Many WorldsSubscribe

Companies

Quantinuum: the trapped-ion quantum computer maker Honeywell helped build

Quantinuum listed on Nasdaq in June 2026. It had $30.9 million of revenue in 2025, 60% of it from one customer, and $1.94 billion in cash on September 30.

Main takeaways

  • Quantinuum (QNT) builds quantum computers from trapped ions and sells access to them, over the cloud and at customers’ sites.
  • It was valued at about $11.4 billion at the November 10, 2026 close, counting both share classes, against 2025 revenue of $30.9 million.
  • The U.S. government and two other customers paid 71% of revenue in the first nine months of 2026.
  • Main risk: its large error-corrected machines are still targets, and operations and equipment used $294.6 million of cash in nine months.
  • Next dates: Sol, a system aiming at 100 logical qubits, in the second half of 2027, and Apollo in 2029.

A trapped ion is an electrically charged atom held in place by electromagnetic fields, and Quantinuum (QNT) uses such ions as the qubits of its computers. Most of its revenue so far comes from research institutions and governments: RIKEN, a Japanese research institute, was 60% of 2025 revenue, according to the IPO prospectus. At the November 10, 2026 Nasdaq close of $43.00 (price from Yahoo Finance), the company was valued at about $11.4 billion. That value counts both share classes in its latest quarterly filing: the Class A shares that trade and the Class B shares held by owners from before the listing.

Key numbers

Metric Value As of
Market value About $11.4 billion, at $43.00 a share November 10, 2026 close
Revenue $30.9 million Full year 2025
Net loss $192.6 million Full year 2025
Cash and cash equivalents $1.94 billion September 30, 2026
Largest system Helios, 98 physical qubits Launched November 5, 2025
Employees About 800 November 5, 2026
Founded 2021
Headquarters Broomfield, Colorado
Revenue by quarter$ million
  • 2025
  • 2026
2025, First quarter: 19.1 $ million2025, Second quarter: 2.1 $ million2025, Third quarter: 2.7 $ million2026, First quarter: 5.2 $ million2026, Second quarter: 8.0 $ million2026, Third quarter: 9.1 $ million19.12.12.75.28.09.1First quarterSecond quarterThird quarter2025, First quarter: 19.1 $ million2025, Second quarter: 2.1 $ million2025, Third quarter: 2.7 $ million2026, First quarter: 5.2 $ million2026, Second quarter: 8.0 $ million2026, Third quarter: 9.1 $ million19.12.12.75.28.09.1First quarterSecond quarterThird quarter

Source: Quantinuum, IPO prospectus (first quarter) and Form 10-Q filings for the second and third quarters of 2026

Loss from operations by quarter$ million
  • 2025
  • 2026
2025, First quarter: 29.9 $ million2025, Second quarter: 51.1 $ million2025, Third quarter: 52.9 $ million2026, First quarter: 77.2 $ million2026, Second quarter: 555.0 $ million2026, Third quarter: 132.8 $ million29.951.152.977.2555.0132.8First quarterSecond quarterThird quarter2025, First quarter: 29.9 $ million2025, Second quarter: 51.1 $ million2025, Third quarter: 52.9 $ million2026, First quarter: 77.2 $ million2026, Second quarter: 555.0 $ million2026, Third quarter: 132.8 $ million29.951.152.977.2555.0132.8First quarterSecond quarterThird quarter

Source: Quantinuum, IPO prospectus (first quarter) and Form 10-Q filings for the second and third quarters of 2026

What the company does

What is sold today is mostly access to its machines and research work done with customers. In the first nine months of 2026, $22.1 million of the $22.3 million of revenue came from what the quarterly filing calls cloud platform, research and support services. Hardware sales are rare and large: the same period of 2025 included $16.5 million from an agreement that gives one customer exclusive use of a machine on its premises for 45 months, which the accounts treat as a sale.

The prospectus names JPMorgan Chase, Amgen, Mitsui & Co. and Honeywell as customers that are also research partners. Most of the money has come from elsewhere: RIKEN was 63% of revenue in 2024 and 60% in 2025, and the U.S. government 16% in 2025.

What is promised is a fault-tolerant machine, one that runs long programs reliably because it corrects its own errors. Quantinuum’s date for it is 2029.

The technology in plain words

A qubit is the basic unit a quantum computer computes with. Quantinuum’s design, called QCCD, “uses electromagnetic fields to suspend qubits just a few microns above the chip”, the prospectus says, and then moves them around. It credits the idea to David Wineland, who proposed it at the U.S. National Institute of Standards and Technology in 2002.

The known strength is accuracy. The number to read is two-qubit gate fidelity: how often an operation on a pair of qubits gives the right result. Helios, the current machine, has 98 physical qubits and a fidelity of 99.921%, so about 8 operations in 10,000 fail. The figure comes from a 2025 paper that the prospectus cites as a preprint, a paper posted before peer review. Quantinuum calls it the highest among commercial machines, by its own reading of rivals’ filings.

The known weakness is speed. Superconducting designs “may achieve faster individual gate speeds”, the prospectus says, and the company’s answer is that it puts accuracy ahead of “raw gate speed”.

The prospectus places the neighbours this way. Alphabet (GOOGL), IBM (IBM) and Rigetti Computing (RGTI) use superconducting qubits: tiny manufactured electrical circuits that must run at extremely low temperatures. Infleqtion (INFQ) uses neutral atoms: uncharged atoms steered with lasers and magnetic fields. IonQ (IONQ) traps ions, as Quantinuum does.

The two ion companies handle them differently. IonQ’s technology page describes a “linear ion trap” that holds the ions in a single chain, each addressed by its own laser beam. Quantinuum moves its ions around the chip, and the prospectus calls its “junction” technology, which lets ion paths cross in a grid, “a crucial enabler for scaling”. On accuracy, IonQ’s system comparison lists a two-qubit gate fidelity of 99.6% for its 36-qubit Forte, against the 99.921% Quantinuum reports for the 98-qubit Helios. Each figure is the company’s own.

Accuracy matters because of error correction, which combines several physical qubits into one more reliable logical qubit. In October 2026 Quantinuum said Helios reached a logical fidelity of 99.9995%, “just five errors in a million operations”. The release does not say whether the result has been peer reviewed.

Date Target Status
2020 H1, the first system Introduced
2023 H2 Introduced
November 5, 2025 Helios: 98 physical qubits, 99.921% two-qubit gate fidelity, 48 error-corrected logical qubits Launched, figures by the company’s own measurement
October 2026 99.9995% logical fidelity on Helios Reported by the company, October 19, 2026
Spring 2027 A Helios system running in Singapore Target, assembly under way
Second half of 2027 Sol: 100 logical qubits, with a target of one logical error in 100,000 operations Target, “on track” on November 5, 2026
2029 Apollo: hundreds of logical qubits, universal and fully fault-tolerant Target, “on track” on November 5, 2026

The business

The public record is short: two full years in the prospectus summary and three quarters of 2026. Revenue was $30.9 million in 2025, up from $23.0 million in 2024, and the net loss $192.6 million, up from $144.1 million.

Revenue comes in lumps. It was $19.1 million in the first quarter of 2025, when RIKEN alone was 90% of it, then $2.1 million and $2.7 million in the next two quarters. The 2026 quarters hold no hardware sale and have risen each time, but nine-month revenue is still below last year’s: $22.3 million against $23.9 million.

On November 5, 2026 the company raised its 2026 revenue outlook to between $31 million and $34 million, from $28 million to $32 million. Signed contracts not yet counted as revenue stood at $149.4 million on September 30, 2026, and the filing expects about 46.9% of that to become revenue within twelve months.

The loss from operations was $765.0 million in the first nine months, against $133.8 million a year earlier, and $484.7 million of it is stock compensation. Employee awards could only be expensed once the company listed, so most of that cost was booked at once: the loss from operations reached $555.0 million in the second quarter, according to the filing for that quarter. The nine-month net loss of $851.0 million also includes a $111.8 million non-cash loss on warrants exercised at the listing.

Quantinuum held $1.94 billion on September 30, 2026, up from $762.6 million at the end of 2025, and its balance sheet shows no debt. Over the nine months, operations used $229.9 million and equipment $64.7 million, $294.6 million together. At that pace, about $33 million a month, the cash covers roughly five years. That arithmetic is ours. The prospectus says the company “will require substantial additional capital”.

Funding and ownership

Quantinuum was formed in November 2021 by combining Honeywell Quantum Solutions and Cambridge Quantum. It raised $300 million and then about $600 million in private rounds, and listed on Nasdaq in June 2026.

Date Round or offering Amount Lead or terms
November 30, 2021 Formation Nearly $300 million invested by Honeywell Honeywell held about 54%
January 16, 2024 Equity round $300 million $5 billion valuation before the money; anchored by JPMorgan Chase, with Mitsui & Co., Amgen and Honeywell
September 4, 2025 Equity round About $600 million $10 billion valuation before the money; new investors included Quanta Computer and NVentures, Nvidia’s venture arm
June 5, 2026 Initial public offering $1,628.8 million after underwriting fees 28.5 million Class A shares at $60.00, including 500,000 from the underwriters’ option
September 2026 Research award from the U.S. Department of Commerce Up to $100 million 2,369,528 Class A shares issued to the Department, at a contract price of $42.20

The listed company owns only part of the operating business: 15.6% on September 30, 2026, with owners from before the listing holding the rest directly. Each of their units comes with one Class B share, which carries a vote but no dividend, and can be exchanged for one Class A share. On November 9, 2026 there were 45.6 million Class A shares and 220.3 million Class B shares, so the listed class is about 17% of the 265.8 million total. The market value on this page counts all of them.

Honeywell was the majority owner before the listing. The prospectus said Honeywell entities would hold about 47.8% of the votes once the offering closed, before any use of the underwriters’ option.

People

Ilyas Khan, the founder of Cambridge Quantum, was the first chief executive. Rajeeb Hazra took over as chief executive on February 14, 2023, after 25 years at Intel and a senior role at Micron Technology. Nitesh Sharan joined as chief financial officer on April 6, 2026, from the same post at SoundHound AI.

Risks

  • Few customers. The U.S. government (30%) and two unnamed customers (24% and 17%) made up 71% of revenue in the first nine months of 2026.
  • A roadmap still to be built. “Our roadmaps and plans for commercialization involve technology that is not yet available for customers”, the prospectus says.
  • Spending ahead of sales. The third-quarter loss from operations, $132.8 million, was about 15 times the quarter’s revenue.
  • Suppliers and licences. The ion traps, a core component, are made in Honeywell Aerospace facilities under a ten-year agreement at cost plus 15%. The prospectus also says the technology depends on a licence from Leonardo DRS.
  • A large holder. The listed class is about 17% of the shares, and Honeywell entities held close to half the votes at the listing.

What to watch

  • The fourth-quarter report. The outlook of $31 million to $34 million leaves $8.7 million to $11.7 million for the quarter, by our arithmetic. No date is announced.
  • The first quarter of 2027, when Sol is due to be assembled, and the second half of 2027, when it is due to be introduced.
  • Spring 2027, when the Helios system in Singapore is due to start running.
  • The Class A share count on each filing’s cover: 36.9 million on August 13, 2026, then 45.6 million on November 9, 2026.

Our read: Quantinuum is betting that accuracy comes first. If each qubit makes fewer errors, fewer are needed to build one reliable logical qubit, and a useful machine arrives sooner than on faster but noisier hardware. Until Sol and Apollo exist, revenue mostly measures the budgets of a few research institutions and governments.

Last checked 2026-11-11

Sources

More in Companies

See all

Get Many Worlds Weekly

The main stories in quantum computing for investors, in one email.

Many Worlds Weekly. Free, every Friday. Unsubscribe in one click.