Nvidia strengthens its AI supply chain amid emerging geopolitical policy risks.

On March 9, NVIDIA Corporation (NASDAQ:NVDA) selected Samsung Electronics and SK Hynix to supply sixth-generation high-bandwidth memory (HBM4) for its upcoming Vera Rubin AI accelerator, leaving Micron Technology out of its supply chain. This decision strengthens Nvidia's next-generation AI platform, which is designed to power advanced data center workloads and large-scale AI models.

High-bandwidth memory is essential for AI accelerators as it enables faster data transfer between processors and memory stacks, a critical factor as demand for AI computing continues to surge. With cloud providers and enterprises investing heavily in infrastructure, Nvidia's Vera Rubin platform is expected to build on its current generation of processors and drive even greater demand for sophisticated memory solutions.

On March 5, reports emerged that the US Commerce Department is drafting regulations that could restrict AI chip exports globally without government approval. The regulations would require Nvidia and other chipmakers to obtain US permission for most AI chip exports, with approval based on the amount of computing power exported. If implemented, the rule could limit Nvidia's industry influence, giving Washington control over foreign AI development. Shipments of Nvidia's latest GB300 GPUs would still face simpler review and possible exemptions.

NVIDIA Corporation is a leading technology company specializing in the design and manufacture of Graphics Processing Units (GPUs) and the development of platforms for artificial intelligence, data centers, gaming, and autonomous vehicles.