NVIDIA shows off its next-gen Vera Rubin Superchip at GTC Washington with two huge GPUs

NVIDIA unveils its next-gen Vera Rubin Superchip at GTC Washington: dual Rubin GPUs, 288GB of HBM4 memory, and a Vera CPU with 88 Arm-based CPU cores.

NVIDIA shows off its next-gen Vera Rubin Superchip at GTC Washington with two huge GPUs
Comment IconFacebook IconX IconReddit Icon
Gaming Editor
Published
2 minutes & 30 seconds read time
TL;DR: NVIDIA unveiled its next-gen Vera Rubin Superchip featuring dual Rubin GPUs with HBM4 memory and an 88-core Arm-based CPU, delivering up to 3.6 Exaflops FP4 inference power in the 2026 NVL144 AI server. The 2027 Rubin Ultra NVL576 platform will boost performance to 15 Exaflops with massive HBM4 memory and enhanced NVLINK connectivity.

NVIDIA has officially shown off its next-gen Vera Rubin Superchip at GTC in Washington, the first time in the flesh, ready with next-gen chips and HBM4 memory for the next dominator in AI chips.

NVIDIA shows off its next-gen Vera Rubin Superchip at GTC Washington with two huge GPUs 102
5

NVIDIA CEO Jensen Huang first showcased the company's next-generation Vera Rubin Superchip, but this is the very first time an actual sample of the motherboard, which NVIDIA calls the Superchip, has been shown. We have a Vera Arm-based CPU joined by not one but two gigantic Rubin GPUs. NVIDIA's motherboard for the Vera Rubin Superchip features LPDDR system memory that gets combined with the on-board HBM4 memory on both Rubin GPUs.

Jensen mentioned that NVIDIA's new Rubin GPUs are back in the NVIDIA labs, which means what he showed off was directly from the semiconductor fab at TSMC in Taiwan. Each Rubin GPU is surrounded with a bunch of power circuitry, with each of the Rubin GPU chips sporting 8 HBM4 sites, and two Reticle-sized GPU dies. The Vera CPU features 88 cores and 176 threads of custom Arm-based CPU cores.

NVIDIA shows off its next-gen Vera Rubin Superchip at GTC Washington with two huge GPUs 105
5

NVIDIA's next-gen Vera Rubin NVL144 AI server platform will debut in 2026, while its monster Rubin Ultra NVL576 AI server drops in 2027 with upgraded Rubin Ultra GPUs, scaling from 144 GPUs to a huge 576 GPUs. Each of the Rubin Ultra GPUs are beefed up to four Reticle-sized chips (double the chips on Rubin) with gigantic 1TB of even faster HBM4E memory on 16 HBM sites.

NVIDIA shows off its next-gen Vera Rubin Superchip at GTC Washington with two huge GPUs 103
5

NVIDIA's upcoming Vera Rubin NVL144 platform will be capable of 3.6 Exaflops of FP4 inference power, and 1.2 Exaflops of FP8 Training capabilities, representing a huge 3.3x increase over the now-shipping GB300 NVL72 AI server.

There's a total of 13TB of HBM4 memory with 75TB of fast memory, which is a 60% increase over GB300 NVL72, with 2 x the NVLINK and CX9 capabilities, rated at up to 260TB/sec and 28.8TB/sec, respectively. NVIDIA's new Vera Rubin NVL144 launches in the second half of 2026.

NVIDIA shows off its next-gen Vera Rubin Superchip at GTC Washington with two huge GPUs 104
5

But the year after, in 2027, NVIDIA will unleash the NVIDIA Rubin Ultra NVL576 AI server platform, which will blow even the next-gen Vera Rubin NVL144. We have 15 Exaflops of FP4 inference power, and 5 Exaflops of FP8 Training capabilities, that represents a crazy 14x increase over the GB300 NVL72, with a whopping 4.6PB (petabytes, not terabytes) of HBM4 memory joined by 365TB of fast memory.

This is an equally as huge 8x increase over GB300 NVL72, with 12x the NVLINK and 8x the CX9 capabilities, rated at 1.5PB/sec and 115.2TB/sec, respectively.