For two decades quantum computing existed in a superposition of its own: simultaneously the most transformative technology of the next century and a research curiosity that might never leave the laboratory. In 2025 and 2026, that superposition began to collapse. IonQ became the first quantum computing company to exceed $100 million in annual GAAP revenue, growing 202% year-on-year. Quantinuum filed confidentially for an IPO that could value it near $20 billion. Google's Willow chip demonstrated a 13,000x speedup over the world's fastest supercomputer. Governments on six continents have committed more than $40 billion to national quantum strategies. None of this means general-purpose quantum computers are here. It means the decade in which we find out has now genuinely begun.
Not investment advice. Data sourced from State of Quantum Computing 2026 (Entangled Future, April 2026), Quantum Computing Report news archive June 2026, Quantum Zeitgeist Companies Directory February 2026, IBM Quantum Innovation Roadmap, IonQ Q4 2025 earnings, Quantinuum SEC filings January 2026, company technical disclosures. All figures current as of June 2026.
Every transformative technology spends years as a promise before it becomes a product. Quantum computing's version of that wait has been unusually long and unusually theoretical — physicists have understood the mathematics of qubits, superposition, and entanglement since the 1980s, but turning that mathematics into a machine that reliably outperforms a classical computer at a problem anyone actually cares about has remained maddeningly elusive. Google's 2019 "quantum supremacy" claim was immediately disputed by IBM and debated by physicists for years afterward. Billions of dollars were raised against roadmaps that, as one industry retrospective put it, "compressed decades of physics into five-year plans."
What changed in 2025 and into 2026 is not that quantum computers became generally useful. It is that the specific, falsifiable, commercially relevant proof points the industry has been promising for twenty years finally started arriving on schedule, in public, with financial and scientific consequences attached. IonQ and Ansys ran a medical device simulation on a 36-qubit trapped-ion computer that outperformed classical high-performance computing by 12% — one of the first documented cases of genuine practical quantum advantage on a problem with commercial relevance, not a contrived benchmark. IonQ became the first quantum computing company in history to exceed $100 million in annual GAAP revenue, growing 202% year-on-year to $130 million — a milestone that directly answers the most persistent criticism of the sector: that quantum companies generate no real revenue. Quantinuum filed confidentially for an IPO that could value the company near $20 billion. This is the year the industry stopped being purely a story about the future and started generating evidence about the present.
Classical computers store information as bits — ones and zeros, definite states. Quantum computers use qubits, which can exist in superposition (a combination of zero and one simultaneously) and can be entangled with other qubits such that measuring one instantly affects the state of another, regardless of distance. These two properties — superposition and entanglement — are what theoretically allow a quantum computer to explore an exponentially large solution space simultaneously, rather than checking possibilities one at a time the way a classical computer must. For certain classes of problems — simulating molecules, factoring large numbers, optimising across enormous combinatorial spaces — this is not a modest speed advantage. It is, in principle, the difference between a calculation taking seconds and one taking longer than the age of the universe.
The catch, and the entire reason quantum computing has taken so much longer to commercialise than classical computing did, is that qubits are extraordinarily fragile. Any interaction with the surrounding environment — heat, electromagnetic interference, even the act of measurement itself — causes decoherence, destroying the delicate quantum state before a useful calculation can complete. Today's machines are best described as NISQ (Noisy Intermediate-Scale Quantum) devices: tens to a few thousand physical qubits, with error rates that accumulate faster than most useful algorithms can run to completion. The industry's entire technical roadmap, across every hardware approach, points toward a single goal: fault-tolerant quantum computing, in which many noisy physical qubits are combined through quantum error correction into a smaller number of highly reliable "logical qubits" capable of sustaining long, accurate computations.
The fidelity leaderboard reveals a pattern that runs against the industry's popular narrative. Superconducting qubits — IBM, Google, Rigetti — are the most heavily funded and most commonly implemented approach, owing to their compatibility with existing semiconductor fabrication. But the logical-qubit leaderboard, where fidelity actually translates into useful computation, is led by trapped-ion and neutral-atom systems — IonQ and Quantinuum chief among them — precisely because their underlying physical qubits start with higher fidelity, requiring less error-correction overhead to reach a usable logical qubit. Fewer, higher-quality qubits are currently outperforming larger numbers of noisier ones — a finding with direct implications for which hardware bets are most likely to reach commercial relevance first.
Synthesised from IBM, IonQ, and Quantinuum public roadmaps · June 2026
The single most important fact about this roadmap, and the one most often missing from popular coverage: the industry has, so far, largely hit the milestones it has publicly committed to. IBM's own characterisation is notable for its specificity — the company states it has "successfully delivered on each of our milestones" across the multi-year Innovation Roadmap, and that track record is what underlies its confidence in the 2029 fault-tolerance target. This does not guarantee the remaining, harder milestones will also land on schedule — engineering difficulty in deep technology has a well-documented tendency to compound near the finish line, not ease — but it is a materially different track record than the sector's earlier decades of roadmaps that simply slipped year after year.
The most mature, most heavily funded approach — leveraging decades of semiconductor fabrication expertise. Requires extreme cryogenic cooling (near absolute zero). IBM's 1,121-qubit Condor is the largest superconducting chip built to date; Google's Willow demonstrated the clearest evidence yet that error rates can fall as qubit count scales, the central requirement for fault tolerance to work at all.
The fidelity leaders. Individual atoms trapped and manipulated with lasers — naturally identical, "perfect" qubits rather than fabricated approximations. Slower gate operation speeds than superconducting systems, but dramatically higher fidelity, meaning fewer physical qubits are wasted on error correction overhead per useful logical qubit produced.
Uses photons (particles of light) rather than matter-based qubits. PsiQuantum's distinctive bet: manufacture quantum chips using existing semiconductor fabrication infrastructure — potentially the most scalable manufacturing path of any approach, if the underlying photonic error correction physics works at scale. Widely regarded as the most ambitious, highest-variance bet in the sector. Xanadu went public via SPAC in March 2026 at a $3.1 billion valuation.
The fastest-improving approach on logical qubit metrics. Arrays of individually trapped neutral atoms, offering a highly scalable architecture with strong European government backing (Pasqal). Infleqtion has publicly targeted 10 true logical qubits — using error correction that actually corrects, not merely detects — by the end of 2026, among the most concrete near-term commitments in the sector.
The highest-risk, potentially highest-reward bet. Microsoft's Majorana-based approach seeks qubits that are inherently, physically resistant to the errors that plague every other architecture — if the underlying Majorana quasiparticle physics holds up, which remains genuinely debated in the physics community. Majorana 1, announced February 2025, remains a prototype; it has not yet demonstrated error-corrected computation. DARPA has selected Microsoft for utility-scale development regardless.
A fundamentally different approach, purpose-built for optimisation problems rather than universal computation. D-Wave already has commercial customers using its systems in production — the most mature commercial deployment of any quantum hardware approach, precisely because it targets a narrower, more tractable problem class than the universal gate-based machines every other company is racing to build.
The most aggressive consolidator in the sector, and the first to prove genuine commercial revenue at scale. Beyond its core trapped-ion hardware, IonQ has spent roughly $2.5 billion acquiring Oxford Ionics ($1.075B), ID Quantique ($250M), Capella Space ($318M), Qubitekk, Lightsynq, Vector Atomic, and Skyloom Global — transforming itself from a pure hardware company into a full-stack platform spanning computing, networking, sensing, and space. A pending $1.8 billion acquisition of SkyWater Technology, a US semiconductor foundry, would bring fabrication capability in-house, pending regulatory approval. CEO Niccolo de Masi testified before Congress in November 2025. The only quantum company to make Deloitte's 2025 Fast 500, with roughly 2,000% revenue growth from 2021 to 2024.
The world's largest integrated quantum company by most measures, and the most significant quantum IPO ever attempted if it completes. Formed from the 2021 merger of Cambridge Quantum and Honeywell Quantum Solutions, Quantinuum's Helios system — 98 trapped-ion qubits, independently validated by Sandia National Laboratories in June 2026 — currently leads the sector's gate fidelity benchmarks. The company's June 2026 announcement of an accelerated roadmap to universal fault-tolerant computing by 2030 places it on the most aggressive public timeline in the industry, ahead even of IBM's 2029 target for a narrower fault-tolerant milestone.
Quantinuum's deepening partnership with Microsoft — including the first chemistry simulation run using reliable logical qubits combined with AI and classical HPC — represents the clearest evidence yet that the "hybrid classical-quantum" computing model the entire industry has converged on as the near-term commercial path is producing genuine scientific results, not just press releases.
The most enterprise-credible player in quantum computing, leveraging an existing global customer base (RIKEN, Boeing, Cleveland Clinic, Oak Ridge National Laboratory) and the broadest software ecosystem (Qiskit, the IBM Quantum Network) of any hardware vendor. IBM's explicit, public claim is unusual in its specificity for the sector: the company states it will be "the only quantum computing organisation in the world" capable of running hundreds of logical qubits and millions of quantum gates by decade's end — a claim grounded in a multi-year roadmap (Loon, Kookaburra, Starling, Blue Jay) the company says it has hit on schedule so far.
The most scientifically rigorous of the major players, with Willow's error-correction demonstration widely regarded as the most technically significant single milestone of the past two years — proof that error rates can fall as qubit count scales, the foundational requirement for fault tolerance to be achievable at all rather than a permanently receding target. Google has consistently prioritised scientific breakthrough publication over near-term commercialisation, a strategy that has produced the field's most credible peer-reviewed results but limited near-term revenue relative to IonQ's commercial push.
The highest-risk, highest-potential-reward architectural bet among the major technology companies, betting that topological qubits — if the underlying Majorana quasiparticle physics holds, which remains genuinely contested in the physics community — could deliver inherently more stable qubits than any competing approach, dramatically shortening the path to millions of logical qubits. Azure Quantum's strategy hedges this bet by also providing cloud access to IonQ, Quantinuum, Rigetti, and Atom Computing hardware — meaning Microsoft profits from the sector's progress regardless of whether its own topological bet succeeds.
The startup most analysts describe as the field's highest-conviction, highest-variance wager: building quantum computers from photons using existing semiconductor manufacturing infrastructure, a bet that — if it works — could offer the most scalable manufacturing path of any architecture in the sector, sidestepping the exotic fabrication and extreme cryogenic requirements that constrain every other approach.
The fastest-improving approach on logical-qubit metrics, with strong European (Pasqal) and US academic and government (QuEra, Infleqtion) backing. Pasqal's 140-qubit system is already operational at CINECA in Italy, with a 256+ qubit Vela system launching across 2026. Infleqtion's public target of 10 genuinely error-correcting logical qubits by end of 2026 — using a code that actually corrects rather than merely detects errors — is among the most concrete, falsifiable near-term commitments anywhere in the sector.
A full-stack, cloud-accessible superconducting platform that reached general availability for its 108-qubit Cepheus-1 multi-chip processor in early 2026, targeting 99.7% fidelity in upcoming updates. Smaller in scale and funding than IBM or Google, but a genuine independent public-market alternative for investors specifically seeking superconducting-architecture exposure without the diversification (and corresponding dilution of pure quantum upside) that comes with IBM or Alphabet shares.
The most commercially mature quantum hardware deployment of any company examined here, precisely because it deliberately targets a narrower problem class — optimisation — rather than chasing universal, general-purpose quantum computation. D-Wave's annealing approach already has customers running production workloads, a genuine commercial proof point that predates most of the sector's recent revenue milestones, even as its narrower technical scope means it will not capture the broadest long-term upside if universal fault-tolerant computing succeeds elsewhere.
IBM, Alphabet (Google), Microsoft, Amazon (via AWS Braket). Quantum computing represents a genuine long-term strategic bet for each, but a small fraction of overall revenue and enterprise value today — meaning quantum-specific setbacks or delays carry limited downside for investors, while genuine breakthroughs provide meaningful optionality. Amazon's approach is distinctive: rather than building only proprietary hardware, AWS Braket offers cloud access across multiple vendors' quantum hardware, positioning Amazon as infrastructure-layer exposure to the entire sector's progress rather than a bet on any single architecture.
IonQ, Rigetti Computing, D-Wave Systems. Direct, undiluted exposure to quantum commercialisation, with correspondingly higher volatility and a valuation entirely dependent on continued technical and commercial progress. IonQ's revenue proof point and aggressive acquisition strategy distinguish it from Rigetti's more focused hardware-only approach and D-Wave's narrower optimisation-specific commercial model.
Quantinuum (confidential S-1 filed), PsiQuantum, Pasqal, QuEra. The companies most closely watched by specialist investors, currently inaccessible to public market participants except through the Quantinuum IPO process now underway. Quantinuum's filing is the single event most likely to reset public valuation benchmarks for the entire sector — a successful, well-received offering would validate years of private capital deployment across every tier above.
"Quantum advantage" remains one of the most contested and frequently misused terms in the entire technology sector, and investors should treat every company's claim of it with the same scrutiny applied to Google's disputed 2019 supremacy announcement. IonQ and Ansys's 12% advantage on a medical device simulation is genuine and well-documented, but it is a narrow, specific result on one problem class — not evidence that quantum computers broadly outperform classical ones today. IBM's own framing — quantum advantage by end of 2026, meaning quantum as an HPC accelerator rather than a standalone replacement — is a more honest characterisation of where the technology actually stands than the more sweeping claims that periodically circulate in sector commentary. The gap between "quantum advantage on a specific, narrow problem" and "quantum computers that are broadly useful" remains the single most important distinction for any investor in this sector to hold onto.
Microsoft's topological approach is the starkest illustration of how much genuine scientific uncertainty remains embedded in even the most well-resourced quantum bets. Majorana 1, announced with considerable fanfare in February 2025, remains a prototype that has not, as of mid-2026, demonstrated error-corrected computation — and the underlying physics of Majorana quasiparticles remains a subject of genuine, unresolved debate within the physics community, not merely an engineering challenge awaiting sufficient capital. This is the clearest reminder in the entire sector that quantum computing investment carries genuine fundamental-science risk, not merely the execution risk familiar from conventional technology investing.
The most credible near-term commercial applications are narrower, less glamorous, and more immediately monetisable than the sector's popular narrative about drug discovery and climate modelling suggests. D-Wave's optimisation-focused annealing approach has the most mature commercial deployment in the sector precisely because it abandoned the goal of universal computation in favour of solving a narrower, more tractable problem class well. IonQ's most credible near-term revenue driver is increasingly its networking, sensing, and space-systems acquisitions — adjacent technology businesses with nearer-term commercial paths — rather than universal quantum computing revenue itself. The investors most likely to be disappointed by quantum computing over the next five years are those expecting the most ambitious long-term applications (molecular engineering, generalised drug discovery acceleration) to arrive on the same timeline as the narrower, more immediately commercial applications (optimisation, specific chemistry simulations, post-quantum cryptography services) that are already generating real revenue today.
Long-horizon thinking on capital, technology, and the forces shaping the next decade of wealth creation. Written from first principles. Not consensus. Not noise.