Most coverage of quantum computing, including a lot of what runs on this site, tracks the companies building the hardware: IBM, IonQ, Quantinuum, D-Wave, and the rest of the field our top companies ranking covers. That is one half of the industry. The other half is the customers, banks, automakers, aerospace manufacturers, drug companies, and utilities paying for hardware access or running joint research, and that side gets covered less carefully, usually as a single line in a vendor press release. This piece flips the lens: which companies outside the quantum industry itself are running workloads on quantum hardware, in which verticals, and how far each effort has gotten past the pilot stage.
The honest framing up front: most of what follows is a pilot, a proof-of-concept, or a research partnership, not a production system replacing a classical one. That is not a criticism of the field. It is the accurate description of where quantum computing sits against most commercial workloads in 2026, and treating a pilot as a deployment is the single most common way quantum coverage overstates itself. Telecom, below, is the one vertical with a documented exception worth reading closely for exactly that reason.
Finance: fraud detection, portfolio risk, and derivatives pricing
Banking has more active quantum pilots than any other vertical, largely because two of its core problems, portfolio optimization and Monte Carlo-style risk simulation, map naturally onto both quantum annealing and gate-model algorithms like QAOA. D-Wave's new partnership with Nasdaq Verafin, covered here, targets anti-money-laundering pattern detection across a customer base of over 2,800 financial institutions, still at the proof-of-concept stage. Quantinuum and SoftBank's joint white paper names telecommunications fraud detection as a graph-analytics use case SoftBank is actively researching on Quantinuum hardware. Our portfolio optimization business case covers why the same combinatorial structure that makes fraud networks a quantum target also makes portfolio rebalancing one. What is missing across the finance vertical so far is a published head-to-head comparison against a classical baseline. Every announcement describes what is being tested, not a measured improvement over what the bank already runs.
Automotive: batteries, fuel cells, and traffic
BMW's relationship with Quantinuum, running since 2021 and expanded into a multi-year partnership this year, is the automotive sector's longest-running and best-documented quantum effort. The target problem is electrochemistry: simulating the oxygen reduction reaction at platinum catalysts, which determines fuel cell efficiency and is genuinely hard for classical methods because of the electron correlation effects involved. Other automakers have run parallel efforts with less public detail: traffic-flow and production-scheduling optimization work using quantum annealing hardware, and battery-chemistry simulation research with gate-model providers. The pattern across the vertical is consistent with BMW's case: automotive quantum work concentrates on chemistry (batteries, fuel cells, catalysts) and on combinatorial scheduling (traffic, paint-shop sequencing, logistics), not on anything resembling a production deployment replacing existing simulation tools.
Aerospace and defense: fluid dynamics and materials
Rolls-Royce, Riverlane, Quantinuum, and the University of Edinburgh's EPCC signed an agreement to test computational building blocks for gas turbine design on Quantinuum's Helios system, building on years of prior fluid-dynamics groundwork between Rolls-Royce, Riverlane, and EPCC specifically. Computational fluid dynamics is one of the more frequently cited plausible quantum use cases industry-wide, since classical CFD simulation is already extremely expensive at the fidelity aerospace design needs, but the Rolls-Royce collaboration's own framing, building blocks first, full simulations later, is a more honest scope than most aerospace-quantum announcements set for themselves. Defense-sector interest tends to run through national-security channels rather than public commercial partnerships: IonQ's memorandum of understanding with Sandia National Laboratories is a hardware co-design relationship rather than an applications pilot, aimed at government use cases that are not disclosed in detail.
Pharma and chemicals: molecule simulation
Drug discovery and materials chemistry are the use case quantum computing's proponents have pointed to longest, on the reasonable logic that simulating molecules is a quantum mechanical problem in the first place, and classical computers approximate it rather than compute it directly. Pharmaceutical and chemical companies have run research collaborations with gate-model quantum providers on molecular dynamics and reaction simulation, generally scoped to specific molecules or reaction pathways rather than end-to-end drug discovery pipelines. The BMW-Quantinuum electrochemistry work sits in this same technical category, chemistry simulation, even though the end application is batteries rather than a drug candidate, and it is the most publicly documented example of this use case producing incremental, published results over several years rather than a single splashy announcement.
Energy and utilities: grid optimization and materials
Energy companies have pursued two distinct quantum use cases that get conflated more often than they should. One is grid and load optimization, a combinatorial problem suited to annealing approaches, similar in shape to the traffic and scheduling problems automakers are testing. The other is materials and chemistry research aimed at things like battery storage and catalysts for cleaner energy production, which overlaps directly with the pharma and automotive chemistry work described above. Utilities exploring quantum computing tend to frame it as multi-year research rather than a near-term operational tool, which is a more accurate characterization than most vendor press releases about the energy vertical offer.
Telecom: network optimization and fraud
Telecom is where this list's usual caveat, no published comparison against a classical baseline, breaks down, and it is worth reading closely for that reason. AT&T expanded its use of D-Wave's quantum annealing technology on July 27, 2026, moving beyond a pilot into broader network operations: outage detection and response, technician routing, network build planning, and traffic management, with evaluation of D-Wave's forthcoming gate-model systems for quantum security and communications underway alongside it. AT&T's Director of Data Science reported cutting a specific network optimization workload from roughly an hour to under 15 seconds in early testing. That is a real, dated, named-workload number, not a framework or a stated research intent, which puts it in a different category than most of what else is in this piece. It is still a vendor-and-customer-reported figure rather than an independently reproduced benchmark, and "a specific workload" is not the same as AT&T's full network optimization stack, but it is the closest thing in this entire roundup to the classical-baseline comparison the rest of this piece keeps noting as absent. SoftBank's white paper with Quantinuum names two further research areas: quantum chemistry (adjacent to the automotive and pharma use cases above) and large-scale graph analytics for fraud detection, a natural fit given SoftBank's own telecommunications business generates exactly that kind of transaction and network-graph data, though that work remains at the research-intent stage AT&T's has moved past.
The pattern across every vertical
Look across finance, automotive, aerospace, pharma, energy, and telecom, and the same two problem shapes keep showing up regardless of industry: chemistry and materials simulation (batteries, fuel cells, catalysts, drug molecules), and combinatorial optimization over networks or schedules (fraud detection, traffic, grid load, portfolio risk). That is not a coincidence. Those are the two categories of problem where quantum mechanics or quantum-inspired optimization has a genuine theoretical edge over classical methods, and it is why the same handful of algorithmic approaches, VQE and chemistry-specific methods for the first category, QAOA and annealing for the second, keep reappearing across industries that otherwise share nothing in common.
How to read the next announcement in this space
Every partnership above is worth the same three questions before treating it as evidence quantum computing works for that use case: is there a published technical result, not only a stated intent, is there a classical baseline it is measured against, and is the claim coming from the hardware vendor, the customer, or an independent third party. AT&T's telecom result clears the first two bars and still fails the third, a vendor and its customer, not an outside party, which is why one dated speedup number changes how far along telecom looks without changing how the other verticals should be read. Our use cases overview tracks which applications across every industry have moved from research-stage to genuinely near-term, and is a useful cross-check the next time a company outside the quantum industry announces a new pilot.