The leading top superconducting quantum computing companies in 2026 build their qubits from superconducting circuits wired around Josephson junctions. Those circuits sit inside a dilution refrigerator, cooled to millikelvin temperatures, thousandths of a degree above absolute zero, in an approach that has anchored the industry’s roadmaps for two decades. It runs from the first Yale and Saclay experiments of the early 2000s through to Google’s Willow chip, which in December 2024 showed errors falling away as the error-correcting code around them grew. Nine commercial vendors make up the top superconducting quantum computing companies. They fall into two camps. Eight build transmons: IBM Quantum, Google Quantum AI, Rigetti, IQM, OQC, SeeQC, QuantWare and Anyon Systems. OQC builds its version, the coaxmon, in three dimensions, so that the control wiring reaches the chip from behind rather than across it. Alice & Bob is the exception, with a bosonic cat-qubit design of its own. Consolidation then reshaped the field across 2025 and 2026, when Google bought Atlantic Quantum and D-Wave bought Quantum Circuits Inc.
Why superconducting still anchors the modality
Superconducting hardware has the deepest deployed-customer base of any approach to quantum computing. The lead is two decades old. It starts with the first transmon experiments at Yale and Saclay, and it has been reinforced ever since by where the chips are made. The manufacturing is not exotic. Superconducting processors come off the same kind of CMOS foundry line that produces ordinary silicon. IBM Research runs one, Google has its own fab in Santa Barbara, and a growing number of European foundries have joined them, among them Forschungszentrum Jülich, VTT and IMS CHIPS. IBM Quantum on its own puts more deployed-qubit hours in front of enterprise customers than the rest of the industry combined. Google supplied the other landmark. Its Willow chip in December 2024 produced the first published exponential error suppression on a working hardware platform.
Through 2025 and 2026 the work has been about one gap. Trapped-ion machines hold the fidelity record at 99.99%, and the superconducting vendors have spent those two years closing on it without surrendering their lead in qubit count. The published numbers show the state of that race. IBM’s Heron R2 runs 156 qubits at 99.5% two-qubit fidelity, IQM’s Radiance 150 qubits at 99.91%, and Rigetti’s Ankaa-3 108 qubits at 99.1%. SeeQC gets 98.4% out of 64 qubits with the control electronics built onto the chip. Alice & Bob is measured on something else entirely. Its Boson 4 cat qubits now hold a bit against an unwanted flip for a full hour, which already crossed the 2030 roadmap threshold a year early.
The route to fault tolerance looks the same at every vendor: add physical qubits, push two-qubit fidelity past the threshold where error correction starts to pay, then show logical qubits doing real work. The codes are where they differ. IBM and Google use surface codes, Alice & Bob uses bosonic codes, and IBM’s 2026 Kookaburra roadmap brings in the newer high-rate qLDPC codes.
How superconducting quantum computing works
A superconducting processor is, physically, a circuit patterned on a chip of silicon or sapphire. By the standards of a qubit that circuit is macroscopic, which means large enough to see. What makes it quantum is the Josephson junction at its heart, the element that turns the circuit into an oscillator with unevenly spaced energy levels, two of which serve as the qubit. The transmon is the dominant design. Fluxonium and cat-qubit encoding are the architectural alternatives. Control arrives as microwave pulses at GHz frequencies, sent down coaxial cable from room-temperature electronics to the millikelvin stage of the fridge, roughly 15 millikelvin in a working machine. Two-qubit gates are usually a controlled-Z through a tunable coupler, or a cross-resonance gate. Reading a qubit out means listening to a resonator coupled to it, and that readout is already quick enough to support mid-circuit measurement on production hardware. The refrigerator is not a luxury. It is the one thing keeping the qubit’s superconducting wavefunction coherent.
The wiring bottleneck and how vendors are solving it
Scaling has always run into the same wall. That wall is wiring. Every qubit needs a drive line, a readout line and one or more coupler lines, and each of those has to run down through the dilution fridge. Package a 1,000-qubit machine the conventional way and you are threading thousands of coaxial cables into it. The fix that emerged between 2024 and 2026 is to move the wiring onto the chip itself. SeeQC put single-flux-quantum control, which is classical electronics, directly on the qubit chip, and QuantWare’s VIO technology takes the connections vertically in three dimensions rather than out sideways. IBM’s Heron generation leans on tunable couplers and chiplet-style connectivity to hold the wire count down. Help is arriving from a fourth direction as well, with Intel, Diraq and other silicon-spin players starting to ship cryogenic CMOS control electronics that answer the same problem across every modality.
The top superconducting quantum computing companies
Nine commercial vendors define the top superconducting quantum computing companies in 2026, and they divide neatly. Two are the research giants with deployed customers, IBM Quantum and Google Quantum AI, and two are publicly-traded pure-plays, Rigetti on Nasdaq since 2021 and IQM on Nasdaq since July 2026. One is the cat-qubit specialist, Alice & Bob. The last four are mid-stage hardware firms, each with its own bet. OQC puts its own machines inside commercial data centres, SeeQC puts the control electronics on the chip, QuantWare supplies chips to other builders, and Anyon Systems deploys full-stack machines in Canada. The QED-C industry consortium tracks the top superconducting quantum computing companies alongside the wider quantum-hardware industry, with published quarterly status updates on QPU access and deployments.
Independent directories of the top superconducting quantum computing companies list much the same shortlist of names. The profiles below take the leading firms one at a time.
What the lineup reveals
Three things stand out in that lineup. The first is reach. Superconducting has by far the deepest deployed-customer base of any modality, anchored by the 250+ enterprise members of the IBM Quantum Network and by Google’s research-access programmes. Between them the two giants account for the majority of all published superconducting-quantum-computing research, which leaves everyone else to compete on architecture rather than on raw qubit count.
The second is the public market. The exposure there is steepening. Rigetti has traded on Nasdaq since its 2021 SPAC, and IQM joined it there in July 2026 through the $1.8B merger with Real Asset Acquisition Corp. SeeQC changed course during 2026, when the Allegro Merger Corp deal was terminated in August and the company filed for a direct listing of its own instead. Alice & Bob has stayed private. It raised a EUR 100M Series B in January 2025 and an undisclosed extension of it in May 2026.
The third is variety, and no other approach runs as many designs at once. Transmons dominate. Beside them, four other routes are in production in 2026. Google has fluxonium qubits through the Atlantic Quantum acquisition, Alice & Bob has cat qubits, Quantum Circuits Inc inside D-Wave has dual-rail cavity qubits, and SeeQC has SFQ on-chip control. Trapped-ion offers a choice between laser-driven and microwave-driven gates, and neutral-atom between analog and gate-model operation, which leaves superconducting looking like the experimental playground rather than a settled architecture.
Transmon, cat, and fluxonium: the qubit-type fork
Every vendor faces the same fork in the road. The question is which Josephson-junction qubit to build on. The transmon is the historical default, and the design behind IBM Heron, Google Willow, Rigetti Ankaa-3, IQM Radiance and SeeQC. Its energy levels are unevenly spaced, which is what lets a microwave pulse address two of them and ignore the rest, and it is comparatively straightforward to fabricate. That has made it the industry consensus since the early 2010s, and the price is sensitivity to charge noise and a fidelity ceiling that has kept transmons just below the trapped-ion gate-fidelity record.
Alice & Bob took the other road. A cat qubit lives in a microwave cavity, in a superposition of coherent states of the mode inside it, and the conservation of photon-number parity protects it structurally against bit-flip errors. The September 2025 Boson 4 milestone, a bit-flip lifetime of more than one hour, is the empirical proof that a bosonic code can suppress errors in the hardware itself, with no active correction running. Fluxonium is the third option, used by Google through Atlantic Quantum and by Quantum Circuits Inc inside D-Wave. It is harder to fabricate. In return it offers a more unevenly spaced energy ladder, better drive selectivity, and room to embed the control electronics in the qubit chip in next-generation architectures.
The 2025-2026 consolidation wave
The most consequential structural change of 2025 and 2026 was a wave of acquisitions. It reshaped the middle of the field. Google Quantum AI acquired Atlantic Quantum in October 2025 and folded its fluxonium qubits and embedded control into the roadmap for Willow’s successor. D-Wave Quantum announced a $550M acquisition of Quantum Circuits Inc on 7 January 2026, made up of $300M in stock and $250M in cash, and completed it on 20 January 2026. That deal gave D-Wave a dual-rail-cavity gate-model platform to sit beside its annealing systems, and it also brought Robert Schoelkopf’s Yale group inside the D-Wave roadmap.
Why the rollup is happening now
The logic is the one that drove IonQ’s rollup in trapped-ion. Enterprise customers want a single supplier for the compute and for the architectural hedge sitting next to it, and the smaller vendors traded independence for capital and for access to those customers. SeeQC, QuantWare and Anyon Systems are the plausible next cohort over the coming 24 months. Alice & Bob sits outside all of it, for the simple reason that no other superconducting vendor builds on the bosonic-code architecture.
Federal money is pushing in the same direction. In May 2026 the Commerce Department and NIST directed $2.013 billion of CHIPS Act awards to nine quantum companies, and took minority equity stakes in those companies in exchange. That list counted IBM, Rigetti, and D-Wave among the superconducting recipients. IBM’s quantum foundry subsidiary, Anderon, then finalised a $1 billion CHIPS award with the Department of Commerce in September 2026, alongside roughly $1 billion of IBM’s own money. The United States is no longer only funding research into superconducting manufacturing. It is putting capital into the domestic capacity at scale.
When superconducting matters for your industry
Pharmaceutical and chemistry
Two properties matter here. A large qubit count and a fast gate time make superconducting hardware a good fit for variational chemistry on small and mid-sized molecules. IBM Quantum hosts the largest deployed pharma-research base, with BASF, Boehringer Ingelheim, BMW, Mitsubishi Chemical, Merck KGaA, ExxonMobil and JPMorgan on the list. Its Heron R2 machines running Qiskit Patterns are the most-used commercial chemistry pipeline, and Google’s published Fermi-Hubbard condensed-matter simulation on Willow is the academic benchmark for the modality.
Optimisation, finance, and operations research
Optimisation work is more forgiving. Quantum approximate optimisation, known as QAOA, rewards a fast gate cycle and a high qubit count, and it tolerates a fidelity that sits below the trapped-ion record. IBM’s partnerships with JPMorgan Chase, Goldman Sachs, HSBC and BMW anchor the financial-services use case. IQM’s Polish HPC partnership runs optimisation on the Radiance fleet, and Rigetti’s C-DAC India deal has made superconducting the dominant modality where a government wants quantum hardware inside its own HPC estate. The same combinatorial problems are home ground for D-Wave’s annealing systems, which is part of why D-Wave acquired Quantum Circuits Inc. The purchase hedges the question of whether the gate-model or the annealing approach wins on optimisation.
Government and defence
In US Defense-Department buying, superconducting is the dominant modality. The orders run through IBM, Google and Rigetti, all three of them funded by the DARPA Quantum Benchmarking Initiative on parallel fault-tolerance roadmaps. In Europe the European Defence Agency funds IQM and the broader EuroHPC quantum push. The UK MOD partners with Rigetti through the $100M UK investment programme, and Anyon Systems supplies the Canadian government quantum testbeds. IBM, Google and Rigetti are aiming at a utility-scale superconducting quantum computer ready for defence work by 2030 to 2033, the same horizon IonQ targets in trapped-ion and PsiQuantum in photonic.
Frequently asked questions
Who are the leading top superconducting quantum computing companies in 2026?
Nine commercial vendors define the modality. IBM Quantum (Heron R2 at 156 qubits and 99.5% 2Q fidelity, Condor at 1,121 qubits, Nighthawk at 120 qubits, Kookaburra qLDPC roadmap 2026) and Google Quantum AI (105-qubit Willow with one verified logical qubit, surface-code milestone December 2024) are the foundational giants. Rigetti Computing (Nasdaq: RGTI, Ankaa-3 + Cepheus-1 108Q) and IQM Quantum Computers (Espoo Finland, 150Q Radiance, Nasdaq: IQMX since July 2026) are the publicly-traded or imminent pure-plays. Alice & Bob (Paris, cat qubits, 16Q Boson 4 with one-hour bit-flip lifetime) builds the bosonic-code alternative. OQC (Reading UK, 32-qubit Toshiko in data centres across the UK, Japan and Spain, £260M Series C June 2026), SeeQC (NY/London, 64Q with on-chip SFQ control), QuantWare (Amsterdam, third-party chip fab supplier), and Anyon Systems (Canada, MonarQ 24Q full-stack) round out the lineup.
What is a transmon qubit?
A transmon is the dominant superconducting qubit type used by IBM, Google, Rigetti, IQM, SeeQC, and most other vendors in the modality. The transmon is an anharmonic LC oscillator built from a Josephson junction shunted by a large capacitance, designed to suppress sensitivity to charge noise while preserving the energy-level anharmonicity that lets microwave pulses address two specific levels as a qubit. Transmons are easier to fabricate than fluxonium and have been the consensus superconducting-qubit choice since the early 2010s, but their fidelity ceiling has stayed below the trapped-ion record because of residual decoherence channels that are intrinsic to the transmon design.
How many qubits do the top superconducting quantum computing companies operate?
IBM Quantum holds the published qubit-count record at 1,121 (the Condor chip, December 2023), with the production Heron R2 at 156 qubits and 99.5% two-qubit fidelity. IQM Radiance runs 150 qubits at 99.91% two-qubit fidelity. Google Willow runs 105 qubits with one verified logical qubit. Rigetti Ankaa-3 and Cepheus-1 run 108 qubits at 99.1% two-qubit fidelity. SeeQC runs 64 qubits at 98.4% two-qubit fidelity with on-chip SFQ control. Anyon Systems MonarQ runs 24 qubits in a data-centre-compatible form factor. Alice & Bob Boson 4 runs 16 cat qubits with bit-flip lifetimes exceeding one hour. QuantWare Contralto runs 25 qubits, with the VIO-40K platform targeting up to 10,000 qubits. OQC Toshiko runs 32 qubits in commercial data centres, and the newer GENESIS runs 16 dual-rail Dimon qubits.
Why is IBM such a dominant player in superconducting?
IBM Quantum has run the deepest superconducting research programme in the world since the early 2000s, anchored by a vertically-integrated stack (chip design at IBM Research Yorktown Heights, fabrication at IBM’s in-house clean room, cryogenic packaging at IBM Quantum, the Qiskit SDK, IBM Cloud delivery, and the IBM Quantum Network of 250+ enterprise members). The roadmap is published in advance (Heron, Nighthawk, Kookaburra, Cockatoo, Starling, Blue Jay) with concrete qubit-count and logical-qubit targets, and IBM’s 1,121-qubit Condor (December 2023) and Heron R2 at 99.5% two-qubit fidelity (December 2024) remain the deepest commercial deployment in the industry. The 2026 Kookaburra programme introduces qLDPC quantum memory plus a Logical Processing Unit for the first time, targeting 7,500 gates on up to 360 qubits.
Are superconducting quantum computing companies publicly traded?
Rigetti Computing is publicly traded on Nasdaq under ticker RGTI following the 2021 SPAC merger with Supernova Partners ($458M). IQM Quantum Computers closed its $1.8B SPAC merger with Real Asset Acquisition Corp on 1 July 2026 and trades on Nasdaq as IQMX, the first publicly-listed European quantum-computing company. SeeQC’s $1B Allegro Merger Corp deal was terminated on 25 August 2026, and the company filed its own registration statement on 11 September 2026 for a direct listing that is not yet effective. Alice & Bob, QuantWare (which raised a $178M Series B in May 2026), Anyon Systems, and the foundational IBM Quantum and Google Quantum AI divisions inside their parent companies all remain private. The 2026 vintage took superconducting from one publicly-listed pure-play to two.
How does superconducting compare with trapped-ion and neutral-atom modalities?
Superconducting wins on deployed-qubit count (1,121 published on IBM Condor versus 256 on QuEra Aquila and tens on trapped-ion systems) and on gate speed (nanoseconds versus microseconds), at the cost of below-99.99% two-qubit fidelity and the cryogenic overhead of a dilution refrigerator. Trapped-ion delivers higher gate fidelities (99.99% on IonQ) but smaller qubit counts (tens to a few hundred) and slower gate cycles. Neutral-atom runs at thousands of qubits (1,180 on Atom Computing Phoenix) with reconfigurable software-defined connectivity, but with somewhat lower gate fidelities than either superconducting or trapped-ion. In 2026 the three modalities complement rather than compete on most production workloads, and the choice is driven more by the application than by raw benchmark comparison.
What is a cat qubit and why does Alice & Bob use it?
A cat qubit is a bosonic-mode encoding where the qubit lives in the superposition of two coherent states of a microwave-cavity mode (the metaphor is the Schrödinger cat in a superposition of alive and dead). The geometric structure of the encoding makes the cat qubit intrinsically resistant to bit-flip errors because flipping requires the entire coherent state to migrate from one phase-space pocket to the other, which is exponentially suppressed in the photon-number expectation. The September 2025 Alice & Bob Boson 4 milestone of a one-hour bit-flip lifetime is the empirical proof that the architecture delivers hardware-level error suppression, and the fault-tolerance roadmap for cat qubits needs far fewer physical qubits per logical qubit than the surface-code approach used by IBM and Google.
What is the relationship between superconducting and quantum error correction?
Superconducting hardware produced the first published exponential error suppression on a working platform when Google’s Willow chip demonstrated in December 2024 that surface-code logical-error rates fall exponentially as the code distance grows. The current published logical-qubit demonstrations span Google Willow (one verified logical qubit on the surface code), IBM (planned 2026 Kookaburra qLDPC), Quantinuum Helios (48 logical qubits on the trapped-ion side), Atom Computing (24 logical qubits via Microsoft), and QuEra, whose neutral-atom processor held up to 96 logical qubits active at once in one demonstration of its 448-atom architecture, reported in Nature. That paper summarises its own deep-circuit results as dozens of logical qubits, so the 96 is the peak of a single experiment rather than a settled record. The IBM 2026 Kookaburra roadmap introduces high-rate qLDPC codes plus a Logical Processing Unit, the architectural primitive that lets useful fault-tolerant computation run with manageable physical-qubit overhead. See our quantum logical-qubit leaderboard for the latest verified numbers across all modalities.
Disclaimer. This article is for informational purposes only and does not constitute investment, financial or professional advice. The quantum technology industry evolves rapidly and information may become outdated. Always conduct your own research and consult qualified advisers before making investment decisions. Investing in quantum computing companies involves significant risk, including the potential loss of your entire investment, and past performance is not indicative of future results.
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
