01 Why These Two Companies?
The two dominant, publicly-traded trapped-ion quantum computing companies, and why they are worth comparing directly.
Trapped-ion qubits are individually charged atoms confined by electromagnetic fields and manipulated with laser or microwave/RF pulses. The approach has long held the industry's best gate fidelities and full qubit-to-qubit connectivity, at the cost of gate speed and, historically, a very complex laser system for every additional qubit. Two companies dominate the commercial trapped-ion field: Quantinuum (formed in 2021 from the merger of Honeywell Quantum Solutions and Cambridge Quantum, majority-owned by Honeywell, IPO'd on Nasdaq in June 2026 under ticker QNT) and IonQ (spun out of Duke University and the University of Maryland in 2015, publicly traded on the NYSE since a 2021 SPAC merger under ticker IONQ). Both are racing toward fault-tolerant quantum computing this decade, but via structurally different bets on how ion-trap hardware should scale.
(Nature 2026, peer-reviewed)
logical qubits (Helios, 2:1 ratio)
(company spec, full system)
physical qubits (Sep 2026)
arXiv:2510.17286 — preprint
02 Architecture & Qubit Control
Both companies trap ions in a QCCD architecture — but they address and gate those ions in very different ways.
Quantinuum's H-series uses a quantum charge-coupled device (QCCD) architecture: ions are held in a linear or racetrack-shaped trap with distinct zones for storage, gates, and readout, and are physically shuttled between zones using time-varying electric fields. Two-qubit gates are driven optically, with tightly focused, individually-steered laser beams addressing pairs of ions once they are transported into a gate zone.
- H1/H2 generation: ¹⁷¹Yb⁺ hyperfine qubits, linear/racetrack QCCD trap company
- Helios generation (2025–): switched the qubit species to ¹³⁷Ba⁺ hyperfine qubits, with co-trapped ¹⁷¹Yb⁺ ions used purely as a sympathetic-cooling species peer-reviewed
- Helios trap geometry: a rotatable ion-storage ring connected to two gate/operation regions by a four-way junction, giving full all-to-all connectivity without a linear-chain qubit-count penalty
- Gate control hardware: acousto-optic and electro-optic beam-steering systems must scale roughly with qubit count and gate-zone count — the laser system is the dominant engineering bottleneck as qubit number grows
- Advantage: laser gates are extremely well-characterized after 15+ years of AMO ion-trap physics; very high fidelities are routine established technique
IonQ's first four hardware generations (Harmony, Aria, Forte, Tempo) also used QCCD-style trapping with laser gates, steered by acousto-optic deflectors (AODs), on ¹⁷¹Yb⁺ ions with all-to-all or many-to-many connectivity. Starting with its sixth-generation Superion platform, IonQ replaced laser beam-steering with control electronics fabricated directly onto the trap chip.
- Forte / Forte Enterprise / Tempo (2023–2026): ¹⁷¹Yb⁺ qubits, AOD laser gate control, all-to-all (Forte) or many-to-many (Tempo) connectivity company spec
- Electronic Qubit Control (EQC): semiconductor-fabricated electrodes drive individual-ion gates directly, eliminating most of the free-space laser-steering optics that otherwise must grow with qubit count; acquired via the ~$1.075B purchase of Oxford Ionics, closed September 2025
- Manufacturing: EQC chips are fabricated at SkyWater Technology, a Minnesota semiconductor foundry IonQ agreed to acquire for ~$1.8B in January 2026 (shareholder-approved May 2026); IonQ reports design-cycle time fell from 9 months to 2 months and wafer-lot output rose 12× over six months after the acquisition company
- Superion 256 (Sept 2026): first EQC-based commercial system, 256 qubits, a standard server-rack footprint, and power draw below a typical GPU rack — explicitly designed to be "built by the hundreds," i.e. as a manufacturable product rather than a one-off lab machine
03 Hardware Generations, Side by Side
Both companies iterate fast. Here is the generation-by-generation timeline, with what's demonstrated versus roadmap clearly marked.
Quantinuum H-Series
IonQ Product Line
Demonstrated physical qubit counts for current, purchasable systems only (log scale). Roadmap targets (Sol, Apollo, Superion 10K) are shown separately in Section 08.
04 Gate Fidelity Deep Dive
The single most important number in trapped-ion QC: how much error does one two-qubit gate introduce?
Two-qubit gate infidelity is the dominant error source in most trapped-ion circuits, and it sets the overhead needed for quantum error correction. Both companies report their numbers as average infidelity across many randomly-selected Clifford circuits (randomized benchmarking), which is the standard, defensible way to characterize a real device — but "average" and "worst-case" can differ, and a full-system number (averaged over many qubit pairs, over hours of drift) is a stronger claim than a single best-pair, best-moment number.
Two-qubit gate fidelity by system and publication status. Dashed line marks the ~99.9% fidelity generally considered a practical threshold for useful near-term circuits.
- H2 (2023–2024): >99.99% 1Q, >99.9% 2Q, all qubit pairs, company-quoted and consistent with peer-reviewed arXiv:2305.03828 peer-reviewed
- Helios (2026): 99.9975% 1Q, 99.921% 2Q, 99.967% SPAM — all averaged across the full 98-qubit device, Nature 655, 81–86 peer-reviewed
- Note: Quantinuum's own Helios launch blog rounds SPAM fidelity to "99.99%"; the peer-reviewed paper's SPAM infidelity of 3.3(5)×10⁻⁴ corresponds to 99.967% — a good example of why this page prefers the peer-reviewed number when the two differ marketing rounding
- Forte / Forte Enterprise: 99.98% 1Q, 99.6% 2Q, full system, all-to-all company spec
- Tempo: 99.99% 1Q, 99.9% 2Q, full system, many-to-many company spec
- EQC record (Oct 2025): >99.99% 2Q fidelity on a single ion pair, without ground-state cooling, arXiv:2510.17286 preprint, component-level
05 Logical Qubits & Fault Tolerance
Physical fidelity only matters because it sets how many physical qubits are needed per logical, error-corrected qubit.
- Apr 2024: 4 logical qubits on H2 (99.8% physical 2Q fidelity at the time), >14,000 circuits run without a single logical error reported by Microsoft; described as the largest reported gap between physical and logical error rates to date company blog
- Sep 2024: 12 logical qubits; circuit error rate 0.0011 (about 22× better than the 0.024 physical baseline); used two of the logical qubits for a real hybrid workflow estimating the ground-state energy of a catalytic P–N–N–P iron intermediate, combined with classical HPC (AutoCAS/AutoRXN) company blog, arXiv preprint referenced
- Nov 2025 (Helios launch): company claims 94 fully-entangled logical qubits at break-even fidelity, 50 error-detected logical qubits above break-even, and 48 fully error-corrected logical qubits at a 2:1 physical-to-logical qubit ratio — the underlying physical-layer numbers are peer-reviewed (Section 04); the logical-qubit-count claims themselves are company-reported, not yet in a peer-reviewed logical-qubit paper company
- Jun 2026: arXiv:2606.06455 reports a trapped-ion demonstration of logical-qubit break-even (logical memory lifetime exceeding the underlying physical qubit lifetime) using a bivariate-bicycle quantum low-density parity-check (qLDPC) code
- Configuration: 40 physical ion qubits encoding 4 logical qubits via the "BB5" code (a weight-5 generalized qLDPC matrix, "4 logical qubits in 18 physical containers"); logical memory lifetime 3.95 s vs 1.1 s for the physical qubits
- Reports logical error rates roughly 4× (Z-basis) and 9× (X-basis) lower than a comparable superconducting surface-code result reported elsewhere in the literature — not a direct comparison against Quantinuum's own logical-qubit numbers above
- Two independent secondary sources describing this preprint both report the physical qubits used were ¹³³Ba⁺ ions, not IonQ's commercial-product ¹⁷¹Yb⁺/EQC line — worth flagging as likely a separate research demonstration rather than a Superion-family result preprint, unusual ion species, verify before reuse
06 Full Head-to-Head Comparison
Every major dimension, side by side. Advantage marked in bold color where the evidence supports a clear edge.
| Category | Metric | Quantinuum | IonQ | Edge |
|---|---|---|---|---|
| Company | Founded / origin | 2021 merger of Honeywell Quantum Solutions + Cambridge Quantum | 2015 spin-out of Duke University / University of Maryland | Different histories |
| Public listing | Nasdaq: QNT, IPO June 2026 | NYSE: IONQ, public since 2021 (SPAC) | IonQ longer public track record | |
| Majority owner | Honeywell (~49% voting power post-IPO) | No single majority owner; widely held | Different structures | |
| Headquarters | Broomfield, Colorado / Cambridge, UK (dual HQ) | College Park, Maryland | Tie | |
| Architecture | Trap type | QCCD racetrack, rotatable storage ring + 4-way junction (Helios) | QCCD-style, chip-integrated electrodes (Superion) | Both QCCD family |
| Gate control | Laser-driven (AOD/EOM beam steering) | Electronic Qubit Control (EQC), chip-integrated, from Superion onward | IonQ for manufacturability; Quantinuum for maturity | |
| Current qubit species | ¹³⁷Ba⁺ (Helios), was ¹⁷¹Yb⁺ (H1/H2) | ¹⁷¹Yb⁺ (Forte/Tempo/Superion) | Different choices | |
| Connectivity | Full all-to-all, 98 qubits (Helios) | All-to-all (Forte) / many-to-many (Tempo) | Quantinuum slightly stronger claim at larger qubit count | |
| Fidelity | 2Q fidelity, full system, peer-reviewed | 99.921% (Helios, Nature 2026) | 99.9% (Tempo, company spec only — not independently peer-reviewed) | Quantinuum |
| 2Q fidelity, best demonstrated anywhere | 99.921% (full 98-qubit system) | >99.99% (single ion pair, preprint) | IonQ if the pair-level result scales | |
| 1Q gate fidelity | 99.9975% (Helios, peer-reviewed) | 99.99% (Tempo, company spec) | Quantinuum (narrowly) | |
| SPAM fidelity | 99.967% (Helios, peer-reviewed; company blog rounds to 99.99%) | Not independently published for Tempo/Superion | Quantinuum (only side with a peer-reviewed number) | |
| Scale | Current physical qubits | 98 (Helios) | 256 (Superion 256) | IonQ on raw count |
| Manufacturing model | Lab-built, laser-aligned systems | Semiconductor-fab-built chips, targeting hundreds of units/year | IonQ (if realized) | |
| 2030 roadmap physical qubit target | "Thousands" (Apollo, imprecise) roadmap | 2,000,000 roadmap | Both unverified projections | |
| Error Correction | Logical qubits demonstrated | 48 fully error-corrected (2:1 ratio, company-reported) | 4 logical qubits, qLDPC break-even (preprint) | Quantinuum on count and maturity |
| Code approach | Color codes / QEC with Microsoft (details vary by demo) | Bivariate-bicycle qLDPC (BB5) | Different code families | |
| Business | Market cap / valuation | ~$14.3B IPO valuation (Jun 2026) | ~$14.7B market cap (Sep 2026) | Roughly comparable |
| Q2 2026 revenue | $8M (+279% YoY) | $80.1M (+287% YoY) | IonQ, roughly 10× larger | |
| FY2026 revenue guidance | $28–32M | $290M | IonQ | |
| Recent capital raised | $1.7B (June 2026 IPO) | ~$2.9B combined for Oxford Ionics + SkyWater acquisitions approximate | Both well-capitalized | |
| Cloud & Access | Cloud platforms | Microsoft Azure Quantum, Quantinuum's own Nexus platform, Oracle OCI (Helios) | AWS Braket, Microsoft Azure Quantum, Google Cloud Marketplace | IonQ broader multi-cloud reach |
| Software stack | TKET compiler/SDK, Guppy, Quantinuum Nexus | Qiskit, Cirq, and native SDK support via cloud partners | Both broadly compatible |
07 Business, Funding & Public Markets
Both companies are now public, giving an unusually transparent window into the economics of trapped-ion QC.
- IPO priced early June 2026 on the Nasdaq Global Market, ticker QNT, upsized to 26.5M shares at $53–55/share, implying a valuation of roughly $14.3B (up from an initial ~$12.7B target)
- Honeywell retains ~49.1% of voting power post-IPO; Cambridge Quantum-linked entities hold ~32.5%; founder Ilyas Khan is the largest individual shareholder, with a stake reported above $2B
- Q2 2026 (first earnings report since IPO, Aug 12 2026): revenue $8M, up 279% YoY from $2M; full-year 2026 guidance $28–32M
- GAAP net loss $597M in Q2 2026 (vs $57M in Q2 2025) — a jump that most likely reflects non-cash, IPO-related stock-compensation and transaction charges rather than a sudden change in operating cash burn; adjusted EBITDA loss was $68M, widened from $43M a year earlier
- Ended Q2 2026 with $2.1B in cash and short-term investments after the $1.7B June IPO raise company filings
- Public since October 2021 via SPAC merger with dMY Technology Group III, trading on the NYSE as IONQ
- Market capitalization ~$14.72B as of September 15, 2026 — down 7.4% from ~$15.90B at end of 2025, illustrating typical volatility for the sector rather than a steady climb
- Q2 2026 revenue: record $80.1M, up 287% year-over-year; full-year 2026 guidance raised to $290M
- Two major 2025–2026 acquisitions reshaped the balance sheet: Oxford Ionics (~$1.075B, closed Sept 2025) and SkyWater Technology (~$1.8B, agreed Jan 2026, approved May 2026) — both aimed at vertically integrating EQC chip design and fabrication rather than at near-term revenue
- Other recent acquisitions (quantum networking): ID Quantique and Qubitekk, expanding into quantum-safe networking and entanglement distribution alongside the core computing business company filings
08 Roadmaps to Fault-Tolerant Quantum Computing
Both companies publish detailed, aggressive multi-year roadmaps. None of the future years below are demonstrated results.
Company-published logical-qubit roadmap targets by year, log scale. IonQ publishes year-by-year figures through 2030; Quantinuum's Apollo target is a single 2029 milestone reported only as an order-of-magnitude ("hundreds" of logical qubits). Every point on this chart is a company projection, not a measurement.
| Year | IonQ roadmap (published targets) | Quantinuum roadmap |
|---|---|---|
| 2025 | 64–100+ physical qubits, 99.9% fidelity | Helios launch (98 qubits) — achieved, not projected |
| 2026 | 100–256+ physical, 99.99% fidelity, 12 logical qubits, logical error <10⁻⁷ | Sol development continues; Helios peer-reviewed (achieved) |
| 2027 | 10,000 physical qubits, 800 logical qubits | Sol target system |
| 2028 | 20,000 physical qubits, 1,600 logical qubits | Scaling toward Apollo |
| 2029 | 200,000 physical qubits, 8,000 logical, error <10⁻¹², photonic interconnect introduced | Apollo target: thousands of physical qubits, hundreds of logical qubits, universal fault tolerance |
| 2030 | 2,000,000 physical qubits, 80,000 logical qubits | No public target beyond Apollo as of this writing |
09 Companies & Ecosystem
Partners, customers, academic roots, and how each platform maps onto the broader quantum-computing job market.
Quantinuum Ecosystem
Honeywell ↗
Majority owner and original parent (Honeywell Quantum Solutions). Retains ~49% of voting power post-IPO. Long history in precision control systems and aerospace, which fed directly into Honeywell's original ion-trap control electronics.
Microsoft / Azure Quantum ↗
Deep logical-qubit collaboration: the April 2024 (4 logical qubits, 800× error reduction) and September 2024 (12 logical qubits, chemistry simulation) demonstrations were joint Microsoft–Quantinuum work. Azure Quantum also offers Quantinuum hardware access to cloud customers.
Sandia National Laboratories
Co-authors on the peer-reviewed Helios Nature paper (2026); a recurring government/national-lab collaborator across Quantinuum's fidelity benchmarking work.
JPMorgan Chase, Amgen, Mitsui & Co., NVIDIA
Named commercial and technology partners across financial random-number generation, drug discovery, and hybrid quantum-classical (CUDA-Q) workflows — illustrative of Quantinuum's enterprise-first go-to-market, distinct from IonQ's broader multi-cloud self-serve access model.
Cambridge Quantum heritage
The 2021 merger partner contributed the TKET compiler and quantum chemistry/NLP software lines that remain part of Quantinuum's software stack today.
IonQ Ecosystem
Duke University / University of Maryland
IonQ's founding academic roots: co-founders Christopher Monroe (Duke) and Jungsang Kim (Duke) built on decades of trapped-ion AMO physics. Strong ongoing recruiting pipeline from both universities' AMO and quantum-information groups.
Oxford Ionics ↗
UK spin-out from Oxford University's ion-trap group, acquired by IonQ in September 2025 (~$1.075B). Originated the Electronic Qubit Control (EQC) technology now central to the Superion product line.
SkyWater Technology
US (Minnesota) semiconductor foundry, agreed acquisition ~$1.8B (Jan 2026, shareholder-approved May 2026). Now fabricates IonQ's EQC ion-trap chips, cutting design-cycle time from nine months to two.
AWS, Microsoft Azure, Google Cloud
IonQ hardware is available through all three major cloud marketplaces — a broader multi-cloud footprint than Quantinuum's more enterprise-direct + Azure-centric model, and a real differentiator for researchers wanting quick, no-contract access.
ID Quantique & Qubitekk
Quantum-networking and entanglement-distribution acquisitions, positioning IonQ beyond standalone computing hardware into the broader "quantum internet" and quantum-safe communications space.
10 Key Papers & References
Primary sources for every claim above: peer-reviewed papers, preprints, and company/press material, clearly separated.
Quantinuum, Landmark Papers & Releases
The Helios physical-layer paper: 98 ¹³⁷Ba⁺ qubits with ¹⁷¹Yb⁺ sympathetic cooling, QCCD architecture with a rotatable ion-storage ring and four-way junction. Single-qubit infidelity 2.5(1)×10⁻⁵, two-qubit infidelity 7.9(2)×10⁻⁴, SPAM infidelity 3.3(5)×10⁻⁴.
Peer-reviewed-adjacent technical description underlying the H2 56-qubit, all-to-all connected system and its 99.9%-class two-qubit gate fidelity claims.
4 logical qubits, error rate reported as 800× better than physical qubits, >14,000 circuits without a logical error, on H2-generation hardware at 99.8% physical 2Q fidelity.
12 logical qubits, circuit error rate 0.0011 (~22× improvement over physical baseline); two logical qubits used to estimate the ground-state energy of a catalytic iron intermediate in a hybrid quantum + classical HPC workflow.
Commercial launch claims for Helios, including the 94 entangled / 50 error-detected / 48 fully error-corrected logical qubit figures, ahead of the June 2026 peer-reviewed physical-layer paper above.
IPO pricing, share count, Honeywell/Cambridge Quantum voting-power breakdown, and comparison to IonQ's market value at the time.
IonQ, Landmark Papers & Releases
The headline EQC fidelity record: two-qubit gate fidelity exceeding 99.99% on a single ion pair, achieved without ground-state cooling. Component-level result; full-system reproduction not yet public. Not yet peer-reviewed.
4 logical qubits from 40 physical ion qubits via a bivariate-bicycle "BB5" qLDPC code; logical memory lifetime 3.95 s vs 1.1 s physical (break-even). Secondary reporting describes the ions used as ¹³³Ba⁺, distinct from IonQ's commercial ¹⁷¹Yb⁺ product line — flagged here as worth independent verification. Preprint, not peer-reviewed.
Launch details for the first EQC-based commercial platform: 256 qubits, server-rack footprint, SkyWater fabrication, 2027 customer deliveries, >300× projected cost-per-qubit reduction across the Superion roadmap.
Official specification sheet for Forte, Forte Enterprise, and Tempo: qubit counts, connectivity, and 1Q/2Q fidelity figures used throughout this page's comparison table.
Year-by-year physical/logical qubit and fidelity targets through 2030, cited in Section 08. Entirely forward-looking; treat every figure as a company projection.
Q2 2026 revenue $80.1M (+287% YoY), FY2026 guidance raised to $290M, and detail on the SkyWater and Oxford Ionics integration.
Independent Surveys & Reviews
Independent survey comparing Quantinuum, IonQ, AQT, eleQtron, Universal Quantum, and QUDORA. Cross-checked against primary sources throughout this page rather than taken at face value.
Independent technology-press analysis of the Helios launch and its error-correction claims, useful context alongside Quantinuum's own launch materials.