ready for NVIDIA's next-gen Rubin R100 AI GPU

SK hynix aims to release its HBM4 memory strip in Q4 2024, ready for NVIDIA’s next-generation Rubin R100 AI GPU arriving in 2025.

VIEW GALLERY – 2 IMAGES

In a new report from ZDnet, we learn that SK hynix is nearing the final phase of commercialization of its next-generation HBM4 memory, with design drawings set to be transferred to the manufacturing process, or “tape out.” According to ZDnet’s industry sources, SK hynix plans to complete production of its HBM4 memory for NVIDIA in October, so we’re just a few weeks away.

HBM4 memory delivers a massive 40% increase in bandwidth and 70% reduction in power consumption compared to the world’s fastest HBM3E memory. HBM4 memory density will be 1.3x higher. With all of these advancements combined, the leap in performance and efficiency is a key driver of NVIDIA’s continued AI GPU dominance.

NVIDIA says its next-gen Rubin AI GPUs and their respective platforms will be available in 2026, while a Rubin Ultra AI GPU offering will launch in 2027. NVIDIA also confirmed that its next-gen Rubin AI GPUs will use next-gen HBM4 memory.

We expect the next-gen NVIDIA Rubin R100 AI GPUs to use a 4x reticle design (vs. Blackwell with a 3.3x reticle design) and be manufactured on TSMC’s industry-leading CoWoS-L packaging technology on the new N3 process node. TSMC recently talked about up to 5.5x reticle size chips arriving in 2026, with a 100 x 100 mm substrate that would handle 12 HBM sites, compared to 8 HBM sites on the current-gen 80 x 80 mm packages.

TSMC will be moving to a new SoIC design that will allow for over 8x reticle size on a larger 120 x 120mm package configuration, but as Wccftech points out, these things are still in the planning stages, so we can likely expect around 4x reticle size for the Rubin R100 AI GPUs.