Demand for sixth-generation high-bandwidth memory (HBM4) is expected to explode if Samsung Electronics (005930.KS) begins full-scale supply following its selection as a priority supplier for AMD's HBM4. With artificial intelligence (AI) infrastructure demand confirmed in the earnings announcements of major big tech companies such as Google and Intel, the importance of HBM4, which determines server performance, will grow further once AMD begins supplying its AI infrastructure to major big tech firms. Bae Kyung-hoon, Deputy Prime Minister and Minister of Science and ICT, held a surprise meeting with AMD Chief Executive Officer Lisa Su, offering support for cooperation between AMD and Korean companies including Samsung Electronics.
Su unveiled the AI server rack system "Helios" in a keynote speech at "AMD Advancing AI 2026," held in San Francisco on the 23rd local time.
Helios is a system that integrates 72 "Instinct MI455X" graphics processing units (GPUs) dedicated to AI computation and 18 "EPYC (codename Venice)" central processing units (CPUs) that control the entire system into a steel rack structure. Helios contains not only AMD's GPUs and CPUs but also networking and software. AMD introduced Helios at "CES 2026" held in Las Vegas in January this year, and disclosed its specific performance on this day.
AI racks are core infrastructure for developing and running AI models such as large language models (LLMs). Until now, companies such as Lenovo, Dell and HPE have built and sold AI servers based on AMD CPUs, but Helios will be supplied as a rack-level bundle containing all of AMD's various AI chips. As the era of agentic AI, in which AI reasons and executes on its own, arrives, CPUs have become more important, and AMD has decided to sell AI infrastructure centered on its own CPUs. This amounts to AMD declaring a head-to-head contest with Nvidia's rack system "Vera Rubin NVL72/NVL144," which combines the GPU "Rubin" and the CPU "Vera."
AMD has already secured major hyperscalers (operators of large-scale computing infrastructure) and mega AI developers as Helios customers, including OpenAI, Anthropic, Microsoft (MS), Meta and Oracle. Helios will begin shipping at the end of the third quarter and enter full-scale supply from the fourth quarter. Sachin Katti, head of infrastructure at OpenAI, who took the keynote stage that day, said, "We plan to begin deploying Helios at scale from late this year and expand it throughout next year." Anthropic has formed a partnership to introduce a Helios system of up to 2 gigawatts (GW).
Particularly noteworthy is that Samsung Electronics' HBM4 will be mounted in Helios, which enters mass production at the end of the third quarter. Samsung Electronics shipped HBM4 for the first time in the industry in February this year and, a month later, signed a priority supply agreement with AMD on the occasion of Su's visit to Korea. Bart Walker, vice president of the memory business division at Samsung Electronics' DS division, held a briefing at the exhibition hall that day and explained that Samsung has maintained a cooperative relationship with AMD for 20 years since supplying GDDR4 for AMD's Radeon HD 2000 series in 2007, and that this led to the priority supply of HBM4 in March. Following its supply of fifth-generation HBM3E to AMD last year, Samsung has secured the status of a key partner for HBM4 as well.
If Samsung begins full-scale supply of HBM4 to AMD, Samsung Electronics and SK hynix (000660.KS), the two leaders in HBM, can further solidify their standing as memory companies in the AI rack competition. This is because if AMD joins in—following the supply of both companies' HBM4 for Nvidia's AI rack "Vera Rubin"—the HBM shortage will intensify, which could lead to long-term contracts. For Korean memory companies, this can also be expected to have the effect of reducing dependence on Nvidia. While one Nvidia GPU, "Rubin," is connected to 288 gigabytes (GB) of HBM4 memory, AMD's GPU "MI455X" is equipped with 432 GB, 50% more, which also increases the supply volume.
At the AMD event that day, Deputy Prime Minister for Science Bae, who oversees sovereign AI (AI sovereignty) policies such as independent foundation models, made a surprise appearance. Su personally introduced AMD's new products including Helios, while Bae toured physical AI (technology combining AI with the physical world) and cloud servers utilizing AMD technology. The originally scheduled time was 30 minutes, but he left the venue after touring booths of Korean companies such as MangoBoost and Moreh for about an hour. "As AMD expands its share in an ecosystem centered on Nvidia, it is meaningful for Korean companies to participate in this open ecosystem," he said. "Since Korea has strengths in semiconductors, we must work to create Korea's own independent ecosystem while cooperating with AMD, and to induce the participation of global companies."
