IBM and University of Chicago Prove Verified Quantum Advantage: 70 Logical Qubits Finish a Task Beyond Classical Reach

Posted on 23rd Aug 2026 06:05:46 in Artificial Intelligence, Machine Learning

Tagged as: quantum computing, quantum advantage, IBM, University of Chicago, error correction, logical qubits

On July 30, 2026, IBM and researchers from the University of Chicago announced a quantum computing demonstration that meets the fundamental criteria for quantum advantage. Their system performed a computation beyond the reach of leading classical simulation methods and, just as importantly, produced statistical evidence that the answer it returned was correct. The experiment encoded 70 logical qubits, ran 2,415 logical two-qubit operations and 468 logical T gates, and completed in roughly 15 minutes on IBM quantum hardware. The team showed that many leading classical simulation approaches would face effectively prohibitive runtimes for the same task.

This matters because quantum advantage claims are not new. Google's 2019 random circuit sampling experiment made headlines worldwide, and several follow-up demonstrations have pushed the frontier since. What has remained stubbornly difficult is verification: proving that a quantum computer's output is trustworthy precisely when the problem has grown too hard for any classical machine to double-check. The IBM-UChicago result attacks that gap directly, and the researchers say it establishes a lower bound on how faithfully the computation was executed. "We are now firmly in the quantum advantage era," said Jay Gambetta, Director of IBM Research and IBM Fellow.

The Verification Problem That Held Quantum Back

For years, the benchmark of choice for quantum-versus-classical comparisons has been random circuit sampling, or RCS. In simple terms, RCS asks a quantum computer to generate patterns so complex that a classical computer cannot efficiently reproduce them. The benchmark made quantum computers famous, but it also created a paradox: the harder the problem, the harder it becomes to confirm the quantum machine got the right answer. Beyond a certain scale, verification becomes infeasible without making strong assumptions about the inner workings of the quantum computer itself.

Earlier experiments often got around this by running smaller or simplified circuits and extrapolating the results to larger, harder regimes. The IBM Quantum blog describes this as a "proxy of a proxy" rather than a direct certification of the classically hard computation itself. Noise and error propagation behave very differently at full scale, so extrapolation does not fully validate a computation in the advantage regime.

"Verification remains one of the biggest challenges in firmly establishing experimental quantum advantage," said Bill Fefferman, associate professor of computer science at the University of Chicago and co-author of the paper. "This experiment develops techniques to better characterize the fidelity of hard quantum states under noise, increasing confidence that the quantum computer is solving a computationally hard problem." Soumik Ghosh, a graduate student in Fefferman's group and co-author, added that stronger verification techniques "have the potential to unlock practical applications for the next generation of quantum computers."

Doped Clifford Sampling: A Structured Answer to RCS

The researchers addressed the verification obstacle with a structured alternative called doped Clifford sampling. The team proved that this alternative retains the same computational hardness criteria as RCS, but with a crucial advantage: the new structure can be used to detect errors during the computation itself.

Here is how it works at a high level. The circuit starts from a Clifford circuit, a class of quantum circuits that can be simulated efficiently on classical computers and whose states can be measured directly. The team then injects, or "dopes," the circuit with a controlled number of T gates, the operations that push the computation into the classically hard regime. The Clifford circuit is embedded in a spacetime code, a construction whose error-detecting regions extend across both the qubits and the circuit's evolution through time. When an error occurs inside a detecting region, it flips an auxiliary qubit, a syndrome, and that run is discarded. Runs that pass are retained and counted.

Because a pure Clifford state's fidelity can be measured directly through stabilizer sampling, the team could compute a guaranteed lower bound on the fidelity of the doped circuit. The encoded computation achieved effective logical error rates ten times lower than the underlying physical error rates, with measured fidelity above 28 percent (an average of roughly 0.32) compared with around 1 percent for an unencoded circuit of the same size. That is the heart of the milestone: a classically hard computation whose correctness is certified, not assumed.

The Scale of the Demonstration

The experiment ranks among the largest error-correction demonstrations ever reported. The team executed 70 logical qubits while shielding them from errors, a scale that required 2,415 logical two-qubit operations and 468 logical T gates. These metrics quantify the complexity of a quantum circuit, and the fact that they were sustained while maintaining measurable fidelity is what separates this result from earlier benchmarks.

The runtime contrast tells the story in practical terms. The IBM quantum computer took approximately 15 minutes to complete the task. The researchers showed that leading classical simulation methods faced prohibitive runtimes, with matrix product state simulations and stabilizer-based simulations estimated to require on the order of centuries even at optimistic speeds. The details appear in the paper "Sampling hard circuits with verifiably high fidelity," authored by Simon Martiel and collaborators and published on the arXiv preprint server. The circuits and results have also been released openly on the Quantum Advantage Tracker, a community resource that pits quantum advantage claims against the best available classical methods.

"We have demonstrated a quantum computation beyond the practical reach of classical computers that establishes, with statistical confidence, a lower bound on how faithfully it was executed," Gambetta said. "This milestone gives scientists, developers and businesses a new foundation for trusting quantum computers as they scale to problems far beyond what we can achieve classically."

An Ecosystem-Wide Shift Toward Trusted Computation

The IBM-UChicago paper did not arrive alone. The same day, IBM ecosystem partners published companion demonstrations built on the same philosophy of validated quantum computation. Qedma, RIKEN and BlueQubit reported observations of quantum phenomena beyond the reach of leading classical simulations using validated error-mitigation techniques, while Algorithmiq demonstrated an approach that validates the computational process itself rather than comparing against a classical answer. The Quantum Advantage Tracker currently lists submissions from Q-CTRL, BlueQubit and the Birla Institute of Technology and Science, Pilani, among others.

The collective message is that quantum computing is entering an era in which beyond-classical results come with rigorous evidence of reliability. For businesses and developers, that shifts the conversation from spectacle to engineering: if a quantum computer's output can be trusted without classical verification, it becomes plausible to use quantum hardware for discovery problems in chemistry, materials science and optimization where no classical ground truth exists.

What Still Stands in the Way

The researchers are careful about the limits of their own result. One honest caveat, noted explicitly in IBM's technical write-up, is that the verification still trusts the hardware to play fair. An assumption-free demonstration, one that does not rely on trusting the quantum device, remains the next open problem in the field. Error correction also remains enormously expensive: 70 logical qubits are a milestone, but fault-tolerant machines capable of running useful algorithms at scale are still years away, and the overhead of encoding logical qubits from physical qubits means practical applications will require much larger systems.

Still, the direction is unmistakable. Quantum computing has spent years producing dramatic claims that outsiders had to take on faith. The IBM and University of Chicago demonstration is a step toward results that can stand on evidence, and that is precisely the kind of progress that turns laboratory milestones into tools that businesses and researchers can eventually rely on.

Sources

• IBM Newsroom — IBM and The University of Chicago Demonstrate Quantum Advantage, Establishing Trusted Quantum Computation on Logical Circuits
• University of Chicago News — IBM, UChicago demonstrate 'quantum advantage,' outperforming traditional computers with a quantum computer
• IBM Quantum Blog — Quantum advantage through trusted quantum computation
• Phys.org — Quantum computer completes verified task beyond practical reach of classical simulations
• arXiv — Sampling hard circuits with verifiably high fidelity (Martiel et al., 2026)