Quantum Advantage and Quantum Supremacy
Quantum supremacy is the milestone of a quantum computer performing some computation faster than any classical computer could, even if that computation is contrived and useless, and quantum advantage is the higher bar of doing so on a problem people actually care about. The terms matter because they are constantly misread as evidence that cryptography is about to fall, and they are not. Google’s 2019 quantum supremacy demonstration on its Sycamore processor solved a problem designed specifically to be hard for classical machines and to have no practical application, and even that narrow claim was partly walked back as classical algorithms and hardware improved. A supremacy or advantage result advances the science of quantum computing without moving the machine any closer to running Shor’s algorithm against RSA, because breaking cryptography is a completely different and far harder task. Keeping these terms straight is part of reading quantum news honestly, the same discipline as How to Tell Real Quantum Progress From Hype.
Source: John Preskill, “Quantum computing and the entanglement frontier,” 2012, arXiv:1203.5813.
The short version:
- Quantum supremacy means a quantum computer beats every classical computer on some task, even a contrived one with no use. The term was coined by John Preskill in 2012.
- Quantum advantage is the stronger claim of a quantum computer beating classical machines on a genuinely useful problem, which no one has convincingly demonstrated for a valuable application.
- Google claimed quantum supremacy in 2019 on its 53-qubit Sycamore processor, running a sampling task in about 200 seconds that it estimated would take a classical supercomputer 10,000 years.
- That estimate was contested. IBM argued a classical supercomputer could do the same task in about 2.5 days, and later classical work cut the gap further, so the “10,000 years” figure did not hold.
- Neither supremacy nor advantage means cryptography is broken, because the demonstrated tasks are sampling problems with no cryptographic relevance, and a CRQC is a different and far larger machine.
Picture a car that sets a land-speed record on a perfectly flat, closed salt flat, doing something no ordinary car could match on that specific surface. It is a genuine engineering achievement, and it proves the car can do something special under ideal conditions. It also tells you almost nothing about whether the car can drive across a mountain range hauling a trailer, because that is a different task with different demands. Quantum supremacy is the salt-flat record. The task is chosen to favor the quantum machine and to have no practical purpose, so the result is real and impressive and still does not mean the machine can do the useful, punishing work of breaking cryptography.
What is quantum supremacy?
Quantum supremacy is the demonstration that a quantum computer can perform a specific computation that no classical computer can perform in any reasonable time, regardless of whether that computation is useful. The term was coined by the physicist John Preskill in 2012, and his paper frames it as a threshold worth crossing to prove the quantum world can outrun the classical one. In his words, “one way to achieve such ‘quantum supremacy’ would be to run an algorithm on a quantum computer which solves a problem with a super-polynomial speedup” over the best classical approach. The key feature of the definition is that usefulness is not required, so a supremacy task can be, and in practice is, a problem constructed specifically to be hard for classical machines.
Source: John Preskill, “Quantum computing and the entanglement frontier,” 2012, arXiv:1203.5813.
The purpose of the milestone is scientific rather than practical. Demonstrating supremacy shows that a quantum computer has crossed into a regime where it does something a classical computer genuinely cannot follow, which validates that the machine is behaving as quantum theory predicts at a scale that matters. That is a real and hard-won achievement. It is also a narrow one, because the task is selected to make the quantum machine look good, so crossing the supremacy line proves the physics works without proving the machine can do anything valuable, let alone break cryptography.
How is quantum advantage different?
Quantum advantage is the stronger and more meaningful claim that a quantum computer outperforms classical computers on a problem that people actually want solved, so it adds usefulness to the raw speed that supremacy demonstrates. Where a supremacy task can be a contrived sampling problem with no application, an advantage result would be a quantum computer beating the best classical methods at something with real value, such as simulating a molecule for drug discovery, optimizing a logistics network, or, in the security context, breaking a cryptographic key. The bar is much higher precisely because the problem has to matter, and the classical competition on useful problems is fierce and well-funded.
The distinction is the one that most affects how a security professional should read the news. As of 2026 there is no convincing demonstration of quantum advantage on a valuable problem, because the useful problems are exactly the ones where classical algorithms remain strong or where the quantum machine is not yet large and clean enough to win. Breaking cryptography would itself be a form of quantum advantage, an enormously consequential one, and it sits far beyond current hardware. So when a headline announces quantum “supremacy,” it is describing the narrower, usefulness-free milestone, and reading it as the useful kind of advantage is the mistake that turns a science result into a false alarm.
What did Google’s 2019 Sycamore demonstration actually do?
Google’s 2019 demonstration ran a specially designed sampling task on its 53-qubit Sycamore processor and claimed it as the first experimental proof of quantum supremacy. The task, called random circuit sampling, asks the machine to run a random quantum circuit many times and produce samples from the resulting output distribution, which is a problem believed to be very hard for classical computers and, crucially, one with no practical application beyond serving as a benchmark for supremacy. Google reported that Sycamore completed the task in about 200 seconds and estimated that the leading classical supercomputer of the day would need roughly 10,000 years to reproduce the same result.
Source: Frank Arute et al. (Google Quantum AI), “Quantum supremacy using a programmable superconducting processor,” Nature 574, 505-510, 2019, doi.org/10.1038/s41586-019-1666-5.
Two features of the demonstration are essential to reading it correctly. The task was chosen to be maximally favorable to the quantum machine and useless in itself, so the result proves the physics at scale rather than any capability anyone would pay for. And the 53 noisy physical qubits involved carry no error correction, so they are the physical qubits that a cryptographic attack cannot use directly, not the error-corrected logical qubits a CRQC would need. The demonstration is a landmark in quantum computing and sits nowhere on the path to breaking RSA.
Why was the supremacy claim walked back?
The claim was walked back because its central figure, the 10,000-year classical estimate, depended on the best classical algorithm known at the time, and classical methods improved quickly once researchers focused on the problem. IBM responded almost immediately, arguing that a classical supercomputer using its full storage cleverly could reproduce Sycamore’s task in about 2.5 days rather than 10,000 years, which shrank the claimed gap from unimaginable to merely large. Later classical work, including improved simulation algorithms, cut the gap further still, so the specific “10,000 years” headline did not survive contact with a motivated classical community.
Source: IBM Research, “On ‘Quantum Supremacy’,” October 22 2019, ibm.com.
The episode is a useful lesson in how to read a supremacy claim, and it does not erase the achievement. A supremacy result is a moving comparison, because it is defined against the best classical method, and that method keeps improving, so the apparent margin can and often does shrink after the announcement. The honest reading is that Sycamore demonstrated a real, hard quantum computation while the precise size of its classical lead was overstated and then corrected. That is a normal part of how the science advances, and it is a reminder to treat the dramatic multiplier in any supremacy headline as provisional rather than settled.
Does quantum supremacy mean cryptography is broken?
No. Quantum supremacy does not mean cryptography is broken, and the reason is that the demonstrated task and a cryptographic attack are entirely different computations with entirely different requirements. Random circuit sampling is a shallow, contrived problem that noisy physical qubits can handle for a benchmark, while breaking RSA with Shor’s algorithm is a deep computation billions of operations long that demands thousands of error-corrected logical qubits, which is millions of clean physical ones. A machine that achieves supremacy on sampling is not one step closer to running Shor’s algorithm at cryptographic scale, because the two tasks stress completely different capabilities.
The clean way to hold this is that supremacy measures progress on one axis, whether a quantum machine can outrun classical ones on a favorable task, and cryptographic risk lives on a different axis, whether a machine has the error-corrected depth to complete a real attack. The resource estimates make the gap concrete: RSA-2048 needs roughly 6,100 logical qubits and a circuit far deeper than any sampling benchmark, which is why a CRQC does not exist in 2026 despite the supremacy milestone being years old. A supremacy or advantage headline is real science, and reading it as a cryptographic threat is the confusion this whole field of terminology exists to prevent.
Common misconceptions
- “Quantum supremacy means a quantum computer beat classical machines at something useful.” Supremacy only requires beating them at some task, and the demonstrated tasks are contrived sampling problems with no application. The useful version is quantum advantage, which no one has convincingly shown on a valuable problem.
- “Google’s 2019 result means cryptography is nearly broken.” Sycamore ran a sampling benchmark on 53 noisy physical qubits with no error correction. Breaking RSA needs thousands of error-corrected logical qubits running a vastly deeper circuit, which is a completely different machine.
- “Sycamore proved a classical computer needs 10,000 years.” That estimate was contested within days. IBM argued about 2.5 days on a classical supercomputer, and later classical work narrowed the gap further, so the headline figure did not hold.
- “Quantum supremacy is a fixed, permanent achievement.” It is a moving comparison against the best classical method, which keeps improving. A supremacy margin can shrink after the announcement, which is exactly what happened to the 2019 claim.
- “Quantum advantage and quantum supremacy are the same thing.” Supremacy allows a useless task; advantage requires a useful one. The difference is precisely usefulness, and it is the distinction that separates a science milestone from a claim that would actually change security.
Questions people ask
What is quantum supremacy? It is a quantum computer performing a computation no classical computer can match in reasonable time, even if the task is contrived and useless. The term was coined by John Preskill in 2012 and requires only speed, not usefulness (arXiv:1203.5813).
How is quantum advantage different from supremacy? Quantum advantage adds usefulness. It means a quantum computer beating classical machines on a problem people actually want solved, which is a much higher bar, and as of 2026 no one has convincingly demonstrated it on a valuable application.
What did Google’s Sycamore actually do in 2019? It ran a random circuit sampling task, a contrived benchmark with no practical use, on 53 noisy physical qubits in about 200 seconds, and Google estimated a classical supercomputer would need 10,000 years (doi.org/10.1038/s41586-019-1666-5). The qubits carried no error correction.
Why did people say the supremacy claim was overstated? Because the 10,000-year figure depended on the best classical algorithm known at the time. IBM argued a classical supercomputer could do it in about 2.5 days (ibm.com), and later classical improvements shrank the gap further, so the dramatic multiplier did not survive.
Does quantum supremacy mean encryption is broken? No. The demonstrated task is a shallow sampling problem with no cryptographic relevance, while breaking RSA needs thousands of error-corrected logical qubits running a circuit billions of operations deep. Supremacy and cryptographic risk are different axes, and a CRQC does not exist in 2026.
Should a supremacy or advantage headline change my migration plan? Not directly. Treat it as progress in quantum computing generally, not as movement toward a code-breaking machine. The migration is driven by lead time and Mosca’s theorem, not by sampling benchmarks, so read these headlines with the discipline in How to Tell Real Quantum Progress From Hype.
Last verified 2026-07-12 · Maintained by Addie LaMarr, LaMarr Labs.