Quantum Computing: what are the top startups?
In our quantum computing market deck, you will find everything you need to understand the market
SUMMARY
PsiQuantum is the top quantum computing startup today, followed by QuEra and Atom Computing; Pasqal ranks fourth while it remains private.
The ranking has been reset by public listings. Quantinuum, Xanadu and IQM have moved out of the private-startup field just as funding has become much more concentrated among the remaining contenders.
That funding surge is unusually large. McKinsey estimates quantum-technology startup investment jumped from about $2 billion in 2024 to $12.6 billion in 2025, with roughly 90% going into quantum computing and around 60% of total quantum startup funding concentrated in the ten largest deals.
Physical-qubit counts are becoming a weaker shortcut for judging leadership. The more useful comparison is whether an architecture can turn physical hardware into reliable logical qubits without creating impossible manufacturing, control or error-correction overhead.
PsiQuantum leads because its utility-scale photonic architecture has reached DARPA Stage C and is backed by more than $2 billion of capital. The big gap is still obvious: it cannot yet show a customer operating the kind of large programmable machine it ultimately wants to build.
QuEra has the strongest public fault-tolerance case among private hardware companies. Its work on logical operations, magic-state distillation and long-running large atom arrays gives it more technical depth than a simple neutral-atom qubit count would suggest.
Atom Computing may be the company most likely to challenge that order soon. Microsoft has already demonstrated 24 entangled logical qubits on Atom hardware, and the planned Magne deployment could turn that research into a customer-facing logical system.
Pasqal stands out for a different reason: commercial execution. Seven installed QPUs, three more in production and €16.5 million of 2025 commercial revenue make it the clearest example of a private quantum hardware company repeatedly shipping complete systems, although its planned public listing may soon remove it from the ranking.
The architecture race is still wide open, but neutral atoms have the deepest private bench. QuEra, Atom Computing and Pasqal have each produced credible results in a different part of the stack: fault tolerance, hardware scale and real deployment.
Alice & Bob and Quantum Motion are more asymmetric bets. Alice & Bob is trying to shrink error-correction overhead with cat qubits, while Quantum Motion is betting that standard CMOS manufacturing will become decisive once quantum computers need hundreds of thousands or millions of physical qubits.
Software companies belong in the ranking, but for different reasons. Riverlane is the strongest quantum-error-correction infrastructure bet, Q-CTRL has the strongest near-term commercial software position, and Classiq is building the higher-level development layer that becomes much more valuable if useful quantum hardware arrives.
No private quantum hardware company has conventional product-market fit yet. The market is still being driven heavily by governments, national computing centers, laboratories and strategic partners, so “early commercial validation” is a better description than broad enterprise adoption.
This market map, featured in our quantum computing market deck, highlights top companies and startups in the quantum computing market
Why is the quantum computing startup ranking changing so quickly?
The quantum computing startup ranking looks very different today because several former leaders have gone public while private funding has suddenly concentrated around a much smaller group of companies.
Quantinuum completed a $1.68 billion IPO and now trades on Nasdaq. Xanadu became publicly listed on Nasdaq and the Toronto Stock Exchange. IQM also became public, with shares trading in the US and Helsinki. Those three companies would have belonged near the top of almost any quantum startup ranking a year ago. They no longer belong in one.
At the same time, investors have become much more aggressive. McKinsey's 2026 Quantum Technology Monitor found that investment in quantum technology startups jumped from about $2 billion in 2024 to $12.6 billion in 2025. Roughly 90% went into quantum computing, which puts the amount directed toward quantum computing startups at around $11.3 billion.
The concentration is even more striking. McKinsey found that roughly 60% of all quantum startup investment went into the ten largest deals. That works out to around $7.6 billion flowing into just ten transactions.
We are therefore looking at a market where funding increased more than sixfold in a year while much of it piled into a small number of companies. Investors are increasingly choosing which architectures they think can make the jump from promising physics to enormous, fault-tolerant machines.
What should "top quantum computing startup" actually mean today?
The top quantum computing startup today should be the company with the strongest credible path to a useful fault-tolerant computer, rather than whichever company can advertise the most physical qubits.
Physical-qubit comparisons get messy very quickly. PsiQuantum uses photons. QuEra, Atom Computing and Pasqal use neutral atoms. Alice & Bob uses superconducting cat qubits. Quantum Motion uses silicon spins. Each architecture has different physical error rates, connectivity, operating speeds and error-correction requirements.
A machine with 1,000 physical qubits can therefore be much further from useful computation than a machine with far fewer qubits. What we ultimately need are logical qubits: error-corrected units that can survive enough operations to run difficult algorithms reliably.
Manufacturing also matters. A processor that works beautifully with 20 qubits may become impossible to wire, cool, calibrate or control with one million. Commercial deployment gives us another useful test because it shows whether complete systems can actually leave the lab.
So our ranking weighs demonstrated progress toward logical computation, ability to scale the architecture, manufacturing credibility, outside technical validation and evidence that complete systems can be deployed.
| What we look at | What it tells us | What can fool us |
|---|---|---|
| Logical-qubit and error-correction results | Whether more hardware is producing more reliable computation | Huge physical-qubit counts |
| Architecture at very large scale | Whether the machine could plausibly reach useful size | Excellent small prototypes |
| Manufacturing | Whether thousands or millions of components can realistically be produced | Lab processes that are difficult to industrialize |
| Outside validation | Whether technical claims survive scrutiny from DARPA, customers or major partners | Company roadmaps |
| Commercial deployment | Whether a complete machine can be installed and operated elsewhere | Partnerships with no operating hardware |
As this chart shows, and as featured in our quantum computing market deck, search interest in quantum computing has grown significantly
Is PsiQuantum really the quantum startup to beat?
PsiQuantum is currently the quantum computing startup to beat because it has the strongest combination of utility-scale ambition, independent technical scrutiny and financial firepower.
The strongest argument for PsiQuantum comes from DARPA rather than its investors. DARPA's Quantum Benchmarking Initiative is examining whether proposed quantum computers can eventually create more computational value than they cost to operate. PsiQuantum and Microsoft are the only two organizations that have reached Stage C, the final phase.
DARPA has been given extensive access to PsiQuantum's technical designs, facilities, manufacturing processes and prototypes. In July 2026, the relationship expanded into a new performance-based agreement worth $125 million.
The financial side is similarly unusual. PsiQuantum raised a $1 billion Series E at a $7 billion valuation and has raised more than $2 billion overall. That capital is financing manufacturing infrastructure, large prototype systems and utility-scale sites.
PsiQuantum's photonic approach also gives it a distinctive manufacturing strategy. The company already fabricates quantum photonic chips in commercial semiconductor foundries and wants to assemble very large numbers of these components into modular systems.
The weakness is clear: PsiQuantum cannot yet point to a customer operating a large programmable machine. QuEra and Atom Computing can show more near-term computation on functioning systems, while Pasqal can show more commercial deployments.
We still put PsiQuantum first because DARPA's Stage C process is the strongest outside evidence we have that its utility-scale architecture deserves to be taken seriously.
If you want more recent data on this point, please see our latest quantum computing market report.
Is QuEra actually ahead on fault-tolerant quantum computing?
QuEra has probably produced the strongest public fault-tolerance evidence of any private quantum hardware startup today.
Research involving QuEra, Harvard and MIT demonstrated logical-level magic-state distillation in 2025. Magic states are needed for universal fault-tolerant quantum computing, so this goes much deeper than simply keeping a logical qubit alive. The same research ecosystem has also shown scalable error correction and fault-tolerant logical operations.
Another useful experiment involved maintaining a roughly 3,000-atom array continuously for more than two hours. Neutral-atom machines naturally lose individual atoms over time. QuEra's architecture can detect those losses and refill the array while the system keeps running.
The company has also raised more than $230 million from investors that include Google, SoftBank and NVIDIA's venture arm. Google deserves particular attention because it already runs one of the world's strongest internal quantum programs and has since added neutral atoms as another hardware research path.
DARPA advanced QuEra into Stage B of the Quantum Benchmarking Initiative, and QuEra has installed a Gemini-class neutral-atom system at Japan's AIST alongside the ABCI-Q supercomputing environment.
PsiQuantum remains our overall number one because its complete utility-scale architecture has advanced further through DARPA's evaluation. If we ranked private companies only on public fault-tolerance results, QuEra would have a serious case for first place.
This chart, included in our quantum computing market deck, illustrates yearly VC funding for quantum computing startups
Is Atom Computing closer to a usable logical quantum computer?
Atom Computing currently has the clearest private-company path to putting a substantial number of logical qubits in front of real users.
Atom's AC1000 system contains more than 1,200 neutral-atom physical qubits and is available as an on-premises machine. The architecture includes mid-circuit measurement, qubit reset and reuse, conditional branching and high connectivity.
Microsoft then added the piece that makes Atom particularly interesting. Using Atom's hardware with Microsoft's qubit-virtualization software, the companies created 24 entangled logical qubits and performed error detection and correction on 28 logical qubits.
They are now turning that research into a product called Magne. Microsoft says Magne will combine more than 1,200 Atom Computing physical qubits with Microsoft's error-correction software stack and is expected to come online in Copenhagen in early 2027 through QuNorth.
DARPA has also moved Atom Computing into Stage B of its Quantum Benchmarking Initiative, while the company says it has raised more than $300 million.
Atom still sits just behind QuEra because QuEra has accumulated a broader set of public fault-tolerance demonstrations. A successful Magne deployment would make that gap much harder to defend.
If you want more recent data on this point, please see our latest quantum computing market report.
Does Pasqal still count as a quantum computing startup?
Pasqal still counts as a quantum computing startup for now, but it is right on the edge of leaving the category.
Pasqal agreed to a business combination that values the company at $2 billion before the transaction. The SEC declared the registration statement effective and Bleichroeder shareholders were scheduled to vote on the deal on August 25.
We checked for the freshest status before updating this ranking. The latest Pasqal and SEC material we can verify still describes the business combination as pending, and we cannot yet verify a completed transaction or the start of trading in Pasqal shares.
The underlying business deserves a high rank regardless of the transaction. Pasqal's SEC materials show €16.5 million of commercial revenue in 2025, up from €3.5 million in 2024. That is roughly 4.7 times more revenue in one year.
Pasqal also reports seven QPUs already installed, three more in production and more than €66 million in booked and awarded business including grants.
Its valuation is still based overwhelmingly on the future. €16.5 million of commercial revenue remains tiny next to a $2 billion proposed equity value, but Pasqal has already crossed an industrial threshold that many quantum companies are still approaching.
If the listing closes, we will simply remove Pasqal from this startup ranking.
This chart, included in our quantum computing market deck, looks at IonQ’s strategy in quantum computing
Could Alice & Bob's cat qubits beat bigger quantum machines with fewer qubits?
Alice & Bob could become a major quantum computing competitor if cat qubits keep reducing the enormous hardware overhead normally required for error correction.
Alice & Bob's cat-qubit architecture is engineered so that bit-flip errors are heavily suppressed at the physical level. Phase errors still have to be handled, so the technology does not remove quantum error correction, but it could make the job much smaller.
That changes how we should read the company's modest qubit count. Alice & Bob unveiled Helium in June 2026 with 18 cat qubits. Helium has been designed to encode the company's first logical qubit, while the next planned 48-cat-qubit processor is supposed to support several logical qubits.
France's GENCI then agreed to buy an 18-cat-qubit Helium system for installation alongside the Joliot-Curie supercomputer. The machine is expected to become accessible to users in 2027.
The company has also raised a €100 million Series B, later expanded with participation from NVIDIA's NVentures.
We rank Alice & Bob below QuEra and Atom because cat qubits have not yet been demonstrated at comparable system scale. The upside is that Alice & Bob may need far fewer physical qubits to reach the same logical result if its error-suppression advantage holds.
If you want more recent data on this point, please see our latest quantum computing market report.
Can Quantum Motion really scale quantum computing with normal chip factories?
Quantum Motion has one of the strongest manufacturing ideas in the quantum startup market, but we still need much more evidence that its silicon qubits can become a large logical computer.
Quantum Motion builds spin qubits using silicon and standard CMOS manufacturing techniques. The attraction is obvious. The semiconductor industry already knows how to make extraordinarily complicated silicon chips at huge scale with repeatable manufacturing processes.
Quantum computers may eventually need hundreds of thousands or millions of physical qubits. At that point, compatibility with existing semiconductor foundries could become a massive advantage.
Quantum Motion delivered a full-stack silicon CMOS quantum computer to the UK's National Quantum Computing Centre in 2025. DARPA then selected the company for Stage B of its Quantum Benchmarking Initiative. In May 2026, Quantum Motion raised another $160 million in a Series C led by DCVC and Kembara.
The company has deepened its manufacturing relationship with GlobalFoundries and is explicitly designing systems that could fit inside ordinary data-center racks.
Some of Quantum Motion's most spectacular claims, including huge reductions in space, cost and energy use, still come from the company itself. We would not use those projections to rank it above companies already showing more advanced logical operations.
What keeps Quantum Motion high on the list is the size of the eventual manufacturing problem. Its silicon CMOS approach could become very powerful if logical performance catches up with the fabrication story.
This chart, included in our quantum computing market deck, illustrates yearly funding for quantum computing startups
Are neutral atoms actually winning the quantum hardware race?
Neutral atoms currently give private quantum startups the deepest group of credible hardware contenders.
QuEra, Atom Computing and Pasqal all use neutral atoms, yet each has reached a different kind of milestone. QuEra has unusually strong public fault-tolerance research. Atom has more than 1,200 physical qubits in a commercial system and is building a logical machine with Microsoft. Pasqal has seven installed QPUs and €16.5 million of annual commercial revenue.
Seeing the same architecture produce strong results in science, hardware scale and deployment across three independent companies is more convincing than one spectacular experiment.
Google's behavior adds another useful clue. Google has spent years developing superconducting quantum processors, but its research organization has now added neutral atoms as another hardware path. Google has also invested directly in QuEra.
Neutral atoms have some natural advantages. Individual atoms are identical, large arrays can be created with optical traps, atoms can be moved around, and connectivity does not depend on physically wiring every qubit to every neighbor.
They also have their own problems. Lasers and optical control systems become complicated at scale, atoms can disappear from traps, gates can be slower than in some competing architectures, and a complete fault-tolerant system still has to coordinate huge numbers of operations reliably.
Among companies that still qualify as startups, neutral atoms currently have the strongest bench.
| Architecture | Leading private companies | Where the evidence is strongest | Biggest open question |
|---|---|---|---|
| Neutral atoms | QuEra, Atom Computing, Pasqal | Large arrays, logical-qubit progress, deployed machines | Can deep fault-tolerant circuits run fast and reliably enough? |
| Photonics | PsiQuantum | Manufacturing strategy and DARPA Stage C | Can the enormous complete system work as designed? |
| Cat qubits | Alice & Bob | Lower error-correction overhead | Does the advantage survive at much larger scale? |
| Silicon spins | Quantum Motion | Semiconductor manufacturing compatibility | Can logical performance catch up with the manufacturing story? |
If you want more recent data on this point, please see our latest quantum computing market report.
Are quantum software startups worth ranking with quantum hardware companies?
Quantum software startups deserve a place in the ranking, but Riverlane, Q-CTRL and Classiq are attractive for completely different reasons from PsiQuantum or QuEra.
Riverlane is our most important private quantum error-correction software company. Its Deltaflow 2 stack is designed to process quantum error data in real time and currently supports experiments involving as many as 250 physical qubits and up to 10,000 error-free quantum operations under its specified test conditions.
The strategic appeal is straightforward. Every serious fault-tolerant quantum computer needs error correction, and Riverlane has worked across several hardware modalities rather than tying itself to one type of qubit.
Q-CTRL has the strongest commercial profile of the three. Its software improves calibration, control and error suppression on imperfect quantum machines, allowing the company to sell something useful before fully fault-tolerant computers arrive. Q-CTRL currently reports $133 million raised, around 220 employees and a business spanning both quantum computing and quantum sensing.
Classiq sits higher in the programming stack. Its software lets developers describe quantum algorithms at a higher level and automatically generates optimized circuits for different hardware systems. The company has now raised more than $200 million, with AMD Ventures, Qualcomm Ventures and IonQ joining its latest strategic financing.
The timing risk is bigger for Classiq because a broad quantum developer ecosystem only becomes valuable when enough organizations have useful quantum workloads to develop.
That leaves us with Riverlane as the strongest infrastructure bet, Q-CTRL as the strongest commercial software company and Classiq as the strongest high-level development platform.
This chart, included in our quantum computing market deck, compares the main business model options for quantum computing hardware startups
Which quantum computing startups are actually selling real products, and do any have product-market fit?
Pasqal currently has the strongest disclosed hardware commercialization among quantum computing startups, but no private quantum hardware company has conventional product-market fit yet.
Pasqal's audited numbers make the comparison unusually concrete. Commercial revenue rose from €3.5 million in 2024 to €16.5 million in 2025. Seven QPUs are already installed and three more are in production.
The company also reports more than €66 million of booked and awarded business including grants and more than 25 quantum-solutions contracts inside that total. We should not confuse booked business with recognized revenue, especially because grants are included, but the figure shows a pipeline substantially larger than current annual sales.
Atom Computing has less financial transparency but an increasingly serious product. Its AC1000 can be ordered for on-premises deployment, while the Microsoft partnership is turning Atom's neutral-atom hardware into Magne for QuNorth.
QuEra has installed a system at AIST in Japan. Alice & Bob has also joined the deployment group after GENCI agreed to purchase its Helium system.
Even so, today's buyers still look very different from those in a mature technology market. National computing centers, governments, universities, laboratories and strategic industrial partners account for much of the demand. They are often buying research capacity, technological sovereignty and access to future capability alongside whatever computing value the machine provides today.
That makes "early commercial validation" a better description than product-market fit. Pasqal has gone furthest in showing that a private quantum hardware company can repeatedly manufacture, sell and support complete systems, but broad enterprise demand still depends on quantum computers becoming useful enough to justify normal computing budgets.
So which quantum computing startups are actually on top today?
PsiQuantum is the top quantum computing startup today, with QuEra and Atom Computing close behind; Pasqal ranks fourth while it remains private, followed by Alice & Bob and Quantum Motion.
PsiQuantum takes first place because no other private company combines DARPA Stage C scrutiny, more than $2 billion of capital, large-scale semiconductor manufacturing and a plan explicitly designed around utility-scale fault-tolerant computing.
QuEra earns second place through the depth of its fault-tolerance research. Logical operations, magic-state distillation, continuous large-array operation, DARPA Stage B and an installed system at AIST make QuEra much harder to dismiss as another promising neutral-atom research company.
Atom Computing comes third and could move higher soon. It already has more than 1,200 physical qubits in its commercial AC1000 platform, Microsoft has demonstrated 24 entangled logical qubits on Atom hardware, and Magne is supposed to turn that work into an operating logical machine for QuNorth.
Pasqal sits fourth because commercial execution matters. €16.5 million of annual commercial revenue, seven installed QPUs and three more in production give Pasqal a level of deployment that most private competitors cannot match. Its planned listing may soon remove the company from this list entirely.
Alice & Bob comes fifth. An 18-qubit machine sounds small until we account for what cat qubits are trying to achieve: far lower error-correction overhead. France's purchase of a Helium system gives the architecture a real external deployment to prove itself on.
Quantum Motion is sixth because industrial manufacturing could become decisive later in the race. Its silicon CMOS approach has reached a full-stack deployment, DARPA Stage B and a $160 million Series C, but logical-computing evidence still trails the companies above it.
Riverlane, Q-CTRL and Classiq complete our top nine. They are unlikely to build the winning quantum processor themselves, but each is trying to own part of the software layer that may sit above several different processors.
A second group deserves close attention: Diraq, Photonic, Nord Quantique and Silicon Quantum Computing. Their inclusion in DARPA Stage B is enough for us to treat them as serious challengers rather than fringe projects.
The ranking is fairly clear as of now even though the eventual technology winner remains wide open. PsiQuantum has the strongest overall private position. QuEra currently has the best case on advanced fault-tolerance research. Atom Computing is closest to turning a large neutral-atom system into a customer-facing logical machine. Pasqal has the clearest disclosed commercial hardware traction.
| Rank | Startup | Why it ranks here today | What could change the ranking |
|---|---|---|---|
| 1 | PsiQuantum | DARPA Stage C, more than $2B raised, semiconductor manufacturing and a utility-scale architecture | The complete photonic system still has to work at the scale promised |
| 2 | QuEra | Strong fault-tolerance results, magic-state distillation, DARPA Stage B and real HPC deployment | Needs to turn research leadership into larger user-facing logical systems |
| 3 | Atom Computing | 1,200+ physical qubits, Microsoft logical-qubit work and Magne commercialization | A successful Magne deployment could move Atom above QuEra |
| 4 | Pasqal | €16.5M commercial revenue, seven installed QPUs and three more in production | A completed public listing would remove Pasqal from the startup list |
| 5 | Alice & Bob | Cat-qubit architecture, complete 18-qubit system and first GENCI order | Must prove its lower-overhead approach works across larger logical systems |
| 6 | Quantum Motion | Silicon CMOS manufacturing, DARPA Stage B and $160M Series C | Needs stronger logical-computing results |
| 7 | Riverlane | Hardware-agnostic real-time quantum error-correction infrastructure | Hardware companies may keep more QEC technology in-house |
| 8 | Q-CTRL | Strong commercial software position and useful control technology on today's machines | Part of its business sits in sensing rather than quantum computing |
| 9 | Classiq | More than $200M raised and a strong hardware-agnostic development platform | Large software demand depends on useful quantum hardware arriving |
If you want more recent data on this point, please see our latest quantum computing market report.
This chart, featured in our quantum computing market deck, illustrates how revenue is divided among customer segments in the quantum computing market
OUR METHODOLOGY
This ranking looks for the private quantum computing companies with the strongest credible paths toward useful, scalable and eventually fault-tolerant quantum computing. We do not use one headline metric because physical-qubit counts, funding, scientific results, manufacturing progress, external scrutiny and commercial deployments can point in very different directions.
We prioritized the freshest meaningful evidence available: demonstrated logical-qubit and error-correction results, working systems, manufacturing progress, customer deployments, disclosed commercial figures and financing where access to capital materially affects the ability to execute. Funding strengthens an execution path, but it does not substitute for technical progress.
DARPA's Quantum Benchmarking Initiative is one of the most useful cross-architecture reference points because it evaluates fundamentally different approaches against the same utility-scale objective while still judging each technical path on its own merits. We therefore gave meaningful weight to Stage B and Stage C status, alongside peer-reviewed research and actual system deployments.
We compared architectures as complete scaling paths rather than forcing photonics, neutral atoms, cat qubits and silicon spins through one artificial qubit-count benchmark. Manufacturing, control complexity, error-correction overhead and the ability to operate complete systems all affect whether impressive small-scale hardware can become a useful large machine.
The final order is an editorial synthesis rather than a mechanical score. We looked for convergence across technical progress, scalability, manufacturing, outside validation and deployment, and we rechecked fast-moving items such as financing, DARPA status, installations and public-listing status close to publication.
Key sources include McKinsey's 2026 Quantum Technology Monitor, DARPA's Quantum Benchmarking Initiative, DARPA's Stage B selection material, DARPA's account of the final-stage evaluation of PsiQuantum and Microsoft, PsiQuantum's DARPA agreement, PsiQuantum's Series E announcement, the Nature paper on logical magic-state distillation, Atom Computing's AC1000 materials, Microsoft's Magne and logical-qubit work, Pasqal's SEC-filed operating figures, GENCI's Helium acquisition announcement, Quantum Motion's Series C announcement, Riverlane's Deltaflow 2 specifications, Q-CTRL's press kit, and Classiq's strategic financing announcement.
This chart, included in our quantum computing market deck, shows how cloud quantum computing access technology has evolved over time
