NVIDIA Introduces Science Agents, Rubin HPC GPU and Quantum Tools at SC26

NVIDIA has unveiled new tools for AI agents in scientific research, including Nemotron for Science, the Rubin HPC GPU and an expanded Vera CPU platform. The announcements came as Ian Buck opened NVIDIA’s week at SC26 in Chicago.
- NVIDIA introduced Nemotron for Science for local scientific research.
- Rubin HPC supports FP64 to FP4 and is expected in 2028.
- DKRZ selected Vera CPUs for its HLRE-5 supercomputer.
- NVIDIA, IBM and Fermilab announced a quantum computing co-design platform.
- CUDA-Q TCBP-JOBS cuts stated research computation from 12 weeks to 45 minutes.
Open models for scientific research
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. According to NVIDIA, researchers can run the system locally on a DGX Spark, a workstation or their own supercomputer.
The toolkit is designed for agents that can plan experiments, run code and check results on researchers’ systems. At Argonne National Laboratory, Buck said, an agent took a single prompt, planned and tested hypotheses, selected simulations and identified two new binding sites. “You own your data. You own the model,” Buck said.
NVIDIA also highlighted agents used in other research settings. An agent from KTH Royal Institute of Technology repaired broken GROMACS setups independently, fixing 81% of test cases in one overnight run on a single DGX Spark. At the Allen Telescope Array, an agent from the SETI Institute responded to a one-line prompt to plot a methanol line and then retuned for Doppler shift on its own.
CPUs and GPUs for the agent loop
Buck said agents rely on CPUs to execute tool calls between model steps. “They need a CPU that can run, evaluate, run the tool, perform the operations, gather the results,” he said.
DKRZ, the German Climate Computing Center, has selected Vera CPUs for its HLRE-5 supercomputer, which is expected in early 2028. NVIDIA said the selection joins Vera systems already announced at Texas Advanced Computing Center.
Rubin HPC is a new GPU in the Vera Rubin platform for science, engineering, analytics and AI. It supports precisions from FP64 to FP4, includes sparse-data acceleration and connects over seventh-generation NVIDIA NVLink. Buck said it has more than 50,000 CUDA cores, about three times as many as Hopper, and is expected in 2028.
NVIDIA also introduced the Nsight AI Porting Agent. Buck said it ported a routine from the Weather Research and Forecasting model to GPUs, speeding it up more than 5x using about 150 million tokens.
Quantum computing collaboration
NVIDIA, IBM and Fermi National Accelerator Laboratory announced plans for a quantum computing co-design platform focused on error-correction and applications research. NVIDIA and IBM are pairing NVIDIA DGX B200 systems with IBM quantum computers, including systems powered by IBM Quantum Nighthawk.
Buck said a simulation on Jülich’s JUPITER supercomputer showed that magic state cultivation can reduce the operations needed for one logical step tenfold. The work is being added to CUDA-Q.
NVIDIA also introduced the CUDA-Q decoding algorithm TCBP-JOBS. According to Buck, it reduces quantum error-correction research computation from 12 weeks to 45 minutes. The company’s stated goal is to cut decoding latency from about 300 microseconds to 10.
The announcements position simulation, AI, quantum computing and agents as complementary parts of NVIDIA’s scientific computing stack. “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.


