Catenaa, Tuesday, March 03, 2026-Scientists at Lawrence Berkeley National Laboratory and the University of California, Berkeley, have completed a record-scale simulation of a quantum microchip, capturing unprecedented physical detail to refine next-generation quantum hardware designs.
The project used over 7,000 NVIDIA GPUs on the DOE’s Perlmutter supercomputer for 24 hours, modeling a multi-layered chip measuring just 10 millimeters square with etchings one micron wide.
The team employed ARTEMIS, an exascale modeling platform, to simulate electromagnetic behavior, signal coupling, and qubit interactions at the physical level.
Unlike conventional “black box” simulations, the researchers accounted for chip material, wiring layout, resonator shape, and nonlinear effects over more than 11 billion grid cells, enabling evaluation of three circuit configurations in a single day.
Simulating chips before fabrication allows early identification of design flaws, improving reliability and reducing costly iterations.
The model also mimicked lab conditions, showing how qubits communicate with each other and the broader quantum circuit. Researchers plan additional simulations to quantify spectral behavior and benchmark frequency-domain responses against future physical chips.
The collaboration leveraged expertise across Berkeley Lab’s Applied Mathematics and Computational Research Division, the Quantum Systems Accelerator, the Advanced Quantum Testbed, and NERSC staff, combining massive computational resources with specialized knowledge.
Scientists say this approach provides one-of-a-kind insights into quantum hardware performance, accelerating the development of more powerful and precise quantum chips.
This breakthrough establishes a foundation for simulating increasingly complex quantum devices, potentially transforming quantum research and hardware design workflows.
