Q-CTRL and IBM Report a 3,000x Wall-Clock Speedup on a 120-Qubit Fermi-Hubbard Simulation, Claiming Practical Quantum Advantage

Overview

Q-CTRL and IBM announced on May 6, 2026 that a 120-qubit simulation of the Fermi-Hubbard model on IBM superconducting hardware ran roughly 3,000 times faster in wall-clock time than the best classical benchmark, according to The Quantum Insider. The teams frame the result as evidence of practical quantum advantage on a problem of genuine industrial interest rather than a contrived benchmark, although the paper has not yet undergone peer review.

The underlying technical manuscript, titled “Fast, accurate, high-resolution simulation of large-scale Fermi-Hubbard models on a digital quantum processor,” was posted on arXiv on May 5, 2026, with eight Q-CTRL authors including company founder Michael J. Biercuk.

What the Demonstration Did

The experiment simulated the one-dimensional Fermi-Hubbard model, a workhorse description of interacting electrons that is widely used in materials science to study superconductors and energy-storage materials. The main run used 120 qubits arranged as 60 lattice sites with two qubits per site, executed more than 10,000 two-qubit gates, and pushed Trotter step depth up to 90, according to Quantum Computing Report. The same source reports that the quantum results agreed with classical reference calculations to within 1% root-mean-square error.

On the timing claim, the quantum processor produced its answer in about two minutes, while the performance-optimized classical baseline, a Time-Dependent Variational Principle (TDVP) tensor-network solver running on a high-performance compute cluster, needed more than 100 hours to reach comparable accuracy, per Quantum Computing Report. The same outlet identifies ITensor as a comparable leading classical tool used in the comparison.

A separate 62-qubit experiment, reported in the same write-up, observed spin-charge separation, the phenomenon in which an electron’s spin and charge degrees of freedom propagate at different velocities through a one-dimensional system. According to Quantum Computing Report, the experiment extracted precise velocity ratios that match classical predictions, an independent check that the quantum hardware is reproducing known physics rather than producing noise that happens to be fast.

The arXiv preprint frames the broader contribution as demonstrating quantum simulation of one-dimensional Fermi-Hubbard dynamics using up to 120 qubits on superconducting hardware, with outputs that “agree quantitatively with approximate classical simulations” using tensor-network methods, according to arXiv.

The Software Story

The headline numbers depend heavily on Q-CTRL’s commercial error-suppression stack. The runtime improvements over the bare quantum hardware were produced by Fire Opal, the company’s performance-management infrastructure software, which implements what Quantum Computing Report describes as runtime error suppression. The arXiv paper refers to the same approach as “efficient overhead-free error-suppression techniques” and credits these techniques for enabling the extended Trotter sequences up to 90 steps, according to arXiv.

The operational angle for customers is that the workflow is being packaged as a Qiskit Function on the IBM Quantum Platform, meaning third parties will be able to call the simulation pipeline without rebuilding the error-suppression stack themselves, per The Quantum Insider.

How the Principals Framed It

Michael J. Biercuk, CEO and founder of Q-CTRL, told The Quantum Insider: “These results mark the beginning of an era of positive ROI from today’s widely available quantum computers on problems that early adopters truly care about.”

Jay Gambetta, Director of IBM Research and an IBM Fellow, was quoted in the same outlet saying: “We’ve moved past the question of whether quantum computers have utility and onto the question of how to use them well,” according to The Quantum Insider.

Industry analyst Andre Konig, CEO of Global Quantum Intelligence, characterized the work to The Quantum Insider by saying that “Q-CTRL’s demonstrations showcase the crucial role of software in unlocking near-term quantum capabilities.”

Jean-Francois Bobier, Partner and Vice President at Boston Consulting Group, said: “This achievement represents a major signal to industry that quantum simulation is both ready and an essential component of the R&D roadmap,” per The Quantum Insider.

What We Don’t Know

The arXiv submission is a preprint dated May 5, 2026, and has not yet completed peer review, according to arXiv. Independent reproduction by a group without access to Q-CTRL’s Fire Opal stack will be needed before the wall-clock claim can be evaluated against alternative classical solvers, including more aggressive tensor-network variants tuned to this exact regime.

The demonstration is also specific to one-dimensional dynamics. The Fermi-Hubbard model in two or three dimensions is what underlies the open problems in high-temperature superconductivity, and the published results do not extend to that regime. Whether the same hardware-plus-software combination scales to harder lattices is not addressed in the available sources.

Neither outlet reports the specific IBM processor name used in the demonstration. The available coverage refers only to “the IBM Quantum Platform” rather than a named device.

Why It Matters

The Fermi-Hubbard model is one of a small number of physics problems that quantum-computing advocates have long pointed to as a near-term target where digital quantum hardware could plausibly beat the best classical methods. A wall-clock claim of 3,000x on 120 qubits, accompanied by an arXiv paper and an integration into a public cloud platform, moves the conversation about “practical quantum advantage” from the future tense to a concrete result that other researchers can now try to reproduce, attack, or extend.