Quantum Computing IPOs: Quantinuum, Infleqtion Lift 2026 Guidance

Two quantum debuts, one test

Quantinuum and Infleqtion took different routes to the public markets in 2026. Both are now reporting as listed companies, and their results show what investors are paying for in quantum computing. Most of the value sits in hardware roadmaps and error-correction milestones. Revenue is still small.

Quantinuum reported its first quarter as a public company on August 11. Second-quarter revenue was $8 million, up 279% from $2 million a year earlier, and the company issued 2026 revenue guidance of $28 million to $32 million.21 Infleqtion reported the next day. It posted record second-quarter revenue of $12.6 million, up 116% year over year, and raised its full-year outlook to about $43 million.2 Both companies beat expectations. Quantinuum's revenue topped a consensus of roughly $7.6 million.25 Infleqtion's beat an estimate of about $10.64 million.10

Then, in September, each company announced an error-correction result that it presented as proof that its roadmap is working. Those results matter more to the investment case than either earnings report. They also deserve closer reading than the press releases invite.

Quantinuum: a large IPO and a large loss

Quantinuum is the bigger company. It priced its upsized IPO at $60 a share, above a range that had already been raised to $53 to $55, and raised $1.68 billion.23 Shares ended the first trading day little changed, giving the company a market value of about $15.7 billion.23 Honeywell kept about 48.1% of the combined voting power.26 Before the deal, an IPOX analyst said the stock would need to price in years of successful execution before revenue could catch up with the valuation.27

The second-quarter loss was large. GAAP net loss was $597 million, compared with $57 million a year earlier.25 Most of the gap came from one-time accounting. The company recorded $447.5 million of stock-based compensation, mostly a catch-up charge triggered when the IPO caused employee equity grants to vest.30 The adjusted non-GAAP net loss was $73 million, and the adjusted EBITDA loss was $68 million.30

The 279% growth figure needs context. Quantinuum's quarterly 10-Q shows first-half 2026 revenue of $13.2 million, down from $21.2 million in the first half of 2025.22 First-quarter revenue had fallen 73%, to $5.24 million.23 Most of that drop came from a $16.5 million sales-type lease that was booked upfront in early 2025.24 Quarterly revenue at Quantinuum depends on a few large system deals. Management said as much, arguing that quarter-to-quarter swings say little about long-term progress.30

The forward-looking numbers are more useful. Remaining performance obligations were about $74 million, and year-to-date bookings were roughly $81 million once deals signed after quarter-end, including Oracle, are counted.30 Cash stood at about $2.1 billion.21 The market's reaction was muted. Shares closed at $56.29 on results day and slipped to about $55 after hours, below the $60 IPO price.30

Infleqtion: smaller and more diversified

Infleqtion went public through a merger with Churchill Capital Corp X and began trading on the NYSE in February.32 It has a different profile from Quantinuum. Its second-quarter revenue was higher than Quantinuum's, all of it organic and all of it from quantum products.2 Much of that revenue comes from sensing rather than computing. Work on NASA's Quantum Gravity Gradiometer program was a major driver of growth, and the company also sells Tiqker optical atomic clocks through a co-selling partnership with Safran.2

Infleqtion's losses are much smaller. GAAP net loss was about $25.5 million, and the non-GAAP operating loss was $17 million.3 The company ended the quarter with $582 million in cash and securities and no debt.2 That figure includes a temporary $27.4 million benefit from payroll taxes collected on stock-option exercises, which the company expects to pay out in the third quarter.2 Without that benefit, operating cash burn was about $14 million.2

The main concern raised about Infleqtion is where its customers come from. One analysis put government contracts at more than 75% of revenue and noted that operating expenses are growing faster than sales.4 The company's own announcements point the same way. Its pipeline includes a proposed $100 million CHIPS Act funding letter of intent and three Department of Energy Genesis Mission projects.32 Government funding gives the company stability, but it is not commercial demand. The Commerce Department's broader plan to take equity stakes in nine quantum companies, including $100 million for Quantinuum, shows how much of this sector's financing now depends on policy.23

Quantinuum's Helix result: strong error correction

Quantinuum has the stronger technical story. Its Helios system launched in November 2025 with 98 barium-ion qubits and two-qubit gate fidelity of 99.921%.2012 Because the physical error rates are low, Helios needs only about two physical qubits for each logical qubit.20

The more important result came in September. On Helios, Quantinuum tested Helix, an error-correction architecture built on a [[20,2,6]] code. It measured 4.6 x 10^-5 errors per logical qubit per correction cycle without discarding any runs.15 Logical two-qubit Clifford operations had an error rate of 2.8 x 10^-4. That is 4.28 times better than the physical gates underneath, again without post-selection.17 A technique called adaptive syndrome extraction cut the number of physical gates by about 33% and run time by about 23%.13 One comparison puts the logical-memory result at roughly one failure in 22,000 correction rounds, against about one in 700 for Google's Willow surface code from December 2024.15 Those are different codes on different hardware, so the comparison is not exact, but it shows the scale of the gap.

The roadmap is concrete. Sol is due in 2027 and Apollo, a fully fault-tolerant machine, in 2029.12 Sol's trap chip is back from fabrication and being validated, while Apollo remains a design.12

Published logical-qubit counts for Helios vary. Some trackers say 50 error-corrected logical qubits.11 Others list 48 and describe them as error-detected.16 Quantinuum's own account separates the experiments: 48 error-corrected logical qubits at about 2:1 encoding, 50 error-detected in a magnetism simulation, and 94 error-detected in an entangled GHZ state.1213 The distinction matters. Error detection throws out bad runs. Error correction repairs them while the computation continues. In my view, the Helix result is more meaningful than any of these headline counts, because it shows logical operations beating physical ones without discarding runs.

Infleqtion's 30 logical qubits: met on time, with caveats

On September 24, Infleqtion said it had entangled 30 logical qubits using 80 physical atoms on its Sqale system. That met its 2026 target, and the company called it a first for a commercial neutral-atom machine.31 The circuit ran about 1,000 physical operations, including four non-Clifford CCZ gates. Roughly 25% of the measured outputs fell within the set an ideal circuit could produce, about 1,000 times the rate of random output.33 The company also said an AI model helped it find a new entangling operation that needs four physical two-qubit gates instead of eight.33

The company's chief technology officer published the technical details. Each logical qubit was prepared in a distance-3 code, but the circuits ran in a distance-2 [[8,3,2]] code. That code can detect any single error or correct for a lost atom, but it cannot correct general errors.33 One independent analysis says the result relied on post-selection. It also notes that the milestone replaced an earlier 2026 target that called for error-corrected logical qubits.32 A second review found that the published write-up gives no shot counts, acceptance rates or error bars. It also found no peer-reviewed paper or independent replication as of late September, and said 30 logical qubits is not a world record.38 Infleqtion says a full paper will follow and that its next goals include real-time control and mid-circuit measurement, so it can move beyond post-selection.33

Infleqtion does have a record of meeting its dates. It reported 2 logical qubits in 2024 and 12 in 2025, beating a 2026 target of 10, before reaching 30 this year.32 It also has three customers running logical-qubit circuits, and the company says it is on track for 100 logical qubits in 2028 and 1,000 by 2030.36

What the comparison shows

Across this coverage, both companies are executing and both are being rewarded for it. The quality of their error-correction results differs. Quantinuum showed below-physical logical error rates without discarding runs, on a machine with a funded path to fault tolerance. Infleqtion showed scale and speed, but its result relies on error detection and still has to show genuine error correction.

As investments, the risks run the other way. Quantinuum's valuation, which reached $15.7 billion at its debut, assumes Sol and Apollo arrive on schedule.23 Infleqtion's market value of roughly $2.5 billion rests on a sensing business that already brings in revenue and on a smaller cash burn.8 For both stocks, the next checkpoints are technical. Sol's validation and Infleqtion's first distance-3 run with error correction during the computation are likely to move the shares more than any quarterly revenue figure.