AMD's Lisa Su signs AI hardware deals with Samsung and Naver in Korea
AMD chief executive Lisa Su visited South Korea on the 18th, marking her first trip to the country since taking the helm at the chipmaker. The meetings centered on expanding AI semiconductor collaboration with Samsung Electronics and Naver as global competition for AI hardware intensifies.
In Seongnam, at Naver’s second campus known as 1784, Su and Naver chief executive Cho Seung-hyeon? No, the article states Choi Soo-yeon, the head of Naver, signed a memorandum of understanding to build a robust AI infrastructure using AMD GPUs to support Naver’s HyperCLOVA X AI model. The plan calls for combining Naver’s data-center and cloud operating experience with AMD’s next‑generation semiconductor technology to deploy the system in real service environments. The two companies also said they would provide AI computing resources to academia and pursue joint research.
Later in the day, Su traveled to Samsung’s semiconductor campus in Pangyo, near Seongnam, to meet Jeon Young-hyun, Samsung Electronics’ Device Solutions division head. They signed a separate memorandum focused on next-generation AI memory and computing collaborations. The agreement contemplates integrating Samsung’s sixth-generation high-bandwidth memory, HBM4, into AMD’s Instinct MI455X GPU. Samsung already supplies HBM3E for AMD’s MI350X and MI355 GPUs, continuing a roughly 20-year partnership in graphics memory.
The talks also touched on potential co-foundry or outsourced manufacturing arrangements. AMD has historically worked closely with Taiwan’s TSMC for fabrication, while Samsung has previously produced AMD graphics chips on a 14nm process in 2016. The discussions indicated an interest in broadening collaboration beyond memory and GPUs, though no final manufacturing commitments were announced.
Both executives emphasized the broader industry significance of collaboration for building out AI infrastructure. Su said the next generation of AI infrastructure requires close cooperation across the sector and expressed enthusiasm about combining Samsung’s memory capabilities with AMD’s GPU and CPU technologies. A Samsung vice chairman who joined the talks said the partnership could expand further as a result of the discussions.
That evening, Su and Samsung Chairman Lee Jae-yong dined in Seoul’s Itaewon district to discuss further cooperation. Their meetings come on the heels of high-level industry gatherings, including Nvidia’s developer conference in the United States where Nvidia CEO Jensen Huang met SK Group Chairman Chey Tae-won, underscoring a broader pattern of cross-border collaboration among leading AI hardware players.
On the following day, Su was scheduled to meet Samsung’s Device Experience head No Tae-moon and, in Seoul, Korean government AI officials, including Ha Jung-woo, the presidential office’s AI futures planning chief, and Lim Moon-yeong, a vice chair of the National Artificial Intelligence Strategy Council. The discussions aimed to align AMD’s collaboration with Korea’s AI strategy before Su returns home.
Why this matters beyond Korea: the deals highlight how major players—AMD, Samsung, and Naver—are shaping the AI computing supply chain, from memory technology to GPUs and system integration. For U.S. readers, the announcements signal potential shifts in where AI hardware is designed, manufactured, and deployed. Samsung’s HBM4 memory could influence memory pricing and availability for AI accelerators, while joint initiatives with AMD illustrate ongoing diversification of fabrication and collaboration sources beyond traditional hubs. The involvement of Korean tech firms and government stakeholders also points to a continuing global realignment of AI infrastructure ecosystems that touch U.S. tech supply chains, chip design ecosystems, and AI services.