Quantum computing.
Pick your physics.

Quantum advantage claims don't last long. In July, IBM said its quantum computer had done something no classical computer could. A month later, researchers in Singapore did the same thing on Nvidia GPUs in 37 minutes.

That's quantum computing going into 2027: real machines, real progress, and a finish line that keeps moving.

The finish line moves because qubits are so hard to tame. A quantum computer runs on qubits, which can hold combinations of 0 and 1 at once and interact in ways ordinary bits can't. For a narrow set of problems, like simulating molecules, some kinds of optimization, and breaking today's encryption, that could mean answers no supercomputer can reach. But qubits are fragile. Heat, vibration, and stray light knock them over, so every useful machine needs error correction, where many physical qubits work together as one reliable "logical" qubit. That's the race everyone is actually running.

Washington is paying for it anyway. In September, it finalized $1.8 billion of chip-research awards across the industry and took stock in at least three of the winners. New listings keep arriving too: Quantinuum in June, IQM in July, Pasqal in August.

Quantum computing stocks are another story. With the 10-year Treasury above 5% for the first time since 2007, money-losing companies with 2029 roadmaps are the first thing investors sell, and Quantinuum already trades below its IPO price.

Every one of these quantum computing stocks is a bet on one kind of qubit, and nobody knows which kind wins. Here they are by physics:

• Trapped ions: charged atoms held by electric fields. The most accurate qubits, but slow and hard to scale.
• Superconducting: tiny chilled circuits. Fast and built like chips, but noisy. IBM and Google's pick.
• Neutral atoms: uncharged atoms held by lasers. They scale to big arrays fast, which made them the new favorite.
• Photons: particles of light. They work at room temperature and travel over fiber, but they're the hardest to build.
• Annealing: a special-purpose machine for optimization. It works today, on a narrow set of problems.
• Any qubit: the company that gets paid whichever physics wins.

The accuracy play. Every ion company's pitch is that fewer, better qubits beat more, noisier ones.

IonQ NYSE: IONQ

IonQ stopped being a quantum startup this year and became a chip company with a quantum division. It closed its $1.8 billion purchase of SkyWater, a US chip foundry, in July, and that's most of why revenue guidance jumped to $450–460 million.

The first IonQ processor built on SkyWater chips, Superion 256, ships in 2027, and Nvidia picked it as the first quantum system for its new quantum research center. You're buying the most revenue in quantum, with the caveat that most of it isn't quantum revenue.

Quantinuum NASDAQ: QNT

Quantinuum builds the machines other quantum companies benchmark against. Its Helios system hit nearly 99.999% fidelity on logical qubits this year, and that's the number that decides whether error correction actually works. Oracle signed a multi-year deal to have a Helios system installed inside its cloud.

It went public in June at $60 and now trades below that, at hundreds of times this year's revenue, with Honeywell still holding about 48% of the vote. You're getting the best physics in the business and paying today for a machine due in 2029.

The approach the giants chose. Superconducting qubits are made like chips, so the semiconductor playbook applies. They're also noisy, which is why everyone in this group talks more about error correction than qubit counts.

IBM NYSE: IBM

IBM has the clearest roadmap in quantum and keeps hitting it. It has more than 90 systems deployed, Cockatoo due in 2027, and Starling, a fault-tolerant machine with 200 logical qubits, scheduled for 2029. In September it finalized $1 billion from Commerce, matched with $1 billion of its own, to run Anderon, a standalone quantum chip foundry in Albany.

It also learned what quantum advantage costs. One of its three July advantage claims was reproduced on GPUs within weeks. IBM had posted the claims on a public tracker so people could try, and that's what a company does when it's playing the long game.

Alphabet NASDAQ: GOOGL

Google made the field's first "verifiable advantage" claim with its Willow chip in October 2025, then added neutral atoms as a second approach in March 2026. It's the only company on this list building two different kinds of qubits. You're buying a search and cloud giant with one of the two best quantum labs in the world attached at no extra charge.

Rigetti Computing NASDAQ: RGTI

Rigetti is the small superconducting company that won't run out of money. It had about $540 million in cash and no debt at midyear, plus a $100 million Commerce deal finalized in September. Its 108-qubit Cepheus-1 is built from twelve smaller chips tiled together, the same chiplet trick that's working in classical computing. Revenue is tiny, and fidelity has to improve before anyone pays up. But Rigetti can afford to keep trying, which most small quantum companies can't.

IQM NASDAQ: IQMX

Finland's IQM listed on Nasdaq in July, and its business is selling whole quantum computers to supercomputing centers, universities and national labs that want a machine on their own soil. It says it has sold 26 and delivered 17. That's a different model from renting time in the cloud, and right now it's where the biggest checks in quantum are being written. If national labs keep buying, IQM is the cleanest way to own that demand.

The newest favorite. Atoms held by lasers can be packed into big arrays fast, and every major lab has added a neutral-atom program in the last two years. The open question is whether they can run error correction as cleanly as ions.

Infleqtion NYSE: INFQ

Infleqtion is the neutral-atom company that already has a business. It sells quantum sensors like atomic clocks, radio receivers and navigation gear to defense and aerospace customers, and that revenue (about $45 million guided for 2026) pays for the computer.

In September it hit its target of 30 logical qubits on 80 atoms. Critics point out that the result detects errors rather than correcting them, which is a real gap. But Infleqtion is the one company in this group you could value on what it sells today.

Pasqal NASDAQ: PSQL

French neutral-atom maker Pasqal listed through a SPAC in August with about $360 million in cash. The stock nearly doubled on day one and now trades below its $10 SPAC price. It has seven processors installed and three more in production. It's the newest ticker in quantum and the cleanest test of whether the market will keep funding roadmaps with 10-year yields above 5%.

Atom Computing (Private)

Atom Computing is the neutral-atom company Microsoft picked as its partner on logical qubits, and Washington has a letter of intent to take a $100 million stake. In August it hired a CFO who had just taken another company through an NYSE IPO. Companies don't make that hire by accident.

Light-based qubits work at room temperature and travel over ordinary fiber, which makes them the natural way to network quantum computers together. They're also the hardest to build at scale.

You won't find Quantum Computing Inc. here. Most of its revenue comes from government subcontracts and conventional photonics, not quantum computing.

PsiQuantum (Private)

PsiQuantum is the biggest bet in quantum: a million-qubit photonic computer made in standard chip fabs. It was valued at $7 billion after a $1 billion raise in September 2025, and it has a finalized $100 million Commerce award plus AU$940 million from Australia. It's building utility-scale sites in Chicago and Queensland. In 2024 it promised an operational Queensland site by the end of 2027. Its latest update only says the cooling plant arrives then. Watch which date it hits.

Xanadu NASDAQ: XNDU

Xanadu is the only public photonic pure-play, with AMD as a partner and C$195 million of signed federal funding, part of up to C$390 million from Canada and Ontario, for a factory in Toronto. In September its insider lockup ended, and the stock fell more than 20% in a day to under $6.

Photon loss is the whole ballgame for this approach, and Xanadu keeps cutting it. It's the highest-risk name on the list, and the one that wins biggest if light turns out to be the right qubit.

One company, one very different machine.

D-Wave Quantum NASDAQ: QBTS

D-Wave is the only company here whose customers run quantum in production today. Its annealers don't do general-purpose quantum computing. They solve optimization problems like scheduling and routing, and they do it now: AT&T cut one network job from an hour to under 15 seconds.

First-half bookings jumped to $35.5 million from $2.9 million a year earlier. Purists dismiss annealing, but customers pay for it. D-Wave also bought a gate-model company in January, in case the purists turn out to be right.

Nvidia NASDAQ: NVDA

Nvidia doesn't build qubits. It builds everything around them. NVQLink connects quantum processors directly to GPUs, CUDA-Q software runs on most of the quantum hardware you can rent, and in September Nvidia picked IonQ's first chip-based system to install in its own quantum research center next year.

Every serious roadmap above is really a hybrid machine, with a quantum processor doing the narrow part and GPUs doing the rest. Whichever qubit wins, it ships next to Nvidia.

The biggest buyers of quantum computers don't need them to beat a GPU. They're the Commerce Department, Australia and Queensland, Canada and Ontario, national labs, state universities, and supercomputing centers across Europe. They're buying because a rival might get there first.

That's the supercomputer playbook. Governments have bought the fastest machines for decades, for weather, weapons and prestige, without asking for a quarterly return. Quantum just joined that list. Revenue can keep growing for years before anyone proves advantage, and the advantage headlines that swing quantum computing stocks matter less to the business than they look.

That also changes what a winning quantum company looks like. It isn't a cloud startup renting time by the hour. It's a contractor that can deliver a whole machine to a government, install it on that government's soil and keep it running for a decade. That's closer to Cray than to Amazon Web Services.

So in 2027, judge quantum computing stocks the way customers judge the companies behind them: on whether they can deliver, install and support a machine, not on who won the latest advantage headline.