A Rare Look at What People Actually Run on Quantum Hardware: 6-Qubit Circuits, Two-Minute Queues and Price Doing the Steering
The quantum computing sector produces an enormous amount of forecasting and very little measurement. Quantum Rings, which sells circuit-simulation software and resells cloud QPU access through its Open Quantum product, has published something closer to the latter: a summer 2026 report drawn from the circuits its public-plan users actually submitted.
Chief executive Bob Wold framed it bluntly in remarks reported by The Quantum Insider: "Quantum computing has no shortage of forecasts and no surplus of measurements." The company's own announcement page, dated 1 September, describes the study as measuring real quantum computing usage from the submitted circuits themselves, across six QPUs from four hardware vendors.
The methodology, and its limits
The window runs 1 June to 31 August 2026 and covers hundreds of active users across 47 countries. The important qualification is one the numbers cannot escape: this is a vendor's measurement of traffic on its own platform, published by that vendor, and it has not been independently audited. It is a census of the users Quantum Rings has - a population weighted heavily toward students, researchers and others working on low-cost access - rather than of the quantum market, and anyone reading it as a proxy for enterprise demand is reading it wrong. What the dataset does capture well is how price and queue behaviour steer a cost-sensitive user base, which is genuinely under-reported.
Circuits are small, and growing fast
Per The Quantum Insider's summary, the median circuit submitted was six qubits. The movement is in the tail: the 95th-percentile circuit grew from 20 qubits in the spring to 96 qubits over the summer window - the nearly five-fold jump in the frontier of real usage that Quantum Rings highlights on its own news page. Jobs of 50 qubits or more rose to 11% of the total from 2% in the previous period, and jobs above eight qubits rose to 39% from 14%. Completed jobs were up 115% and active users up 206%.
Mostly benchmarks, not applications
The composition of the workload is the finding most likely to be uncomfortable for the sector's marketing. Among the traffic the study could classify, 66% of circuits were designed mainly to prepare quantum states or benchmark hardware, and applications accounted for 20%. That qualifier carries weight: 37% of jobs contained circuits the study could not classify at all, so the two-thirds figure describes the recognised share of the workload rather than the whole of it.
Of the work users did declare a category for, 77% was physics simulation, with machine learning, optimisation and finance accounting for much smaller shares. That distribution sits awkwardly against the finance-and-logistics framing that dominates quantum sales decks.
Price is the routing algorithm
The pricing spread across the network is extraordinary. Rigetti Computing's Cepheus-1 was the cheapest machine on Open Quantum at $0.000425 per shot and took 57% of all completed jobs. Retail pricing across the network ran from that floor up to eight cents per shot, a 188-fold range. The behavioural consequence shows up in shot counts: median shots per job were 2,000 on Cepheus-1 against 100 on IonQ's more expensive Forte-1. Users on subsidised plans do not simply run fewer jobs on costly hardware, they run thinner ones, which degrades statistical quality on exactly the sampling-heavy algorithms the field cares about.
One durable piece of folklore does not survive. Quantum Rings summarises its own finding as evidence that "the two-minute quantum queue is real," and the per-system figures The Quantum Insider reports are shorter still: a median queue wait of 60 seconds on IQM's Garnet and 38 seconds on IonQ's Forte Enterprise. Median execution times ranged from three seconds on Garnet to ten minutes on Forte Enterprise. For a field that has spent years explaining that access is the bottleneck, waits measured in seconds on this tier of traffic suggest supply is not the constraint at this end of the market - demand and budget are. That is useful to know, and only usage data can tell you. Frontier Tech Wire has relied on The Quantum Insider's account of the report's figures alongside Quantum Rings' own announcement page; the underlying dataset is the vendor's, describes the vendor's own platform, and has not been independently audited. Nothing in this article is investment advice.
The queue myth takes a hit
One durable piece of folklore does not survive. Median queue wait times across all six systems in the study were no more than two minutes, and median execution times ranged from three seconds on IQM's Garnet to ten minutes on IonQ's Forte Enterprise. For a field that has spent years explaining that access is the bottleneck, a two-minute median wait on public-tier traffic suggests supply is not the constraint at this end of the market - demand and budget are. That is useful to know, and only usage data can tell you. Frontier Tech Wire has relied on The Quantum Insider's account of the report's figures alongside Quantum Rings' own announcement page; the underlying dataset is the vendor's and has not been independently audited. Nothing in this article is investment advice.
