AMD Chair and CEO Lisa Su will meet Korean AI chip companies in Seoul on October 8, seven months after her March visit, in a push to expand cooperation with the country's AI and semiconductor ecosystem and loosen Nvidia's grip on the AI semiconductor market.

According to industry sources cited by Chosun Ilbo on September 30, the Seoul meeting will draw Korean neural processing unit (NPU) developers including Rebellions, DeepX and Mobilint. AMD is also coordinating sessions with local AI startups such as Upstage, and Su is scheduled to sit down with Deputy Prime Minister Bae Kyung-hoon, Minister of Science and ICT, and Ha Jung-woo, vice chairman of the Presidential Council on National Artificial Intelligence Strategy.

The visit builds on groundwork laid earlier this year. AMD and the Ministry of Science and ICT first met in Korea in March, then signed a memorandum of understanding in Silicon Valley in July. Under the MOU, the two sides agreed to jointly build and demonstrate AI computing infrastructure that pairs AMD CPUs and GPUs with Korean NPUs, establish AI research hubs in Korea, and develop local talent.

"We plan to create a successful model for open AI computing infrastructure by combining South Korea's AI semiconductor technology with AMD's high-performance computing capabilities and global partnerships," the Ministry and AMD wrote in a recent joint op-ed.

Industry analysts read Su's quick return as an attempt to capture business opportunities in an accelerating AI infrastructure race and to open cracks in what remains a highly concentrated market dominated by Nvidia. Korea offers two advantages at once. The country has seen rapid AI adoption among businesses and individuals, making it a significant consumer of AI infrastructure. At the same time, Samsung Electronics and SK Hynix supply critical memory and other components into that same infrastructure, positioning Korea as both supplier and customer.

For AMD's Korean NPU partners, the alliance offers a route into computing stacks that would otherwise be difficult to reach. The July MOU explicitly links domestic NPU silicon with AMD's CPU and GPU platforms, giving startups such as Rebellions, DeepX and Mobilint a potential path into large-scale AI computing deployments alongside a top-tier accelerator vendor.

Su's visit is also stirring activity in Korea's conglomerate sector. Meetings with major business leaders, including Samsung Electronics Chairman Lee Jae-yong, are reportedly being coordinated. During her March trip, Su met Lee at Seungjiwon in Seoul's Itaewon district and held separate talks with Jun Young-hyun, Samsung Electronics CEO and head of the DS (Device Solutions) Division — the unit that houses Samsung's foundry and memory businesses.

The Samsung connection carries obvious commercial weight. Samsung is a leading supplier of high-bandwidth memory and foundry capacity that AI accelerator vendors depend on, and any deepening of ties between the two companies would touch the supply chain behind AMD's data center product lines.

How far the October meetings move beyond MOU language toward concrete infrastructure projects and commercial deployments will signal whether AMD can convert its Korean partnerships into measurable share gains against Nvidia in the AI computing market.