IBM quantum task takes 19 seconds, with a 110-year supercomputer estimate

The benchmark behind the headline

The headline number is deliberately dramatic: an IBM quantum processor completed a narrowly defined sampling task in 19 seconds, while the researchers behind the experiment estimated that a comparable classical reproduction could require roughly 110 years on a leading supercomputer [1]. The processor was IBM’s Nighthawk r2, a 120-qubit superconducting system available through IBM’s cloud platform, and the work was led by Tigran Sedrakyan of U.S. quantum software company BlueQubit [1].

What actually ran was not a business application, a drug-discovery workflow or a code-breaking calculation. It was random-circuit sampling, a benchmark in which a quantum computer applies complex, mostly random operations to qubits and then measures the resulting quantum state many times [2]. In this case, the reported central run used 61 qubits on Nighthawk r2 and produced one million measurement results in 19 seconds under a 36-cycle condition [2].

That distinction matters. Random-circuit sampling is designed to be hard for classical machines to imitate as circuits grow wider and deeper; it is not meant to be a direct prototype of a conventional enterprise workload. Its value lies in stress-testing whether a quantum device can produce outputs that would be prohibitively expensive for classical simulation, not in immediately solving a practical industry problem.

What “19 seconds” means

The reported 19 seconds refers to the time spent generating one million samples on the quantum processor, not to a full end-to-end project timeline including research design, cloud queueing, verification and analysis. The test selected 61 qubits from the 120-qubit Nighthawk r2 processor and ran increasingly complex random circuits, with the key comparison based on a 36-cycle circuit [1].

Several recent reports identify the same technical center of gravity: the run used random-circuit sampling, selected a subset of Nighthawk r2’s qubits and compared the quantum sampling time with an estimated classical simulation burden [1][2][3]. Gigazine’s coverage notes that the researchers also considered circuits up to 40 cycles and estimated that calculating a single output amplitude could require about 10^22 complex operations under the model used [3].

The benchmark therefore has two components. The first is the measured quantum side: the processor produced one million samples in 19 seconds. The second is the estimated classical side: researchers used simulation-cost methods to infer how much work a state-of-the-art classical system would need to reproduce comparable samples [2].

The 110-year comparison is an estimate, not a stopwatch

The 110-year figure is the most memorable part of the story, but also the easiest to overread. VnExpress reports that the comparison was based on Frontier, a leading U.S. supercomputer, and that the estimate did not come from actually running the full job on Frontier for 110 years [1]. Instead, the researchers used a particular approach to estimate how difficult it would be for conventional computing methods to reproduce the quantum output [1].

Digital Today similarly emphasizes that the same performance gap should not be generalized to all calculations, data-analysis tasks, optimization problems or drug-discovery workloads [2]. The comparison is limited to the specific benchmark of random-circuit sampling, and the result remains a preprint that has not yet been peer-reviewed [2].

This is why the most careful reading is not “quantum computers are now 180 million times faster than supercomputers.” It is closer to this: on one carefully chosen random-circuit sampling benchmark, a commercially accessible IBM quantum processor produced samples in seconds, while one modeled classical strategy was estimated to need an impractical amount of compute time.

Why the result still matters

Even with those caveats, the claim is significant because of accessibility. The experiment was reported to use IBM’s standard cloud service rather than a one-off laboratory machine specially built only for the benchmark [1]. Hit and Hot News also reports that the work used IBM’s commercially accessible Nighthawk r2 processor, with 61 qubits from the 120-qubit device and a 36-cycle circuit involving 918 two-qubit gates [4].

That accessibility changes the benchmark’s practical meaning. Earlier quantum-advantage demonstrations often relied on specialized hardware, narrow experimental settings or systems unavailable to ordinary users. Here, the reported point is that the task ran on a commercial cloud-accessible processor, which makes the claim easier for outside researchers to examine and potentially reproduce [1][4].

The researchers also reportedly made circuits, measurement results and analysis resources available for examination, a key step because random-circuit sampling claims are only as strong as their verification and classical comparison [4]. If outside groups can rerun the circuit, test the fidelity estimates and improve the classical simulation baselines, the field gets a more concrete measure of progress than qubit count alone.

What random-circuit sampling does and does not prove

Random-circuit sampling creates complicated quantum states, measures them and asks whether a classical machine can generate samples from the same probability distribution. As qubit numbers and gate depth increase, directly calculating that distribution becomes extremely expensive [2]. This is why the benchmark has become a standard battleground for quantum advantage claims.

But a sampling win is not the same as general-purpose quantum advantage. It does not mean that today’s quantum computers can replace supercomputers in climate modeling, logistics, genomics or financial risk analysis. It also does not mean that a classical algorithm will never catch up. VnExpress notes that earlier quantum-advantage estimates, including after Google’s 2019 Sycamore demonstration, were later narrowed as researchers developed more efficient classical simulation techniques [1].

That historical pattern is important. Quantum hardware improves, but so do classical algorithms. Every big sampling claim gives classical-computing researchers a new target, and many of the most useful debates begin after the headline result appears.

The enterprise angle: hybrid, not replacement

For companies watching the quantum sector, the lesson is not to buy into a sudden “supercomputer killer” narrative. The better takeaway is that quantum hardware is becoming capable enough to justify serious hybrid-workflow planning. In that model, classical systems continue to handle most computing tasks, while quantum processors are reserved for narrow workloads where quantum structure may offer an advantage.

The Nighthawk r2 result strengthens that argument because it gives buyers, cloud users and researchers a concrete benchmark beyond raw qubit counts. A processor that can be accessed commercially and run a difficult sampling circuit at measurable scale is more relevant to enterprise planning than a laboratory milestone that cannot be repeated outside one institution [1][4].

Still, the procurement question remains conservative: What problem does the quantum device solve, how is output quality verified, what is the classical baseline, and can the workflow tolerate probabilistic results? Those questions matter more than the 19-second headline.

What to watch next

The next phase is scrutiny. Peer review must test the assumptions behind the fidelity estimates, the classical-cost model and the comparison with Frontier-like performance [2]. Independent researchers will also look for faster classical simulation methods, because the “110 years” number could shrink if a better algorithm or decomposition strategy is found [1].

If the claim holds up, the result becomes a strong marker that commercially accessible quantum processors are entering a more serious benchmarking phase. If the gap narrows, the experiment will still be useful: it will have provided a public target, a test circuit and a sharper understanding of where quantum advantage is real, fragile or overstated.

For now, the safest conclusion is also the most interesting one. IBM’s Nighthawk r2 appears to have delivered a striking 19-second result on a narrow but important quantum benchmark, and the reported 110-year classical comparison is a challenge to be tested, not a slogan to be accepted uncritically. Somewhere, a classical cluster may indeed be filing for 110 years of overtime, but the auditors have only just opened the spreadsheet.

Sources from the last 72 hours

1. [1]IBM quantum computer completes 110-year supercomputer task in 19 secondsOct 2, 2026, 9:36 AM
2. [2]IBM quantum processor generates 1 million samples in 19 seconds in random circuit sampling testSep 30, 2026, 8:31 AM
3. [3]IBM's quantum computer completes a calculation in 19 seconds that would take a supercomputer 100 years.Sep 30, 2026, 6:35 AM
4. [4]IBM Quantum Processor Completes 19-Second Calculation Estimated to Take a Supercomputer 110 YearsSep 30, 2026, 2:00 AM

AI-generated article based on recent web research, then preserved as a dated editorial snapshot.