Quantum hardware
Six qubit modalities compete on different axes (fidelity, connectivity, coherence, gate speed, scalability). No modality has won; the practical winner will likely be whoever reaches hundreds of fault-tolerant LOGICAL qubits first. Physical-qubit counts are marketing until paired with fidelity and error correction.
Superconducting transmon
Microwave-driven Josephson-junction circuits at ~15 mK; fast (~microsecond) gates, chip-fab-friendly, but short coherence (~100 us) and mostly nearest-neighbor connectivity.
Strengths: Fast gates; mature fabrication; largest physical-qubit counts.
Weaknesses: Short coherence; limited connectivity; cryogenic overhead; wiring fan-out.
| Who | Metric | Best known | Date |
|---|---|---|---|
| Google Willow | below-threshold surface code | distance-7, 101 qubits, 0.143%/cycle; Lambda=2.14 per +2 distance; 2.4x over best physical qubit | 2024-12 (Nature 638:920, 2025)fact |
| IBM | two-qubit fidelity (Heron) | materially improved; roadmap to Kookaburra ~4,158 physical qubits + qLDPC memory (~360 qubits, 7,500 gates) | 2026 targettarget |
| 2Q physical fidelity | 99.88% | 2026fact |
Roadmap: IBM: modular Quantum System Two -> 100,000+ qubits via chip-to-chip links; 200 logical qubits by 2029 (target). Google: fault-tolerant logical qubits before 2030 (target).
Trapped ion
Individual ions confined in EM traps, laser/microwave gates; the highest fidelities and all-to-all connectivity, but slow (~millisecond) gates limit shot throughput.
Strengths: Highest gate fidelity; all-to-all connectivity; long coherence (s-min).
Weaknesses: Slow gates (throughput); laser/optics complexity; scaling ion counts is hard.
| Who | Metric | Best known | Date |
|---|---|---|---|
| Quantinuum Helios | error-corrected logical qubits | 50 logical at 2:1 encoding | 2025-11demo |
| Quantinuum + Microsoft (H2) | logical qubits below physical error | 12 logical at ~2e-3 logical error ('reliable quantum computing') | 2026-03demo |
| Quantinuum H2 | logical qubits beyond break-even (postselected) | up to ~94 error-protected; 48 from 98 physical at 2:1 | 2026-03demo |
| IonQ | 2Q gate fidelity | 99.9923% ('four nines', EQC) - world record | 2025-10fact |
| IonQ Tempo | algorithmic qubits | 64 AQ, all-to-all | 2026-Q1fact |
Roadmap: Quantinuum Apollo: 100+ qubits; long-term 1,000+ ions below threshold (target). IonQ: photonic interconnect to 1,000+ ions; 800 logical qubits by 2027 (target; no below-threshold demo yet).
Neutral atom
Neutral atoms in optical tweezer arrays, Rydberg-state entangling gates; reconfigurable connectivity and large arrays, strong recent fault-tolerance progress.
Strengths: Large arrays (1000+); reconfigurable/movable qubits; transversal gates; strong 2025-26 FT results.
Weaknesses: Atom loss; slower cycle times; measurement/reset engineering.
| Who | Metric | Best known | Date |
|---|---|---|---|
| Atom Computing | physical qubits | >1000 neutral-atom qubits | 2024-25fact |
| Harvard (Lukin) | integrated fault-tolerance ingredients | 448-atom: below-threshold surface-code-style QEC + transversal gates + teleportation universality + mid-circuit reuse + constant-entropy deep circuits | 2025demo |
| Infleqtion | error-corrected logical qubits | 12 logical on neutral atoms | 2025demo |
Roadmap: Rapid FT progress; considered a top contender for first hundreds of logical qubits alongside superconducting/ion platforms.
Photonic
Qubits encoded in light (single photons / squeezed states); room-temperature-ish, network-native, measurement-based; PsiQuantum bets on fusion-based FT at fab scale.
Strengths: Coherence-immune (photons don't decohere like matter); networking/interconnect native; silicon-photonics fab.
Weaknesses: Probabilistic gates / photon loss; huge resource-state overhead; detectors need cryo.
| Who | Metric | Best known | Date |
|---|---|---|---|
| PsiQuantum | architecture | fusion-based fault-tolerant photonic; utility-scale system build-outs announced (Brisbane AU, Chicago US) | 2024-26target |
| Xanadu | photonic milestones | Borealis quantum-advantage sampling; Aurora networked-modules demo | 2022-25demo |
Roadmap: Long-horizon bet; if the photonic FT approach pays off it scales via manufacturing rather than qubit-by-qubit.
Silicon spin qubit
Electron/nuclear spins in silicon quantum dots; CMOS-compatible, tiny footprint, high fidelity but early on counts.
Strengths: CMOS/foundry-compatible; smallest qubits; high fidelity; potential for dense integration.
Weaknesses: Very early on qubit counts; uniformity/crosstalk; control wiring.
| Who | Metric | Best known | Date |
|---|---|---|---|
| Silicon Quantum Computing | gate fidelity | ~99.99% (fidelity-leaderboard co-leader) | 2026fact |
| Intel | platform | Tunnel Falls spin-qubit chip on 300mm CMOS line | 2023-25fact |
Roadmap: Betting on foundry manufacturability to scale once per-dot control is solved.
Topological
Information stored non-locally in anyons/Majorana modes; intrinsically error-protected IF the physics holds - long-contested experimentally.
Strengths: Would be hardware-level error protection (fewer physical qubits per logical).
Weaknesses: Existence/measurement of the underlying quasiparticles remains scientifically contested; least mature.
| Who | Metric | Best known | Date |
|---|---|---|---|
| Microsoft | Majorana 1 / topological qubit | claimed topological-qubit device; underlying Majorana evidence disputed in the literature | 2025interp |
Roadmap: Highest-risk / highest-reward; not yet a demonstrated computing platform.
2Q gate-fidelity leaderboard (2026-03)
2Q gate fidelity (higher = better)
| Who | Fidelity | Modality |
|---|---|---|
| IonQ | 99.99% | trapped ion |
| Silicon Quantum Computing | 99.99% | silicon spin |
| Quantinuum | 99.97% | trapped ion |
| IQM | 99.91% | superconducting |
| Infleqtion | 99.73% | neutral atom |
Curated + graded knowledge base, aggregated from the research corpus and refreshed by a scheduled tracker. Grades: fact demo target estimate interp. The live feed is machine-collected and unverified. contact resistant@tuta.com