SK Group Chairman Chey Tae-won recently met with Nvidia CEO Jensen Huang in the US to discuss potential collaborations on High Bandwidth Memory (HBM) supply and broader artificial intelligence (AI) ventures, according to industry insiders.
Sources indicate the meeting occurred earlier this month at a fried chicken restaurant in California. It is believed that the executives discussed supply plans for HBM4, the next-generation HBM technology anticipated for use in Nvidia’s forthcoming AI accelerator, codenamed “Vera Rubin.”
SK hynix Inc., a crucial HBM supplier to Nvidia, has stated that they are progressing with the mass production of HBM4 according to the agreed-upon timelines with their clients.
“Despite full-scale production, the company might not be able to satisfy 100% of customer demand, potentially allowing competitors to secure supply contracts,” said an SK hynix official. “However, we expect to maintain our leadership and dominant position as a supplier, backed by our product performance, manufacturing capabilities, and quality.”
Samsung Electronics Co., a competitor in the HBM market, is projected to commence HBM4 production later this year.
Industry sources suggest that SK hynix is likely to remain Nvidia’s primary HBM supplier this year, with their estimated share of the HBM4 market around 70 percent.
The discussions also touched upon possible collaboration related to SK Group’s strategy to broaden its AI presence beyond memory chips, encompassing comprehensive AI solutions, sources revealed.
