IBM Quantum Computer Solves in 19 Seconds a Calculation That Would Take a Supercomputer 110 Years

The world's fastest supercomputer would need an estimated 110 years to complete a computation that IBM's quantum processor finished in 19 seconds. The achievement is particularly significant because it represents the first demonstration of quantum advantage using a commercial processor accessible to ordinary users through the cloud, rather than specialized laboratory equipment.

A research team led by Tigran Sedrakyan, a theoretical condensed matter physicist at the U.S. quantum software company BlueQubit, conducted random circuit sampling (RCS) experiments using IBM's 120-qubit superconducting quantum processor, the Nighthawk r2. The results were posted last week on arXiv, the preprint server, and have not yet undergone peer review.

The researchers selected 61 qubits and applied random circuits up to 40 cycles deep. The most meaningful results emerged at 36 cycles, where 918 two-qubit gates were used. The Nighthawk r2 took 19 seconds to generate 1 million samples.

Random circuit sampling works by repeatedly applying randomly selected operations to qubits, then measuring the quantum state to obtain samples composed of 0s and 1s. As the number of qubits and operations increases, the computational effort required to reproduce the same results on a classical computer grows exponentially.

To estimate how long a classical computer would take, the researchers employed tensor-network contraction techniques. They calculated that reproducing the same 1 million samples would require approximately 1.2×10²⁷ operations. Applying this figure to the actual sustained performance of Frontier—the world's first exascale supercomputer and the fastest machine through 2024—yields an estimated runtime of roughly 110 years.

Since the experiment cannot be validated by computing the exact answer directly, a separate verification procedure was necessary. The researchers used two distinct methods. The first involved "patch circuits," where portions of the circuit connections were severed to create 3–4 smaller clusters—small enough for classical computers to simulate exactly, providing reference values. The second used "mirror circuits," where the same operations were executed forward and then immediately reversed; on ideal hardware, qubits should return to their starting state, so any deviation directly measures the error magnitude. The fidelities estimated by both methods agreed at every measured depth.

The fidelity of the surviving quantum signal at 36 cycles was approximately 0.23%. In noisy quantum circuits, information gradually washes out, so this fidelity value represents not an accuracy score but rather how much of the ideal probability distribution's signature remains detectable. What matters is whether that signature persists at measurable levels—and whether a classical computer could reproduce the distribution at the same fidelity. The researchers explain that 36 cycles represents the point where classical simulation costs saturate while the quantum signal remains detectable.

The key lies in the experimental conditions. The experiment was conducted in IBM's standard cloud execution environment, with no custom calibration or special adjustments. The researchers published everything—circuits, samples, and analysis code. While previous quantum advantage experiments were often performed on laboratory-dedicated equipment, this one achieved results on a processor accessible to anyone through a commercial cloud platform.

Japan hosts a system from the same vendor. In June 2025, RIKEN's Center for Computational Science (R-CCS) became the first site outside the United States to install an IBM Quantum System Two. Located in the same building as the Fugaku supercomputer and connected to it at the instruction level, the system features a 156-qubit Heron processor. While the Nighthawk r2 itself is only accessible via the cloud, IBM notes that new users can start with a free account on the same platform.

In their paper, the researchers state: "To the best of our knowledge, this is the first demonstration of quantum advantage for generic random circuit sampling on a commercially accessible quantum processor that can be readily reproduced by most non-expert quantum computer users."

What Made 19 Seconds Possible: The Reset

The speed did not come from adding more qubits. According to IBM, the Nighthawk r2's core innovation lies in reducing the time required to reset qubits to their initial state between circuit executions. The previous Heron generation used conditional resets—measuring qubits and then applying inversion operations based on their states—which required holding circuits idle for hundreds of microseconds to fully reset qubits outside the computational region.

The Nighthawk r2 attaches a dedicated reset element to each qubit, actively extracting energy into the cold environment. When activated, this element reduces the qubit's energy retention time (T1) from a median of approximately 200 microseconds to about 25 nanoseconds, and the inter-circuit wait time drops to as little as 1 microsecond. IBM reports that this architecture increased circuit executions per second from roughly 4,000 on Heron to over 100,000—an approximately 25-fold improvement.

In terms of raw qubit count, the Nighthawk r2 has fewer than Heron. IBM explains that this generation's improvements lie not in scale but in speed and initialization accuracy. With 120 programmable qubits, 218 couplers, and 120 reset elements, the processor contains 458 physical components—the most complex processor IBM has ever manufactured.

Differences from Google's 2019 Experiment

Quantum advantage refers to a situation where a quantum computer performs a computation that would be impractical for a classical computer to complete within a realistic timeframe. In 2019, Google announced it had achieved quantum advantage using its 53-qubit Sycamore processor in a random circuit sampling experiment.

At the time, Google used a dedicated processor at its own research facility. The IBM experiment differs in that it leveraged commercial hardware delivered as a cloud service—quantum advantage confirmed on production infrastructure rather than a purpose-built research environment.

The paper also notes that the 110-year figure is an estimate based on the assumption of no memory constraints.

Limitations and Caveats

The researchers are careful to note that these results do not permanently define the limits of classical computing. The 110-year figure is an estimate based on the computational effort required to reproduce the quantum circuits on a classical computer using a specific approach. If more efficient algorithms are developed in the future, the computational effort and runtime required on classical computers could decrease substantially.

Furthermore, this achievement is limited to a specific benchmark: random circuit sampling. It does not mean quantum computers broadly outperform classical computers in general-purpose computation. Random circuit sampling is a representative test designed specifically to demonstrate quantum advantage, and it is far from a direct indicator of practical problem-solving capability.

Nevertheless, the experiment is being interpreted as a signal of the maturity of commercial quantum computing infrastructure. The fact that researchers completed the experiment in a standard cloud environment without custom calibration—and published their code—lays the groundwork for other research teams to independently reproduce similar experiments.

The preprint status, not yet peer-reviewed, also means further validation is needed. The scientific community's review process will likely scrutinize the experimental methodology and the validity of the estimates in the coming months.

Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.