NextFin

SK Hynix and NVIDIA Accelerating HBM Development, Supply Volume Finalized for 2025

AsianFin — During CES 2025 on Thursday, SK Group Chairman Chey Tae-won emphasized the pivotal role of artificial intelligence (AI) in driving transformation across all sectors. "AI is no longer an option; it's essential. Whether we lead or follow in the AI market will determine the future," Chey stated. He stressed the importance of building foundational AI infrastructure and talent domestically, urging self-reliance over dependency on foreign countries.

Chey also disclosed details of his meeting with NVIDIA CEO Jensen Huang during the event, where they discussed high-bandwidth memory (HBM) and emerging concepts like "physical AI." SK Hynix, a key supplier of HBM to NVIDIA, made significant advancements in this domain last year. In March, the company became the first to supply fifth-generation 8-layer HBM3E products and, by October, began mass-producing 12-layer stacked HBM3E memory—another global first.

Acknowledging past delays in SK Hynix's HBM development compared to NVIDIA's expectations, Chey revealed that NVIDIA had urged SK Hynix to accelerate its pace. Recently, however, SK Hynix has not only closed the gap but has also begun to outpace NVIDIA's development timeline.

Both companies are now intensifying their efforts in HBM research and development. According to Chey, work-level meetings between SK Hynix and NVIDIA have finalized supply volumes for 2025, marking a critical step in meeting growing AI demands.

Explore more exclusive insights at nextfin.ai.

Open NextFin App