Samsung’s next generation HBM4 memory is moving closer to deployment in Nvidia’s upcoming Rubin artificial intelligence accelerators, with a South Korean report claiming that Samsung Electronics has recently passed the final quality test for HBM4 from Nvidia plus AMD, and will begin mass production next month. The sixth generation High Bandwidth Memory, built on a 4 nm node process, reportedly completed development in 2025 and is now expected to feed directly into Nvidia’s roadmap. Industry sources suggest that this verification could allow initial Samsung HBM4 batches to be demonstrated at the GTC 2026 artificial intelligence conference, scheduled for March 16-19, where they would be immediately used in performance demonstrations of next generation Rubin accelerator products.
The report frames HBM4 as a chance for Samsung to regain ground in a market where it was seen as trailing rivals during the HBM3 and HBM3E era, even as Nvidia’s current Blackwell artificial intelligence GPUs continue to rely on HBM3E. At CES 2026, Nvidia announced an upgrade of Vera Rubin memory bandwidth, and it is not officially clear whether this jump from 20.5 TB/s (September 2025 spec) to 22 TB/s was enabled by the utilization of Samsung HBM4 modules. According to SBS Biz, Samsung’s HBM4 is said to have the industry’s highest specifications over the closest competition, operating at 11.7 Gb/s, which is faster than the 10 Gb/s (JEDEC standard) required by Nvidia and AMD, and the publication reports that Samsung passed customer verification without requiring a redesign despite a request for increased performance last year.
SBS Biz’s anonymous sources also indicate that full scale supply will commence by June time, tying HBM4 availability closely to Nvidia’s Vera Rubin manufacturing schedule. This aligns with expectations from Quanta Computer president Mike Yang, who has anticipated first shipments of Nvidia Vera Rubin artificial intelligence server hardware heading to hyperscaler customers starting around late summer. Taken together, the reported qualification, performance figures, and production timing point to Samsung playing a central role in the memory subsystem of Nvidia’s next wave of artificial intelligence accelerators, with GTC 2026 likely to be the first high profile showcase of the collaboration.
