US Quantum Computing Companies 2026

IBM has made a deliberate bet that the race to fault tolerance will be won not just by better qubits, but by better classical integration — and in late 2025 it began to prove the point.

Where Google and Microsoft are chasing hardware purity, IBM Quantum under VP Jay Gambetta has built the most commercially embedded quantum programme in existence. The IBM Quantum Network spans over 300 organisations — including CERN, Airbus, Daimler and MIT — running real experiments on cloud-connected processors as part of production workflows. This operational breadth gives IBM something no competitor can match: a vast empirical dataset about what quantum circuits actually fail in practice. The November 2025 Nighthawk processor (120 qubits) introduced next-generation tunable couplers delivering roughly 30% more circuit complexity than its Heron predecessor. More telling was a separate IBM-AMD breakthrough: using commercial FPGA chips to achieve real-time quantum error correction a full year ahead of schedule — demolishing the assumption that bespoke quantum control hardware is a prerequisite for fault tolerance.
IBM’s roadmap is the most detailed in the industry. Kookaburra (2026) introduces multi-chip quantum communication links. Starling (targeting 2028–29) aims for 200 logical qubits from roughly 10,000 physical qubits using LDPC codes that IBM claims require 90% fewer physical qubits than surface codes. Stated goals: verified quantum advantage by end-2026, a large-scale fault-tolerant machine by 2029. A Cisco partnership targets networked quantum infrastructure. In February 2026, IBM Ventures backed SQK and QodeX Quantum from Chicago’s Duality accelerator as part of its regional anchor strategy at the Illinois Quantum and Microelectronics Park.

Google is, at this moment, the undisputed leader on the single most important metric in quantum computing: demonstrating a computation that classical machines provably cannot replicate.

Google Quantum AI made two defining announcements in quick succession. In December 2024, the Willow chip — a 105-qubit superconducting processor — delivered what the field had been seeking for thirty years: as qubits are added, error rates fall rather than rise. Single-qubit fidelity reached 99.97%, two-qubit fidelity 99.88%. Then in October 2025, Google published in Nature the first verifiable quantum advantage on hardware: the Quantum Echoes algorithm running 13,000 times faster on Willow than the best classical algorithm on a supercomputer, with results that can be independently replicated. Critically, the team used the same algorithm to study molecular behaviour with UC Berkeley, taking a tangible step toward drug discovery applications rather than benchmark theatre.
A third milestone, less reported but equally significant: Google achieved 99.99% fidelity in magic state cultivation in 2025 — a 40-fold improvement that makes fault-tolerant computation materially more resource-efficient. Google also acquired Atlantic Quantum to strengthen its hardware team, and made a strategic investment in QuEra Computing as a hedge into neutral-atom architectures. Its next stated milestone is a long-lived logical qubit — the missing link between current hardware performance and practical quantum computing.

Microsoft is playing the longest game in the room — betting on a qubit physics so different from IBM and Google that, if it works, it makes every other architecture look inefficient. The risk is commensurate with the ambition.

Microsoft Azure Quantum, under quantum lead Krysta Svore, introduced the Majorana 1 chip in February 2025 — the first processor built on a Topological Core using topoconductor materials, a novel class of matter that exploits semiconductor-superconductor nanowire junctions to host Majorana zero modes. Topological qubits store information in the non-local topology of the quantum state, making them inherently resilient to local perturbations. The theoretical upside is profound: far fewer physical qubits per logical qubit than any error-correction code requires today, with Microsoft claiming a path to one million qubits on a single chip. Independent scientists have noted that producing Majorana zero modes reliably has been an elusive challenge for over a decade; DARPA, however, selected Microsoft for the final phase of its Underexplored Systems for Utility-Scale Quantum Computing (US2QC) programme, which is not a trivial endorsement.
In January 2026, Microsoft released an updated open-source Quantum Development Kit (QDK) with chemistry-aware algorithms that reduce circuit gate counts from thousands to single digits for molecular simulation — a practical near-term contribution regardless of when Majorana hardware matures. Azure Quantum also provides cloud access to IonQ, Quantinuum, Rigetti and Atom Computing systems, capturing commercial quantum revenue at scale while its own hardware matures. CEO Satya Nadella projects commercial quantum systems operating in data centres by 2029.

AWS is the only major tech company running a credible quantum hardware research programme while simultaneously operating the quantum industry’s dominant cloud marketplace — a structural position that lets it profit from the sector’s development regardless of which qubit modality wins.

The Ocelot chip (February 2025) from the AWS Center for Quantum Computing in Pasadena is a superconducting processor built on cat qubits — microwave photons engineered to suppress bit-flip errors exponentially, leaving only the far simpler phase-flip errors for classical correction overhead. AWS claims this architecture reduces the physical qubit requirement per logical qubit by up to 90% compared to conventional transmon qubits, a figure that — if it holds at scale — changes the economics of fault-tolerant quantum computing fundamentally. The demonstration showed error-correction benefits scaling with code size, mirroring Google’s Willow below-threshold result but using significantly fewer resources per logical qubit.
In parallel, Amazon Braket is the most hardware-agnostic quantum cloud platform commercially available, offering pay-per-use access to systems from IonQ, Rigetti, QuEra and IQM alongside GPU-backed simulators. Integration with SageMaker enables quantum-classical hybrid workflows in production ML pipelines. AWS is, in essence, the Switzerland of the quantum hardware wars: it sells access to all combatants and quietly develops its own weapons in the background.

On 25 February 2026, IonQ became the first quantum computing company in history to report more than $100 million in annual GAAP revenue. That is not a rounding error in a press release — it is a structural milestone that separates IonQ from every other pure-play in the sector.

Full-year 2025 revenues came in at $130 million, up 202% year-on-year, with Q4 alone generating $61.9 million — a 429% surge, beating IonQ’s own guidance by 55%. The company ended 2025 with $3.3 billion in cash and investments, more financial runway than any pure-play quantum hardware company in the world. Some 60% of revenue is now from commercial customers and 30% international, indicating genuine market breadth rather than dependence on a single government programme. For 2026, IonQ is guiding for $225–245 million in revenue, implying 73–88% growth. At that trajectory, IonQ is not a speculative quantum play. It is a fast-growing technology company.
The technical story supports the commercial one. IonQ achieved a world-record 99.99% two-qubit gate fidelity in 2025, enabled by its shift from ytterbium to barium atoms and integration of microwave control IP from its Oxford Ionics acquisition — replacing laser arrays that make conventional trapped-ion systems difficult to scale. South Korea’s KISTI purchased a 100-qubit IonQ system to anchor that country’s flagship quantum-classical computing facility. Most ambitiously, IonQ announced plans to acquire SkyWater Technology — the leading quantum chip foundry — to create what it calls the world’s first full-stack quantum merchant supply chain. DARPA selected IonQ for Phase B of its Quantum Benchmarking Initiative.

Quantinuum produces the most accurate commercial quantum computer on earth — and it is about to become a public company at a $10 billion valuation. Its Helios processor sets the performance benchmark that every competitor is chasing.

Formed in 2021 from the merger of Honeywell Quantum Solutions and UK startup Cambridge Quantum Computing, Quantinuum launched its third-generation Helios processor in November 2025 to specifications that demand attention: 98 all-to-all connected physical qubits, 99.9975% single-qubit gate fidelity and 99.921% two-qubit gate fidelity across every qubit pair. Helios supports 94 logical qubits in error-detected mode and demonstrated 48 fully error-corrected logical qubits via code concatenation. Launch customers include Amgen, BMW Group, JPMorganChase and SoftBank Corp — a group spanning pharma, automotive, financial services and telecom simultaneously. NVIDIA has integrated its GB200 superchips with Helios via NVQLink for real-time quantum error decoding, treating error correction as a dynamic GPU-accelerated process rather than a static classical post-processor. DARPA advanced Quantinuum to Stage B of its Quantum Benchmarking Initiative.
On 14 January 2026, Honeywell announced Quantinuum’s confidential S-1 submission to the SEC, formally beginning the most anticipated quantum IPO in history. The company last raised at a $10 billion pre-money valuation in September 2025, backed by $600 million from JPMorgan, Mitsui, NVIDIA and Amgen. Honeywell CEO Vimal Kapur has stated publicly that Honeywell plans to exit its approximately 54% stake within 12–36 months of listing, giving it a clear divestiture incentive to support the offering price. On Quantinuum’s technical roadmap: Sol (2027) targets a 2D-grid ion trap design at 192 physical qubits; Apollo aims for thousands of physical qubits as the company’s first genuinely fault-tolerant computation vehicle.

In 2025, QuEra and its academic partners at Harvard, MIT and Yale resolved three barriers to fault-tolerant quantum computing that the field had been working around for a decade. They did so in a single year, with results published in Nature.

The first barrier was atom loss: neutral-atom systems had previously struggled to operate continuously because atoms escape their optical traps, corrupting the computation. A Harvard-MIT team demonstrated a 3,000-qubit array operating for over two hours by replenishing lost atoms mid-computation — effectively the first quantum computer that runs continuously without stopping to reload. The second barrier was below-threshold error correction: increasing system size reduces logical error rates, demonstrated with up to 96 logical qubits — the result that proves neutral-atom architectures can scale toward fault tolerance. The third barrier was magic state distillation on a neutral-atom platform, demonstrating that the expensive state preparation required for universal quantum algorithms can be performed with acceptable overhead. All three results, published in Nature, represent the comprehensive architectural validation that CEO Andy Ory described as “the year fault tolerance became real.”
QuEra closed over $230 million from Google Quantum AI, SoftBank Vision Fund 2 and NVIDIA’s NVentures, earmarked for manufacturing scale and global deployment. On the commercial side, QuEra completed its first on-premises HPC installation at Japan’s AIST — operating its Gemini gate-based system alongside an NVIDIA-powered supercomputer — and delivered a second Gemini system to the UK’s National Quantum Computing Centre at Harwell. QuEra was selected for NERSC’s 2026 Research Access Programme and forged enterprise alliances with Deloitte Japan and BCG X. Third-generation systems are targeted for 2026–2027. The neutral-atom structural advantage over superconducting systems is not trivial: room-temperature operation, low power consumption, and wirelessly controlled identical qubits that eliminate the wiring complexity that increasingly constrains superconducting scaling.

Infleqtion solves a problem that IonQ, QuEra and Quantinuum do not: it generates quantum technology revenue from defence and sensing contracts even before its quantum computers achieve advantage. That dual-platform model makes it the most commercially protected hardware pure-play now trading on US exchanges.

Infleqtion (formerly ColdQuanta, founded 2007) became the first neutral-atom quantum company to list on a US exchange when it began trading under INFQ on 17 February 2026, following its SPAC merger with Churchill Capital Corp X that raised over $550 million, valuing the company at approximately $1.8 billion. The computing platform centres on the Hilbert quantum computer (1,600-qubit neutral-atom array) paired with the Superstaq cloud software platform that optimises circuit performance across multiple hardware backends. The long-term Sqorpius roadmap targets over 100 logical qubits by 2028 using up to 40,000 physical qubits. The sensing platform produces quantum sensors for defence, aerospace and navigation — including the Tiqker commercial atomic clock — that generate near-term revenue that hardware-only peers cannot match.
Two announcements in the week of listing illustrated the commercial pipeline’s breadth. A $6.2 million ARPA-E contract (ENCODE project) will deploy 1,600-qubit neutral-atom arrays alongside Superstaq to optimise the US power grid — applying quantum computing to one of the most consequential real-world optimisation problems in existence. And in partnership with NASA JPL, Infleqtion is developing the Quantum Gravity Gradiometer Pathfinder: the world’s first space-based standalone quantum gravity sensor, funded at $20 million, designed to map Earth’s mass changes — from melting ice to shifting groundwater — with ten times the sensitivity of any classical gravimeter. Both projects exemplify why the dual computing-sensing model gives Infleqtion a commercial durability that pure-play hardware companies will spend years trying to replicate.

Atom Computing is the quietest significant quantum company in the United States. It has minimal public profile and two partnerships that validate it as one of the most technically serious neutral-atom hardware developers anywhere: DARPA and Microsoft.

Atom Computing uses nuclear-spin qubits in strontium atoms — a technically demanding choice that delivers among the longest coherence times in the industry, enabling the sustained operations that error correction requires. In a landmark Microsoft collaboration, Atom Computing achieved 24 fully entangled logical qubits — a world record for neutral atoms at the time and a milestone that moved the field’s debate from “can neutral atoms do error correction?” to “how fast can they scale?” DARPA subsequently selected Atom Computing for its US2QC programme, validating the Phoenix platform as a credible alternative architecture for practical quantum computing. In a concrete international deployment, Atom Computing and Microsoft are jointly delivering an error-corrected quantum computer to Denmark’s Export and Investment Fund and the Novo Nordisk Foundation — one of the earliest physical deployments of a neutral-atom error-corrected machine outside a research setting. CPO Justin Ging has noted that both QuEra and Atom Computing anticipate placing 100,000 atoms in a single vacuum chamber within a few years, a number that would put neutral-atom systems well ahead of superconducting approaches in raw qubit scalability.

PsiQuantum is building quantum computers in the only way it believes a billion-qubit machine can actually be constructed: using the same semiconductor fabs that make iPhones, scaled to the physics of single photons rather than electrons. The ambition is extraordinary. So is the capital required to pursue it.

PsiQuantum entered 2026 with a pivotal leadership transition. On 10 February 2026, Victor Peng — former CEO of Xilinx and President of AMD, an executive who guided two major semiconductor platform transitions — became Interim CEO, while co-founder Jeremy O’Brien moved to Executive Chairman. The signal is clear: the company is shifting from decade-long R&D into deployment execution. The backdrop is substantial: a $1 billion Series E closed in September 2025, groundbreaking at the Illinois Quantum and Microelectronics Park (IQMP) in Chicago — the largest quantum computing infrastructure project in US history — and simultaneous construction at a Brisbane, Australia site anchored by $620 million AUD in combined government funding. Linde Engineering is building the world’s largest cryogenic plant for a quantum computer at the Brisbane campus.
PsiQuantum’s technological distinction is manufacturability. Its Omega silicon photonic chipset is fabricated using standard processes at GlobalFoundries in New York — meaning the supply chain scales with the global semiconductor industry rather than being constrained by specialist quantum hardware facilities. Photonic qubits carry quantum information as single photons through on-chip waveguides, interacting via beamsplitters and phase shifters. They are naturally robust against thermal noise, but photon loss requires large physical qubit counts to produce a practical logical qubit count — PsiQuantum’s architecture requires millions of physical qubits for useful computation, which is precisely why it needs semiconductor-grade manufacturing. DARPA advanced PsiQuantum to QBI’s final validation phase. The company launched Construct in 2025, a software platform for designing fault-tolerant algorithms on the Omega architecture. Industrial partners include Airbus, Lockheed Martin and NVIDIA. A public offering is widely anticipated in 2026.

Quantum Computing Inc is the most discussed company in the quantum retail investor community — a publicly traded name carrying a $1.88 billion market capitalisation that reported $682,000 in full-year 2025 revenue. That gap between valuation and fundamentals defines both the investment thesis and the sceptic’s case.

QCi (QUBT) is evolving from a quantum software company toward a vertically integrated photonics hardware business, driven by CEO Dr Yuping Huang, who assumed the permanent role on 1 January 2026. The commercial pivot centres on thin-film lithium niobate (TFLN) chip technology — the photonic integrated circuit material used in ultra-high-speed optical communications — which QCi is positioning as the substrate for quantum-enabled photonic computing. The Fab 1 facility in Tempe, Arizona, opened in 2025, is currently operating as an R&D and small-batch manufacturing site. In Q4 2025, QCi unveiled Neurawave, a photonic reservoir computing system for AI workloads that operates without cryogenic cooling, offering a genuinely near-term differentiator. Q4 revenue of $198,000 (up 219% year-on-year) missed the $394,000 consensus; full-year 2025 revenue was $682,000. On the same day those results were released, QCi completed the $110 million acquisition of Luminar Semiconductor, adding laser, detector and packaging manufacturing capabilities alongside an established defence and aerospace customer base.
The balance sheet tells a different story from the income statement. QCi raised $1.55 billion in 2025, ending the year with $1.52 billion in combined cash and investments — runway measured in years. Analysts project approximately $6.9 million in 2026 revenue, implying a forward price-to-sales ratio still exceeding 270x at the current market cap. The company’s trajectory — from quantum software startup to photonic chip manufacturer to (potentially) quantum hardware producer — reflects a coherent if ambitious strategy. Whether the photonic quantum optics technology delivers at commercial scale, and whether the current valuation reflects that potential appropriately, is a question investors must resolve for themselves.

Rigetti occupies a peculiar position in the quantum landscape: it is the only publicly traded company that designs, fabricates and operates its own full-stack superconducting systems, giving it manufacturing capabilities that IBM and Google access only through much larger internal programmes. That vertical integration is a double-edged sword.

Rigetti Computing, led by CEO Dr Subodh Kulkarni, controls its entire stack from chip design through the Fab-1 semiconductor facility in Fremont, California, to cloud delivery via Quantum Cloud Services. This means Rigetti can innovate at the fabrication process level in ways that fabless quantum companies cannot — but every capital expenditure hits its own P&L, creating a cost structure that investors scrutinise closely. The current Ankaa-3 system operates at 84 qubits. The 108-qubit Cepheus-1-108Q chiplet-based system, delayed from its original late-2025 target after a technical revision in January 2026, is expected to reach general availability around the end of Q1 2026. The roadmap calls for 150+ qubits at 99.7% two-qubit fidelity by end-2026, and a 1,000+ qubit multi-chip architecture at 99.8% fidelity by end-2027.
The most significant recent commercial development was an $8.4 million purchase order from India’s Centre for Development of Advanced Computing (C-DAC) in January 2026 — Rigetti’s largest international hardware sale and a validation that non-US government quantum procurement markets are beginning to open at meaningful deal sizes. Rigetti participates in NVIDIA’s NVQLink GPU-QPU integration initiative. The company holds approximately $450 million in cash as of Q3 2025, sufficient runway for the 2027 multi-chip roadmap. Full-year 2025 results are due 4 March 2026. Analysts do not project profitability before 2030, making Rigetti a pure technology and roadmap execution bet.

D-Wave entered 2026 as the only quantum company in the world claiming simultaneous commercial traction in both quantum annealing and error-corrected gate-model computing — a dual-platform position it secured through an audacious acquisition that fundamentally changes what D-Wave is.

D-Wave reported full-year 2025 revenue of $24.6 million on 26 February 2026, up 179% year-on-year, with an 82.6% gross margin that would be extraordinary even for a pure software company. Revenue came from over 135 customers including 70+ commercial enterprises and more than two dozen Forbes Global 2000 companies, suggesting genuine enterprise penetration. Advantage2 system usage surged 314% year-on-year. The operational story, though, is dominated by the January 2026 acquisition of Quantum Circuits Inc. — the Yale spinout co-founded by Rob Schoelkopf, the inventor of the transmon qubit — making D-Wave the world’s first and only dual-platform quantum company spanning annealing and superconducting gate-model architectures. QCI’s dual-rail qubits use built-in erasure detection that identifies 90% of errors before propagation, delivering gate fidelities above 99.9% at speeds approaching superconducting systems. An initial gate-model system is targeted later in 2026 from a new R&D centre in New Haven, Connecticut.
First-quarter 2026 bookings already exceeded $32.8 million as of 25 February — before Q1 even closed — including a $20 million Advantage2 purchase by Florida Atlantic University and a $10 million two-year QCaaS agreement with a Fortune 100 company. D-Wave is relocating its corporate headquarters from Palo Alto to Boca Raton, Florida, establishing a new US government business unit to pursue federal defence contracts, and building a quantum R&D facility at the Boca Raton Innovation Center.

SandboxAQ may have the smartest business model in the quantum sector: it sells urgently needed products today — post-quantum cryptography tools, quantum navigation systems, quantum-inspired materials simulation — without waiting for fault-tolerant hardware that may still be years away.

Spun out of Alphabet in 2022 and backed by Eric Schmidt’s personal investment, SandboxAQ has raised over $500 million at a valuation of approximately $5.75 billion. Its three commercial pillars are distinct and revenue-generating. Post-quantum cryptography is the largest near-term market the quantum sector has created: NIST standardised its first PQC algorithms in 2024, and enterprises now face a multi-year migration of cryptographic infrastructure — a migration that SandboxAQ’s Security Suite and PQC audit tools help manage. Quantum sensing is the company’s second pillar: a US Air Force SBIR contract will develop a quantum navigation prototype tested live aboard Air Force aircraft, addressing the urgent vulnerability of GPS-dependent military systems in contested environments. The third pillar — quantum-inspired and AI-augmented materials simulation using large quantitative models (LQMs) — accelerates drug discovery and materials design at speeds that even today’s limited quantum hardware cannot yet match.
SandboxAQ generates real revenue from paying enterprise and government customers now — something only IonQ and D-Wave can claim among the companies in this guide. Its commercial partnership with evolutionQ co-distributes quantum network security tools across critical infrastructure clients. If a fund manager needed to explain a quantum investment to a sceptical board, SandboxAQ would be the easiest conversation in the room: it sells products that solve documented security problems, today, regardless of when a fault-tolerant quantum computer arrives.

QC Ware solves a problem that hardware companies often ignore: enterprise quantum users do not want to write quantum circuits — they want to run quantum applications. Its Forge platform translates optimisation, financial simulation and chemistry problems into circuits that run on whatever hardware the customer has access to today.

Founded in 2014 by Matt Johnson and backed by Goldman Sachs, Airbus Ventures and BMW i Ventures, QC Ware is one of the most institutionally credible quantum software companies in existence. The investor consortium is not accidental: Goldman uses Forge for portfolio optimisation and Monte Carlo simulation on quantum hardware; Airbus explores computational fluid dynamics for aircraft design; BMW targets battery materials simulation. These are production-grade enterprise engagements with strategic investors whose own commercial interests are served by QC Ware’s success. Forge runs across IBM, IonQ, Quantinuum, Rigetti and D-Wave, giving customers hardware optionality as the competitive landscape evolves and protecting the investment as individual hardware companies rise and fall. QC Ware’s algorithms achieve dramatic circuit depth reductions, enabling useful computations on today’s NISQ devices while remaining forward-compatible with fault-tolerant architectures.
In an ecosystem where most companies chase hardware milestones and qubit count records, QC Ware chases application milestones: the first demonstration that a quantum algorithm delivers a commercially meaningful result for Goldman’s traders, for Airbus’s engineers, for BMW’s materials scientists. That discipline — orienting technical development around end-user value rather than physics benchmarks — is increasingly recognised as the missing capability that will separate quantum companies that generate sustained enterprise revenue from those that perpetually promise it.