AMD has struck two collaborations in South Korea with NAVER Cloud and AI lab Upstage, deepening its position in the country’s growing market for domestically operated AI infrastructure.

The agreements cover CPU and GPU deployments, early access to upcoming AMD accelerators, and joint software optimisation. They also align with Korea’s broader interest in sovereign AI, where organisations want tighter control over data location, system operations, and model development.

NAVER Cloud plans

NAVER Cloud plans to expand its use of AMD EPYC processors across its AI and cloud services, and expects to deploy AMD’s upcoming sixth-generation EPYC processor, code-named “Venice”.

AMD will also give NAVER Cloud early access to its next-generation Instinct MI455X GPUs for use across NAVER Cloud’s cloud and production environments. Together, the companies plan to optimise NAVER Cloud’s AI services and software stacks on AMD platforms using AMD’s ROCm software.

The collaboration also includes research and development on new AI services and solutions for NAVER customers.

AMD Chair and CEO Dr. Lisa Su framed the partnership as part of a broader national technology agenda.

“Sovereign AI infrastructure plays a critical role in accelerating how nations build and deploy advanced AI,” said Dr. Lisa Su, Chair and CEO, AMD. “Our expanded collaboration with NAVER Cloud brings together AMD’s leadership AI compute platforms and open software ecosystem with NAVER’s cloud and AI capabilities to enable scalable infrastructure and accelerate the next generation of AI in Korea.”

NAVER emphasised scale and openness across infrastructure and services.

“NAVER Cloud is focused on building scalable and open AI infrastructure to support the rapid growth of AI innovation,” said Soo Yeon Choi, CEO, Naver. “Through our collaboration with AMD, we will leverage high-performance AMD compute platforms to enhance our AI and cloud services and deliver new capabilities to customers and developers.”

NAVER Cloud, part of NAVER, provides cloud and platform services. The group has described itself as an early builder of hyperscale large language models and positions NAVER Cloud as a provider across infrastructure, data centres, and AI services.

Upstage expansion

Separately, AMD is expanding a strategic collaboration with Upstage, a South Korea-based AI lab. Under the agreement, Upstage will begin deploying AMD Instinct MI355 GPUs under a multi-phase roadmap over the next year.

Upstage plans to use the deployment to develop and scale its large language models and document-processing engines. It also linked the work to Korea’s sovereign AI initiatives, including contributing to a sovereign AI model as part of a government-led Proprietary AI Foundation Model project.

AMD and Upstage also plan to use ROCm as part of the software environment for model development, aiming to strengthen Korea’s domestic AI ecosystem and broaden sovereign AI infrastructure options.

Upstage CEO Sung Kim said the partnership ties the company’s model development to local infrastructure plans.

“AMD is a cornerstone of global AI computing infrastructure, and this collaboration will be instrumental in both advancing our proprietary Solar LLMs and building Korea’s sovereign AI model,” said Sung Kim, CEO of Upstage. “Pairing that with domestic infrastructure gives us the foundation to make Korea a genuinely competitive force in global AI.”

Su also positioned the Upstage work around compute, software, and domestic AI development.

“Upstage is at the forefront of AI innovation in Korea, building advanced language models and AI engines that serve enterprises, governments and regulated industries,” said Su. “This collaboration brings together AMD Instinct GPUs and the ROCm open software with Upstage’s expertise to advance Korea’s sovereign AI capabilities and deliver the performance, efficiency and open ecosystem needed to accelerate AI innovation.”

Together, the deals give AMD multiple routes into Korea’s AI build-out, spanning cloud infrastructure procurement and demand from model developers. Over the next year, deployments at Upstage and planned upgrades at NAVER Cloud should provide early signals of how quickly AMD hardware and ROCm gain traction in Korean sovereign AI projects.