NVIDIA introduces next-gen Rubin, Rubin Ultra and Blackwell Ultra GPUs and powerful Vera CPUs

NVIDIA recently confirmed its next-gen Ruby GPU architecture, while also unveiling Ruby and Blackwell GPUs and the latest Vera CPUs. This is a significant step for the technology giant, which aims to take the performance and efficiency of AI and data center applications to a new level.

The most important points:

Ruby architecture: The new GPU architecture follows on from the Blackwell platform and is expected to go into mass production in the fourth quarter of 2025.
Rubin R100 GPUs: The first generation of Rubin GPUs, codenamed “R100”, is expected to use 4x reticle design and HBM4 memory.
Vera CPUs: New CPUs called “Vera” will replace the Grace CPUs and will be based on TSMC’s 3nm process.
Performance improvements: NVIDIA’s focus is on improving the power efficiency and AI capabilities of the new GPUs.
Timeline: The Ruby GPUs are not expected to be unveiled before the next GTC in 2025.

More details:

Ruby Ultra: NVIDIA is also planning an “Ultra” variant of Ruby GPUs with 12Hi memory stacks in 8 locations and a 2027 launch.
TSMC CoWoS-L packaging technology: The R100 GPUs are expected to be manufactured using TSMC CoWoS-L packaging technology on the N3 process node.
HBM4 memory: The next generation HBM4 DRAM will power the R100 GPUs and enable up to 16Hi stacks.
Grace Superchip Module: The GR200 Superchip module will include two R100 GPUs and an updated Grace CPU.
Vera CPU: The new ARM CPU solution is based on TSMC’s 3nm process and offers 72 cores.

The announcement of the Ruby GPU architecture and related products marks a turning point for NVIDIA. With a focus on performance, efficiency and AI capabilities, NVIDIA is well positioned to meet the demands of the growing data center and AI market.

While it is still early days, early information suggests that the Ruby architecture will be a significant advance over current products.

Source: NVIDIA via Youtube