Nvidia CEO Hints at Partnership With Samsung for AI Memory Chips

Samsung Electronics is poised to make a significant leap in the AI semiconductor market as it edges closer to supplying advanced memory chips to Nvidia, the global leader in GPU technology.

Progress in Certification and Collaboration

At an event at the Hong Kong University of Science and Technology, Nvidia CEO Jensen Huang confirmed that the company is “working as quickly as possible to certify Samsung’s AI memory chips.” Huang noted that Nvidia is evaluating Samsung’s cutting-edge HBM3E (high-bandwidth memory) chips in both 8-layer and 12-layer configurations, highlighting the potential for collaboration between the two tech giants.

HBM chips are vertically stacked DRAM packages designed to enhance the performance of GPUs, particularly for AI-driven applications. Samsung’s fifth-generation HBM3E chips, capable of powering Nvidia’s Blackwell GPUs, represent the pinnacle of this technology.

HBM3E Chips and Their Significance

Nvidia’s Blackwell GPUs, among the most advanced on the market, are essential for powering AI systems, and their cost—ranging between $30,000 and $40,000 per unit—reflects their cutting-edge capabilities. Samsung’s HBM3E chips are expected to complement this advanced hardware, boosting performance for Nvidia’s AI customers.

Currently, SK Hynix, another South Korean semiconductor giant, serves as Nvidia’s primary supplier of HBM chips. SK Hynix began supplying 8-layer HBM3E chips to Nvidia in March and has benefited immensely from the growing demand in the AI sector.

Samsung, however, is aggressively positioning itself as a competitive supplier. It aims to expand its HBM3E sales in the fourth quarter of this year, with these chips projected to account for approximately 50% of Samsung’s total HBM revenue during the same period.

Samsung’s Progress and Industry Outlook

Samsung’s commitment to quality has already yielded results. During the company’s October 31 business report, Kim Jae June, Vice President of Samsung’s memory chip division, confirmed “meaningful progress” in passing critical quality certification stages for a “major customer.” While Kim did not explicitly name Nvidia, the timing and context strongly suggest the graphics chip maker is the client in question.

Samsung’s success in securing Nvidia as a client could reshape the competitive landscape of the HBM market, where it competes with SK Hynix and U.S.-based Micron Technology.

Why This Matters

The partnership between Nvidia and Samsung could bring Samsung closer to dominating the high-bandwidth memory sector. With Nvidia’s GPUs powering the world’s most advanced AI systems, including its Blackwell line, a reliable supply of certified HBM3E chips will be crucial for maintaining Nvidia’s edge in the AI market.

This collaboration also underscores the rising importance of AI hardware innovation, as companies race to meet the growing computational demands of machine learning and artificial intelligence technologies.

Samsung’s anticipated role as a supplier for Nvidia signals not only a potential increase in revenue but also a significant step in becoming a key player in the evolving AI chip ecosystem.