NVIDIA Corp. and SK Hynix Inc. have entered a multi-year technology partnership to co-develop next-generation memory for AI infrastructure [1, 2].
This collaboration is critical as the demand for high-performance memory chips grows to support massive AI factories and complex computing platforms. By securing a stable pipeline of advanced memory, NVIDIA aims to sustain its leadership in the AI boom while SK Hynix solidifies its role as a primary supplier of the hardware that powers generative AI.
The agreement was announced June 8, 2024 [4], during a visit by NVIDIA CEO Jensen Huang to SK Hynix facilities in Seoul, South Korea [2]. The partnership specifically targets the development of memory technologies for the NVIDIA Vera Rubin platform and the creation of AI factories [1, 3].
Huang highlighted South Korea's readiness for the next evolution of the industry. "The age of physical AI has finally arrived. And no country is better prepared for it," Huang said [2].
Beyond specific hardware, the deal seeks to establish Asia's largest AI infrastructure [2]. The companies intend to use South Korea as a primary launchpad for "physical AI," a concept that integrates artificial intelligence with physical systems, and robotics [2]. This strategic alignment allows NVIDIA to secure the specialized components necessary for its long-term ambitions while providing SK Hynix with deep integration into NVIDIA's future product roadmaps [1, 2].
“The age of physical AI has finally arrived.”
This partnership signals a shift toward 'physical AI,' where the focus moves from purely digital LLMs to AI that interacts with the physical world. By tying its hardware roadmap to SK Hynix, NVIDIA is mitigating supply chain risks for the high-bandwidth memory required for its next-generation platforms, ensuring that memory bottlenecks do not slow the rollout of AI factories.
