NVIDIA CEO Confirms Partnership with Samsung & SK Hynix on HBM AI Memory

Meta Description

NVIDIA CEO Jensen Huang confirms long-term partnerships with Samsung and SK Hynix to develop next-gen AI memory chips — HBM4, HBM5, and beyond.

NVIDIA Strengthens AI Memory Partnership with Samsung and SK Hynix

NVIDIA CEO Jensen Huang has reaffirmed the company’s deep collaboration with Samsung and SK Hynix to advance high-bandwidth memory (HBM) technology for AI chips. The announcement reinforces that both South Korean tech giants are long-term partners in NVIDIA’s journey to power the next generation of artificial intelligence hardware.

During a recent discussion, Huang praised the two companies, noting that Samsung’s strength lies in its diversity, while SK Hynix excels in focus and precision. He expressed complete confidence in their collaboration, saying:

“The strength of Samsung is diversity, and SK Hynix’s strength is focus; I am 100% confident that we will continue to develop HBM4, HBM5, and even HBM97 together.”

Pushing the Boundaries of AI Memory Technology

The partnership focuses on HBM3 and HBM4 memory, which are crucial for AI and data center workloads that demand high-speed processing and massive data bandwidth. NVIDIA has already approved Samsung’s HBM3E chips for use in its latest AI accelerators and is reportedly working with both Samsung and SK Hynix to bring HBM4 to life.

While SK Hynix remains a step ahead in HBM qualification, Samsung is catching up rapidly and is expected to play a significant role in NVIDIA’s future lineup of HBM-based AI processors.

Building the World’s Largest AI Infrastructure

Jensen Huang also highlighted how South Korea’s advanced memory technologies are critical to sustaining NVIDIA’s AI infrastructure business, which he described as “the largest in history.” He emphasized that the success of this ecosystem relies not only on industry giants but also on startups, universities, and research institutions working together under an open AI model ecosystem.

“To empower nations with AI, we need a broad ecosystem driven by innovation from all corners,” Huang added.

Future Outlook: From HBM4 to HBM97

While HBM4 and HBM5 are already on the horizon, Huang’s mention of “HBM97” signals NVIDIA’s long-term commitment to pushing the boundaries of AI computing performance. These advancements could revolutionize how GPUs handle data-intensive tasks, from generative AI to cloud computing and autonomous systems.

As AI continues to evolve, NVIDIA’s collaboration with Samsung and SK Hynix stands as a testament to the power of global partnerships driving technological innovation forward.