A New Score to Gauge Quantum Computer Usefulness

When you hear "quantum computer", you probably picture a futuristic beast that can instantly crack encryption or simulate the chemistry of life‑changing molecules. In practice, today’s devices are still stumbling over basic tasks, and figuring out just how far off they are has been something of a guessing game.

Now a group of researchers headed by Timothy Proctor at Sandia National Laboratories says they have built a ruler for that purpose. They call it the QUOPS score – short for Quantum Universal Operation Performance System – and it aims to translate a machine’s raw hardware specs into a single, comparable number.

The idea is surprisingly straightforward. A quantum computer runs a predefined suite of circuits, designed to tease out two things: how many physical qubits it houses, and how many logical operations it can carry out before errors ruin the calculation. Those two ingredients are folded together into the QUOPS score. Then you do the same exercise for the algorithm you actually want to run – be it a simulation of the FeMoco enzyme that fixes nitrogen, or a cryptographic attack on RSA‑2048. If the algorithm’s required QUOPS exceeds the machine’s, the computation is unlikely to succeed. If the machine’s score matches or beats the algorithm’s, the problem becomes, at least in principle, tractable.

Proctor’s team ran the benchmark on a handful of the world’s leading machines – Google’s Sycamore, IBM’s latest superconducting processor, and Quantinuum’s trapped‑ion Helios‑1. The scores were modest: none crossed the 2,000‑operation mark, whereas cracking current cryptography or accurately modelling FeMoco would demand scores more than 100,000 times larger.

They also put Quantinuum’s Helios‑1 into a fault‑tolerant mode, where the system actively detects and corrects its own errors. Even then, the device scraped a QUOPS of roughly 40, underscoring how much extra overhead fault‑tolerance brings.

Why does this matter? As engineers add more qubits, the machines become ever more error‑prone. Fault‑tolerant architectures – which bundle many noisy physical qubits into a smaller set of reliable logical qubits – are widely seen as the only way forward. The QUOPS framework gives a concrete way to track when logical qubits finally start outpacing their noisy cousins in useful performance.

“The inflection point will be when logical‑qubit‑based machines routinely post higher QUOPS scores than the best noisy devices,” Proctor notes. “That’s when quantum computing will start to feel truly practical.”

Industry voices are taking note. Sam Stanwyck of Nvidia, who helped develop the benchmarking software, says the tool lets designers see how tiny tweaks – say, a marginal improvement in gate fidelity – ripple through to the overall score. That feedback loop could accelerate progress toward higher‑score machines.

Not everyone thinks QUOPS will be the final word, however. Wolfgang Mauerer of the Technical University of Applied Sciences Regensburg cautions that as the field matures, new performance dimensions may emerge that a single number can’t capture. “Quantum power is still a fuzzy concept,” he remarks. “Benchmarks help, but they’re never the whole story.”

In any case, the QUOPS score provides a reality check for the hype surrounding quantum computing. It reminds us that, while we’re building ever larger qubit arrays, the road to solving the big, world‑changing problems is still long and winding.