Nvidia Leads the Way in HBM4 Adoption as AI Memory Stock Secures 70% of Orders

Key Points

• Nvidia is seeking to address high-bandwidth memory (HBM) bottlenecks as AI workloads scale, focusing on a multiyear partnership with SK Hynix for HBM4 solutions.

• Industry estimates suggest SK Hynix could secure 50% to 70% of Nvidia’s HBM4 orders, significantly impacting its revenue trajectory.

• Three major companies—SK Hynix, Samsung, and Micron Technology—are in the race to produce HBM4 at scale, which is critical for training large AI models.

Nvidia’s CEO, Jensen Huang, highlighted the necessity of advanced memory solutions as the demand for data transfer in AI applications increases. HBM technology, particularly HBM4, is essential for enhancing model sizes and powering next-generation platforms like Nvidia’s Vera Rubin architecture.

With SK Hynix positioned as a favorable partner, the company aims to double its wafer capacity by 2030. This partnership is crucial as AI workloads grow significantly, establishing SK Hynix as a potential leader in the emergent HBM segment.