Jensen Huang: "Three Memory Companies Pass HBM4 Qualification... Vera Rubin Enters Full-Scale Mass Production"
[Jensen Huang's Second Meeting with Kanbu]
Jensen Huang of NVIDIA Visits South Korea "to Coordinate Supply Chains"
Samsung and SK Suggest They Have Secured Qualification to Supply HBM4
"More HBM Needed… Verarubin Enters Mass Production Phase"
[Edaily Reporters Kim So-yeon and Lee Bae-woon] Jensen Huang, CEO of NVIDIA, visited South Korea on the 5th and announced that Samsung Electronics, SK Hynix, and Micron Technology have all secured qualification to supply High Bandwidth Memory (HBM) 4, which will be incorporated into the next-generation artificial intelligence (AI) accelerator "Vera Rubin." Jensen Huang, CEO of NVIDIA, holds a press conference upon his arrival at Gimpo International Business Aviation Center on the 5th. (Photo: Reporter Lee Bae-woon) Upon his arrival at the Gimpo International Business Aviation Center that day, CEO Huang stated, “All three HBM suppliers have completed qualification testing,” adding, “All three companies have begun production and are competing to supply components for Vera Rubin.”
The AI accelerator Vera Rubin, set to be produced in the second half of this year, will be equipped with HBM4, the sixth-generation HBM. CEO Huang indicated that both Samsung Electronics and SK Hynix have passed the qualification tests and have begun supplying the chips. NVIDIA is developing the Grace Blackwell AI accelerator platform and plans to produce the next-generation AI accelerator, Vera Rubin, in the second half of this year.
This marks his return to South Korea for the first time in about seven months since late October of last year. Regarding his visit to Korea, he explained, “The main purpose is to coordinate the supply chain,” adding, “This is because we need to produce massive quantities of HBM memory and AI chips.” CEO Huang stated, “The ‘Grace Blackwell’ system is operating smoothly, and ‘Vera Rubin’ has entered full-scale mass production,” noting, “We will be very busy in the second half of the year.”
Regarding the future outlook for HBM, he said, “We will be using HBM on a massive scale,” adding, “Korea’s strength is undoubtedly memory.” He emphasized, “We will continue to collaborate with our Korean partners to secure as much supply as possible in a stable manner.” Demand for AI memory is surging as investment in AI infrastructure expands. With the growing need for various high-performance, high-capacity memory types—including not only HBM but also low-power DRAM and graphics DRAM—CEO Hwang’s visit to Korea is seen as an effort to ensure stable supply chain management.
He also mentioned that a surprise gift was prepared for Korean companies. When asked, “Do you have a gift for Korea this time as well?” CEO Huang replied, “I’ve brought a great deal of business for Korea.”
The biggest bottleneck for AI servers is memory. As the KV cache—where large language models (LLMs) store past computations—accumulates, the required memory capacity increases exponentially. This prob…
Concerns about “tax risks” have been raised in some quarters of the financial investment, pharmaceutical, and biotech markets regarding Genosco, a subsidiary of OSCOTEC Inc.(039200)specializing in new…
As July began (June 29–July 3), the pharmaceutical and biotech industries turned their attention to Celltrion Pharm Inc.’s large-scale investment in production facilities and AriBio’s successful fundr…