(June 8): Nvidia Corp and SK Hynix Inc have agreed to partner on designing future generations of memory chips for AI, a win for a South Korean leader vying with Samsung Electronics Co in a red-hot arena.
The two companies signed a multi-year agreement covering both chip design and manufacturing. Nvidia will help its partner diversify into new arenas, encompassing infrastructure and physical AI, as well as memory for Vera Rubin, Nvidia’s most powerful accelerator.
The broad tie-up gives SK Hynix a boost as it prepares to expand significantly into the next generation of high-bandwidth memory, known as HBM4. Nvidia chief executive officer Jensen Huang confirmed for the first time last week that it has cleared Samsung, SK Hynix and Micron Technology Inc to supply that product, indispensable for top-end systems.
The trio, which dominates the global market for memory, compete fiercely for a slice of that lucrative business. Set for deliveries in the third quarter of this year, Vera Rubin is now in full production, Huang said during the Computex trade show in Taiwan. The new systems are built around clusters of Vera central processing units and Rubin graphics cores, allied with terabytes of HBM4 in each server system.
SK Hynix’s shares slid 10% on Monday, tracking a broader selloff in Asian tech. Stock in the company and its memory sector rivals have skyrocketed over the past year, driven north by surging chip prices. SK Hynix and its peers are racing to supply HBM to Nvidia, which is in turn scrambling to supply the accelerators that hyperscalers such as Meta Platforms Inc need to train and operate AI services. Memory has emerged as “probably the toughest” bottleneck to resolve for the tech industry, Arm Holdings Plc CEO Rene Haas said last week.
“Together, we will co-develop the next generation of memory for AI factories and support the accelerating global expansion of AI infrastructure — from frontier model training to agentic and physical AI,” Huang said in the statement.
See also: Anthropic calls for AI pause button to let humans take stock
Huang is pushing a slate of products in coming years, and Asian companies — including South Korean firms — will play a critical role. In Taipei, Nvidia’s CEO took the unusual step of hosting a dinner with partners including SK Group chairman Chey Tae-won.
Since arriving in Seoul in past days, Huang similarly dined with several high-profile industry names. He called on gaming studios Krafton Inc and NC Corp, whose endorsement may be key to ensuring widespread adoption of Nvidia’s RTX Spark chip — its foray into a PC sphere dominated by Intel Corp and Advanced Micro Devices Inc.
Nvidia announced a series of tie-ups Monday with big local names other than SK Hynix. The US company will help SK Telecom Co and Naver Corp build AI cloud services, and team up with Doosan Group on robotics.
Uploaded by Liza Shireen Koshy
