Trapped Ion

Physics / mechanism

Trapped-ion qubits confine individual ions — typically ytterbium (Yb⁺), barium (Ba⁺) or calcium (Ca⁺) — in RF Paul traps using oscillating electromagnetic fields. Qubit transitions (hyperfine or optical states) are driven either by lasers or by microwaves (with a magnetic-field gradient). Two-qubit gate fidelities now routinely exceed 99.9% and the best published figure is 99.99% (IonQ Tempo); coherence times reach minutes to hours, orders of magnitude beyond superconducting qubits, and every ion is identical (no fabrication-variability problem). The structural weaknesses are gate speed (~1 ms vs ~10 ns for superconducting, capping circuit throughput) and wiring/optics overhead per ion — which is what makes integration the whole game (below). Performance is increasingly quoted as logical-qubit counts: Quantinuum’s Helios showed 48 logical qubits via color-code QEC; IonQ/Oxford Ionics target 256 physical qubits in 2026 and >10,000 (networked) by 2027. 2026 06 16 Trapped Ion Landscape 2026

Where it stands (2026) — context

Trapped ion is, alongside superconducting, the most commercially mature modality, and arguably leads on the metrics that matter for fault tolerance (fidelity, coherence, all-to-all connectivity). The field has split into two control philosophies, and the competitive question is which scales:

• Laser-driven gates (IonQ, Quantinuum, AQT) — the established route; highest fidelities, but needs lots of precisely-aligned optics per ion.
• Microwave / electronic control (Oxford Ionics, Universal Quantum, eleQtron) — drives gates with on-chip electronics/microwave fields instead of lasers, betting that semiconductor-fab-style electronic control scales better than optics.

The binding constraint is shifting from qubit count to integration and interconnect. Two scaling paths, not mutually exclusive: (1) QCCD (quantum charge-coupled device) — shuttle ions between zones on a single chip (Quantinuum’s approach); (2) photonic networking — entangle ions across separate trap modules over optical links (the route to >10k qubits in every public roadmap). This is why the modalities thesis predicts interconnect/networking becomes the openly-acknowledged “next wall” — see Quantum Computing Modalities.

Conventional ion traps deliver control light with bulk free-space optics — large optical tables, mirrors, manual alignment — which does not scale past a lab. The frontier is putting the optics on the trap chip: integrated waveguides, grating couplers, splitters and modulators delivering multi-wavelength light (Doppler cooling, state prep, gates, readout) to each zone. Demonstrated 375–866 nm delivery in alumina and silicon-nitride waveguides; leading groups are MIT Lincoln Laboratory, ETH Zurich, Sandia and UC Berkeley (integrated optical MEMS). 2026 06 16 Trapped Ion Landscape 2026

Competitive landscape — trapped-ion vendors

vs other modalities

Superconducting (Superconducting Qubits — IBM, Google) leads on speed and fab scalability but trails on fidelity/coherence. Neutral atoms (Atom Computing, QuEra, Planqc) offer similar coherence with faster Rydberg gates, lower demonstrated fidelity. Photonic QC (Photonic Quantum Computing — PsiQuantum, Xanadu) is room-temperature but probabilistic. Silicon spin is CMOS-compatible but early on fidelity.

Investability & routes

Substance first: if trapped ion is a winning modality, the durable value is split between (a) the full-stack machines — already captured by scaled players (IonQ is public; Quantinuum is heading to a ~$20B+ IPO; Oxford Ionics sold for $1.075B), so this layer is largely not a pre-seed opportunity; and (b) the enabling hardware layer — photonic I/O chips, ion-trap MEMS, control ASICs, vacuum/packaging — which is where new, capital-efficient companies can still form. Per Quantum Computing Modalities, the enabling silicon/photonics is the fund-shaped layer in a non-photonic modality, not the machine itself.

Companies using

Connected ideas

Sources

Frontier (open questions)

• To be added.