem, including building GPU infrastructure optimized for Naver’s HyperCLOVA X large language model.