NVIDIA’s New Artificial Intelligence Monster: R100 “Rubin” GPU with HBM4 Memory Coming in 2025!
New information has emerged about Rubin, the new generation GPU architecture developed by NVIDIA to maintain its leadership in the field of artificial intelligence.
Famous analyst Ming-Chi Kuo to the report According to reports, mass production of the R100 GPU, which will be equipped with HBM4 memory technology, will begin in the last quarter of 2025. System and server solutions are expected to enter production in the first half of 2026.
Rubin’s Strengths
- Rubin, which will be developed with TSMC’s N3 production process, will have a more advanced processor architecture compared to the Blackwell B100.
- The GPU, which will be produced with CoWoS-L packaging technology, will have a 4-layer design.
- Rubin, which will be equipped with eight HBM4 memory units, will open new ground in memory bandwidth and capacity.
- NVIDIA also aims for significant improvements in power consumption with the Rubin architecture. In this way, the energy needs of data centers will be reduced.
With the launch of Rubin, a big leap is expected in the field of artificial intelligence. Thanks to its advanced processor architecture, high memory bandwidth and low power consumption, it provides output in many fields from scientific research to autonomous driving systems. He will open a fire.
