Quantum computing has made undeniable progress over the past year. Logical qubits are being demonstrated, error correction milestones are being reached, and investment continues to flow into hardware start-ups. Yet, for all the momentum, the field remains far from the transformative technology many headlines suggest. If you want to evaluate the latest breakthrough without being swept up in the excitement, you need a clear-headed framework. This guide offers a practical checklist for judging quantum claims and avoiding hype, centered on the most important question: what quantum computing still can’t do today.
Why Quantum Claims Need Extra Scrutiny
Quantum computing is uniquely difficult to assess because the technology is both deeply counterintuitive and moving quickly. A paper that shows a tiny logical qubit array might be described by its institution as “a giant leap toward universal quantum computing.” A company might report “quantum advantage” using a contrived sampling task that has no real-world use. Without a robust set of evaluation criteria, even experienced technology leaders can mistake a clever physics experiment for a practical computing breakthrough. The gap between a physics milestone and a useful machine is still enormous, and that gap is where unrealistic expectations are born.
To navigate this landscape, you don’t need a PhD in quantum mechanics. You need to ask the right questions — questions that reveal whether a claim is about actual capability or just a carefully worded experiment. The checklist below is designed to do exactly that.
The Five-Point Hype Checklist
1. Look for Logical Qubits, Not Just Qubit Counts
The first thing to check in any quantum computing announcement is whether the qubits mentioned are physical or logical. Physical qubits are the raw hardware elements — superconducting circuits, trapped ions, or neutral atoms. They are noisy, fragile, and prone to errors. Logical qubits, on the other hand, are made from groups of physical qubits that work together through quantum error correction to store information more reliably. A machine with 10,000 physical qubits might still have only a handful of logical qubits, or none at all.
When a company boasts about a high qubit count, ask whether that number refers to physical or logical qubits. If the answer is physical, the claim is far less impressive than it appears. Real progress today is measured in logical qubits and error suppression per operation — and even the best demonstrations remain in the single-digit logical qubit range. Until vendors routinely report logical qubit counts and error rates, any claim about “million-qubit machines” or “fault tolerance” should be viewed with skepticism.
2. Demand a Fair Classical Baseline
Another common red flag is an unfair comparison against classical computers. Many quantum advantage demonstrations use problems that are specifically tailored to quantum hardware — for example, random circuit sampling — and then compare their performance against a classical simulation running on a single GPU or a modest CPU cluster. But a truly convincing claim must compare against the best classical algorithms running on the best available supercomputers, ideally with the same problem instance and the same constraints.
In the past few years, classical algorithms have improved dramatically for problems that were once thought to be quantum-hard. A quantum result that edges out a 2019-era simulator is not a compelling result. Always ask: what was the classical baseline? Was it the state of the art at the time of the comparison, or a strawman that makes the quantum machine look better than it really is?
3. Check Whether the Benchmark Is Reproducible
Reproducibility is a cornerstone of scientific credibility, yet many quantum computing announcements come with no public benchmark data. A vendor might claim “X times faster than classical” without releasing the full input parameters, the exact algorithm used, or the hardware configurations on both sides. Without these details, the claim is essentially a press release, not a verifiable result.
Before taking a quantum claim seriously, look for independent replication or at least a detailed preprint that explains the experiment well enough for others to reproduce it. Also check whether the benchmark has a clear, well-defined figure of merit — such as task completion time, energy consumption, or cost per solution — rather than a vague metric like “quantum operations per second.” If the benchmark can only be run by the vendor itself, it should not be treated as solid evidence.
4. Ask Who Actually Wants the Result
A quantum computer that solves a highly abstract mathematical problem faster than a classical machine is not necessarily useful. The real question is whether anyone has been waiting for that computation to be done. In fields like cryptography, chemistry, materials science, and optimization, there are known hard problems that quantum computing might someday accelerate. But many touted use cases are still hypothetical: no company is currently losing money because a quantum computer is not available.
When a vendor presents a use case, ask whether it is based on an actual business requirement or an academic curiosity. A practical quantum application should solve a bottleneck that classical computers have struggled with for years, not a contrived problem that perfectly matches the quantum hardware’s strengths. The best way to avoid hype is to stay focused on problems that are already important, not on problems invented to make quantum computing look good.
5. Scrutinize the Roadmap
Every quantum company has a roadmap, and almost every roadmap shows a curve heading steeply upward toward fault tolerance and commercial usefulness. But roadmaps are promises, not capabilities. They often depend on multiple unproven technologies — from improved error decoding to advanced cryogenic packaging — being developed simultaneously. A single missed milestone can push the entire timeline back by years.
When judging a roadmap, pay attention to concrete engineering milestones rather than broad statements like “we expect to reach 1,000 logical qubits by 2029.” Look for evidence that the company has actually built the smaller components it plans to scale up. Also note that many roadmaps do not include the massive classical compute overhead required to run quantum error correction in real time. A realistic roadmap should show not just the quantum chip, but the entire system: control electronics, wiring, cooling, and classical co-processors.
What Quantum Computing Still Can’t Do Today
Even if a claim passes the checklist, the technology remains limited in fundamental ways. Here are a few things quantum computing still can’t do, and likely won’t be able to do for several years:
- Run large-scale, error-corrected algorithms: The systems that can run meaningful applications without errors do not yet exist. Every current quantum computer is still in the “noisy intermediate-scale quantum” (NISQ) era, meaning its results are probabilistic and require extensive error mitigation.
- Outperform classical computers at most real-world tasks: For the vast majority of workloads — from database search to machine learning inference — classical computers remain faster, cheaper, and more reliable. Quantum speedups have been shown only for a handful of carefully selected problems.
- Scale to practical size without major engineering breakthroughs: A fault-tolerant quantum computer that can break RSA encryption or simulate a complex drug molecule will likely need millions of physical qubits. Today we have machines with thousands, and the path to millions is not yet clear.
- Operate independently of classical systems: Quantum processors are not standalone computers. They require classical supercomputers to calibrate them, control them, and decode their error-correction syndromes. The quantum part is just a peripheral, not yet a standalone platform.
- Guarantee speedups for your specific problem: A theoretical speedup does not always translate into an actual end-to-end advantage. Input/output, data loading, and communication overhead can wipe out any quantum gain, especially on problems with large amounts of classical data.
Conclusion
Quantum computing is progressing fast, but it is still a physics-driven research effort with occasional flashes of utility, not a general-purpose computing revolution. The most reliable way to evaluate quantum claims is to apply a simple set of checks: look for logical qubits, demand fair classical baselines, require reproducibility, focus on real problems, and treat roadmaps with skepticism. The next time you read that quantum computers are about to change the world, remember how much they still can’t do — and use that knowledge to separate meaningful progress from marketing excitement.
