NVIDIA is sprinting to integrate Samsung’s cutting-edge High Bandwidth Memory 3E (HBM3E) chips into its arsenal. This aggressive push, confirmed by CEO Jensen Huang in a recent Bloomberg TV interview, signals a major development in the high-stakes world of artificial intelligence hardware. The announcement came on the heels of Huang receiving an honorary doctorate in engineering from the Hong Kong University of Science and Technology.
Huang revealed that NVIDIA is actively exploring procurement of both 8-layer and 12-layer HBM3E chips from Samsung. This is a notable shift, considering that currently, NVIDIA primarily relies on SK hynix for the HBM memory crucial to its powerful AI GPUs. Notably, during NVIDIA’s Q3 2024 earnings call, Samsung wasn’t mentioned among the company’s key suppliers – a fact that now stands in stark contrast to their current efforts.
The urgency is palpable. Huang emphasized NVIDIA’s commitment, stating they are “working as fast as they can” to certify Samsung’s HBM3E. This heightened interest aligns perfectly with Samsung’s own recent announcement. During their Q3 2024 earnings call on October 31st, Samsung Electronics Executive Vice President Kim Jae-june reported “significant progress” in supplying HBM3E to major clients, including NVIDIA, and expressed confidence in expanded sales during the fourth quarter. He further highlighted plans to release improved HBM3E chips timed with the rollout of next-generation GPUs from their key partners.
The HBM market has long been dominated by SK hynix, leaving Samsung playing catch-up. Recent leadership changes within Samsung’s semiconductor and foundry divisions have added to the complexities. NVIDIA’s swift action, however, could be a game-changer, potentially propelling Samsung into a more significant player in the HBM3E arena far sooner than initially anticipated.
This collaboration is strategically crucial for NVIDIA. The company is currently selling every AI GPU it produces, a testament to the booming demand for AI processing power. Securing a reliable supply of HBM memory chips from a third major manufacturer, alongside SK hynix and Micron, is not merely beneficial; it’s essential to sustain this momentum and fuel continued growth throughout 2025 and beyond. The move ensures a diversified supply chain, mitigating potential risks associated with reliance on a single or even two major suppliers. This strategic diversification positions NVIDIA for continued dominance in the rapidly evolving landscape of artificial intelligence.
