Lisa Su, AMD chief executive officer (CEO and chair), said on the 7th that the company will establish an artificial intelligence (AI) research center in Korea and hire hundreds of AI researchers. It will also invest in computing infrastructure for domestic AI startups. AMD will build a heterogeneous AI computing infrastructure that combines its central processing units (CPU) and graphics processing units (GPU) with a domestic Neural Processing Unit (NPU), and it will push measures to make it a case that can be used in overseas markets.
Su met with Ryu Jemyeong, Vice Minister of the Ministry of Science and ICT, on the morning of the 7th at Four Seasons Hotel Seoul in Jongno-gu, Seoul, and then attended the Korea AI Semiconductor & Startup Roundtable with the heads of 13 domestic AI semiconductor and infrastructure corporations, including FuriosaAI and DEEPX. Domestic corporations proposed linking AMD’s semiconductors and software with their own technologies and entering global markets together.
The meeting is a follow-up to the memorandum of understanding (MOU) signed between the Ministry of Science and ICT and AMD on the occasion of President Lee Jae-myung’s AI summit in San Francisco in July. It was arranged to check the progress of establishing the research center and academia-industry collaboration, and to seek new business and technology cooperation opportunities with domestic corporations.
◇ Pushing to establish the research center within the year… investing in AI talent and computing infrastructure
Meeting with the Vice Minister Ryu, Su said, “We are establishing an AI research center (CoE, Center of Excellence) in Korea and plan to hire hundreds of AI researchers,” adding that it is “to advance cutting-edge AI technologies while supporting the Korean AI ecosystem.” At the subsequent corporate roundtable, Su also outlined a plan for a Korea research hub that will involve hundreds of engineers over the next few years.
The Ministry of Science and ICT and AMD are conducting working-level consultations with the goal of establishing the center within the year. The location of the center has not yet been decided, and AMD plans to invest in hardware and more. Lee Do-gyu, Director General for ICT Policy at the Ministry of Science and ICT, said at a media briefing that day, “There will be another steering committee meeting next month, and we should be able to get a bit more specific then.”
AMD will also increase AI compute resources that domestic startups and others can use. Su said, “We are also investing in installing computing resources in Korea for AI startups and other users,” introducing the first cloud build-out case with MangoBoost. Su also said a recent MOU was signed with the Seoul National University AI Institute and noted, “We will expand cooperation with domestic universities and startups.”
The Vice Minister Ryu said, “The market is expanding rapidly into the inference market,” and added, “It would be good if you could help expand the global AI ecosystem through active and vigorous cooperation with domestic NPU startups.”
◇ Combining AMD semiconductors and domestic NPU… government to push a demonstration project next year
AMD is also cooperating with the Ministry of Science and ICT to build a heterogeneous computing infrastructure that ties together different semiconductors and software. The plan is to build a system that uses AMD’s CPU and GPU together with NPUs developed by domestic corporations, linking it to domestic demonstration projects, and to develop it into a “reference model” that can be used in global markets.
An NPU is a semiconductor specialized for AI computation. When different types of compute devices are used together in one system, the CPU can handle general-purpose tasks, while the GPU and NPU handle AI computations suited to their strengths. The Ministry of Science and ICT explained that the goal of this cooperation is to go beyond boosting the performance of individual chips to jointly optimize semiconductors and software to lower inference expense and power consumption.
Next year, the government plans to push a large-scale demonstration project that spans CPU, GPU, NPU, memory consolidation technologies, and software. Multiple corporations will form consortia to participate, and the government is preparing to select multiple projects through an open call. Through this, it intends to connect domestic corporations’ individual chip supply to real-world experience in system building and operation.
Deputy Minister Lee said, “There are limits to global expansion if we only supply chips, so the idea is that having a full stack and validating it in the field will help overseas expansion,” adding, “Multiple projects will roll out next year through an open call.” Full stack refers to a configuration that includes the necessary technologies from semiconductors to systems and software.
Attendees at the roundtable were ▲FuriosaAI ▲HyperAccel ▲DEEPX ▲Mobilint ▲MangoBoost ▲Panacea ▲Dinoticia ▲More ▲NotaAI ▲Lablup ▲FADU ▲Elice Group ▲Upstage. The attending corporations also discussed cooperation using AMD’s open-source-based AI and high-performance computing software platform ROCm (Radeon Open Compute). ROCm is a software platform that supports AI model development and execution using GPUs and is AMD’s key software ecosystem corresponding to Nvidia’s “CUDA.”
FuriosaAI, HyperAccel, Mobilint, and others met Su at the roundtable and proposed using various types of chips tailored to diverse AI inference tasks. Su indicated an openness to cooperation on consolidation of CPUs with various accelerators. Lablup requested CPU-GPU integrated optimization, while Upstage requested support for software engineers and a domestic joint research hub. HyperAccel proposed a sovereign AI infrastructure that would secure independent AI capabilities by supplementing Korea’s short supply of CPUs and training GPUs through cooperation with AMD.
Su also laid out a direction to expand ROCm support from data center GPUs to PC chips and CPUs. Seeing physical AI as a field that will become very important over the next three to five years, Su explained a plan to prepare hardware and solution designs that customers can easily use together with WorldLabs.
Deputy Minister Lee said regarding specific chip supply plans, “We explored the possibilities for cooperation, but we did not get to the stage of specifying how many chips would be supplied.”
◇ Meeting with Samsung Electronics as well… a Korea trip compressed into one day
Su will also meet with Samsung Electronics representatives that day. Jack Huynh, senior vice president and general manager of AMD’s Computing and Graphics Group, answered a reporter’s question about meetings with Samsung Electronics or SK hynix by saying, “We plan to meet with Samsung Electronics contacts during this visit to Korea.”
Su’s Korea schedule, initially set for two days, was reduced to one day. As the Ministry of Science and ICT meeting originally slated for the 8th was moved up, Vice Minister Ryu attended in place of Bae Kyung-hoon, who is both deputy prime minister and Minister of the Ministry of Science and ICT. Deputy Minister Lee explained, “Lisa Su had an urgent matter in Washington, so the schedule, which was originally two days, was reduced to one.”
In the afternoon, Su will attend “AMD Gamer Day” at Seoul Dragon City in Yongsan-gu, Seoul, to meet Korean gamers and introduce AMD’s gaming technologies and vision.
Before visiting Korea, on the 6th (local time), Su met with supply chain corporations including Foxconn in Taiwan to discuss expanding CPU and GPU production. Su told local reporters that, expecting strong demand to continue over the next few years, AMD is drawing up supply chain plans three to five years ahead and is working closely with memory companies to secure sufficient supply.
By Jeong Du-yong at Chosun Biz
https://biz.chosun.com/en/en-it/2026/10/07/6QHHA5EKDJETHC3ZEIGTYOIEEU