Lisa Su, Chair and CEO of AMD, met with the Ministry of Science and ICT to announce plans to establish an AMD AI Center of Excellence (CoE) and hire hundreds of talents this year. The two sides also discussed strategies for building heterogeneous computing infrastructure.
On October 7, the Ministry of Science and ICT revealed that the meeting with Su was a follow-up to the memorandum of understanding (MOU) signed during President Lee Jae-myung's AI summit in San Francisco in July.
During the meeting, Ryu Je-myung, the Second Vice Minister of the Ministry, and Su agreed to proceed with key collaborative tasks, including the establishment of the CoE and the training of graduate talents. Su stated, "We plan to establish a CoE in Korea and bring hundreds of AI researchers to our center."
Following the meeting, Lee Do-kyu, head of the Ministry's Information and Communication Policy Bureau, said, "We are in discussions at the working level with the goal of establishing it this year. Next month, we will discuss specifics about where it will be set up in the operating committee."
Su also outlined plans for collaboration with South Korean startups and university research institutes. She emphasized, "We plan to invest significantly to increase computational capacity so that AI startups and general users can easily utilize AI. As a first step, we partnered with MangoBoost to build cloud infrastructure."
She added, "We recently signed an MOU with an AI research institute and plan to link various activities with universities and startups in Korea."
Vice Minister Ryu noted, "There are many promising university research institutes and startups in Korea, including Seoul National University and MangoBoost. I hope we can actively collaborate."
The two sides also discussed cooperation on building heterogeneous computing infrastructure. As the AI market evolves from learning to inference, the need for efficient integration of various semiconductors and software into AI infrastructure is becoming increasingly important.
Ryu explained, "As the focus of the AI market shifts, heterogeneous computing infrastructure that utilizes various computing resources based on service and purpose is becoming crucial. Since AMD is expanding an open computing ecosystem based on open source and standards, we can create new collaboration opportunities with our companies."
As a result, both parties decided to develop a reference model that combines AMD's CPUs and GPUs with domestic AI semiconductors to establish AI computing infrastructure that can be utilized in the global market.
After the meeting, Su attended a roundtable discussion with domestic AI semiconductor companies. Thirteen companies representing all areas of AI infrastructure, including domestic AI semiconductors (NPUs), software, memory and storage, and networking, participated in the discussion.
The meeting was designed to explore how domestic AI semiconductor companies can link their technologies and services with AMD's global AI ecosystem. During the open discussion, inquiries from AI semiconductor companies continued.
Key NPU companies, such as FuriosaAI, expressed that as inference workloads diversify, various types of chips will be needed and asked AMD about participating in the open ecosystem. In response, Su stated, "It is AMD's philosophy to allow users to easily choose the hardware they want, whether it be GPUs or accelerators, and we are open to collaboration that connects various accelerators."
Amid the growing burden on GPU computations due to the spread of agent AI, Shin Jeong-kyu, CEO of Level Up, inquired about AMD's plans for CPU and GPU optimization. Su replied, "We will support the entire product line, including data center GPUs and PC chips."
However, it was also clarified that AMD is not currently in a position to support large-scale supply to domestic companies. In response to a reporter's question on this matter, Lee stated, "AMD aims to build an open ecosystem related to AI computing infrastructure and has created a platform to secure domestic companies. While we have explored collaboration possibilities, we are not discussing specific chip supply amounts at this time."
On October 7, the Ministry of Science and ICT revealed that the meeting with Su was a follow-up to the memorandum of understanding (MOU) signed during President Lee Jae-myung's AI summit in San Francisco in July.
During the meeting, Ryu Je-myung, the Second Vice Minister of the Ministry, and Su agreed to proceed with key collaborative tasks, including the establishment of the CoE and the training of graduate talents. Su stated, "We plan to establish a CoE in Korea and bring hundreds of AI researchers to our center."
Following the meeting, Lee Do-kyu, head of the Ministry's Information and Communication Policy Bureau, said, "We are in discussions at the working level with the goal of establishing it this year. Next month, we will discuss specifics about where it will be set up in the operating committee."
Su also outlined plans for collaboration with South Korean startups and university research institutes. She emphasized, "We plan to invest significantly to increase computational capacity so that AI startups and general users can easily utilize AI. As a first step, we partnered with MangoBoost to build cloud infrastructure."
She added, "We recently signed an MOU with an AI research institute and plan to link various activities with universities and startups in Korea."
Vice Minister Ryu noted, "There are many promising university research institutes and startups in Korea, including Seoul National University and MangoBoost. I hope we can actively collaborate."
The two sides also discussed cooperation on building heterogeneous computing infrastructure. As the AI market evolves from learning to inference, the need for efficient integration of various semiconductors and software into AI infrastructure is becoming increasingly important.
Ryu explained, "As the focus of the AI market shifts, heterogeneous computing infrastructure that utilizes various computing resources based on service and purpose is becoming crucial. Since AMD is expanding an open computing ecosystem based on open source and standards, we can create new collaboration opportunities with our companies."
As a result, both parties decided to develop a reference model that combines AMD's CPUs and GPUs with domestic AI semiconductors to establish AI computing infrastructure that can be utilized in the global market.
After the meeting, Su attended a roundtable discussion with domestic AI semiconductor companies. Thirteen companies representing all areas of AI infrastructure, including domestic AI semiconductors (NPUs), software, memory and storage, and networking, participated in the discussion.
The meeting was designed to explore how domestic AI semiconductor companies can link their technologies and services with AMD's global AI ecosystem. During the open discussion, inquiries from AI semiconductor companies continued.
Key NPU companies, such as FuriosaAI, expressed that as inference workloads diversify, various types of chips will be needed and asked AMD about participating in the open ecosystem. In response, Su stated, "It is AMD's philosophy to allow users to easily choose the hardware they want, whether it be GPUs or accelerators, and we are open to collaboration that connects various accelerators."
Amid the growing burden on GPU computations due to the spread of agent AI, Shin Jeong-kyu, CEO of Level Up, inquired about AMD's plans for CPU and GPU optimization. Su replied, "We will support the entire product line, including data center GPUs and PC chips."
However, it was also clarified that AMD is not currently in a position to support large-scale supply to domestic companies. In response to a reporter's question on this matter, Lee stated, "AMD aims to build an open ecosystem related to AI computing infrastructure and has created a platform to secure domestic companies. While we have explored collaboration possibilities, we are not discussing specific chip supply amounts at this time."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.