Test report DSG-7457 · Rev B · tested November 17, 2026

Processors & AcceleratorsDevice under test

NVIDIA Opens SC26 With Rubin HPC GPU, Nemotron for Science

At SC26 in Chicago, NVIDIA launched Nemotron for Science, the Rubin HPC GPU and expanded Vera CPUs, plus a quantum push with IBM and Fermilab. Rubin lands in 2028.

Read
3 min
Words
637
Node
28nm
Operator
Grace Kim

Spec summary

  1. NVIDIA announced the Rubin HPC GPU with 50,000+ CUDA cores, about 3x Hopper, expected in 2028
  2. A KTH-built GROMACS agent fixed 81% of broken test setups in one overnight DGX Spark run
  3. CUDA-Q's TCBP-JOBS decoding algorithm cuts error-correction research computation from 12 weeks to 45 minutes
  4. DKRZ selected Vera CPUs for its HLRE-5 supercomputer, due in early 2028
  5. Magic state cultivation, simulated on JUPITER, cuts operations per logical step tenfold
Ian Buck Kicks Off NVIDIA’s Week at SC26 as Science Enters the Age of Agents
Fig. AIan Buck Kicks Off NVIDIA’s Week at SC26 as Science Enters the Age of Agents — launchpresso / pixabay

NVIDIA used its SC26 special address in Chicago on Monday to launch three products at once: the Nemotron for Science agent toolkit, the Rubin HPC GPU and an expanded Vera CPU platform for high-performance computing. Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA, also announced a quantum computing collaboration with IBM and Fermi National Accelerator Laboratory.

"The frontier of computing and the new scientific stack is going to be simulation, as always, but combined with AI and the future of quantum," Buck said. Simulation "will always be on the top," he added, with AI, quantum and agents working alongside it.

Buck framed the moment with a look back. He gave his first public talk on CUDA (NVIDIA's GPU programming platform) at an SC06 tutorial in Tampa, Florida, 20 years ago. The company now offers more than 1,000 CUDA-X libraries built on that foundation.

What do science agents actually do?

Nemotron for Science combines the Nemotron 3.5 Super VL open model with a new open chemistry dataset and post-training recipes for life sciences and chemistry. Researchers can run it locally on a DGX Spark workstation or their own supercomputer. "You own your data. You own the model," Buck said.

At Argonne National Laboratory, an agent built with the toolkit took a single prompt, planned and tested hypotheses, chose which simulations to run and identified two new binding sites.

"[Agents] have been just a game changer in how we interface with our supercomputers, with our devices, our instruments and how we can do the research," Buck said.

Deployments he cited:

  • GROMACS: An agent from KTH Royal Institute of Technology repairs broken molecular dynamics setups on its own. In one overnight run on a single DGX Spark, it fixed 81% of test cases.
  • SETI Institute: At the Allen Telescope Array, an agent took a one-line prompt to plot a methanol line, then retuned for Doppler shift on its own.
  • Nsight AI Porting Agent: NVIDIA's new agent ported a routine from the Weather Research and Forecasting (WRF) model to GPUs, speeding it up more than 5x using about 150 million tokens.

Why CPUs matter for agents

Agents lean on CPUs to run tool calls between model steps. "That's exactly what agents need," Buck said. "They need a CPU that can run, evaluate, run the tool, perform the operations, gather the results."

DKRZ, the German Climate Computing Center, has selected Vera CPUs for its HLRE-5 supercomputer, expected in early 2028. It joins Vera systems already announced at Texas Advanced Computing Center.

Inside the new GPU

Rubin HPC is a new GPU in the Vera Rubin platform, built for science, engineering, analytics and AI. It supports precisions from FP64 down to FP4, adds sparse-data acceleration and connects over seventh-generation NVLink. It carries more than 50,000 CUDA cores, roughly three times as many as Hopper. Availability is expected in 2028.

Quantum: error correction and decoding

A simulation on Jülich's JUPITER supercomputer showed that a technique called magic state cultivation can cut the operations needed for one logical step tenfold. The work is being added to CUDA-Q, NVIDIA's quantum software platform.

NVIDIA, IBM and Fermilab plan a quantum-classical co-design platform for error-correction and applications research. NVIDIA and IBM are pairing DGX B200 systems with IBM quantum computers, including those powered by IBM Quantum Nighthawk.

On decoding, Buck set a clear target: cut latency from about 300 microseconds to 10. A new CUDA-Q decoding algorithm, TCBP-JOBS, cuts error-correction research computation from 12 weeks to 45 minutes, he said.

Networking

Buck also showed one of the first deployments of NVIDIA Photonics co-packaged optics switches, at Lambda. The switches are 15x more reliable on bit errors and link flaps, he said.

via nvidianews.nvidia.com (Original)

Filed under

  • nvidia
  • rubin-gpu
  • vera-cpu
  • nemotron-for-science
  • hpc
Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering regulation, policy and their impact on hardware layer of artificial intelligence at Die Signal.

665 articles

Same lot · LOT-362D

« Previous articleNext article »