Technologies · 110

1 · Qubit carrier

• Transmon transmon — Anharmonic LC oscillator; ~200–300 MHz anharmonicity bounds gates at ~10 ns; T1 ~70–100 µs.
• Fluxonium fluxonium — Low-frequency (0.2–1 GHz) superconducting qubit with large anharmonicity; longer T1, flux-biased.
• Bosonic cavity mode cavity — Harmonic mode of a 3D/planar superconducting resonator; single-photon T1 10 ms (Al, Yale 2013), cavity-qubit T2 34 ms (Weizmann 2023); errors = photon loss (str
• Trapped atomic ion ion — Yb⁺/Ba⁺/Ca⁺ ions in rf traps; motional coupling (MHz) bounds gates at µs; hour-scale memory.
• Alkali atom (Rb/Cs) in optical tweezers alkali — Hyperfine qubit; Rydberg interaction (MHz–GHz) allows sub-µs gates; T2 ~1–13 s; dominant error is atom loss.
• Alkaline-earth(-like) atom (Yb/Sr) — erasure-native ae_atom — Nuclear-spin / clock-state qubits with metastable levels; decays are detectable → erasures (98% theory, 56% shown).
• Single photon (discrete variable) photon — Heralded (SFWM) or quantum-dot photons; entangling is probabilistic (fusion); the error is loss = erasure.
• Squeezed light mode (CV) squeezed — Continuous-variable modes for GKP/cluster states; needs ~10 dB *effective* squeezing for FT (0.62 dB on chip today); raw on-chip squeezing is a different quanti
• Gate-defined quantum-dot spin (Si/SiGe, SiMOS, Ge) qd_spin — Electron/hole spin in a CMOS-fabricated dot; exchange (10–100 MHz) gives ns–100 ns gates; readout via charge sensor.
• Donor spin (P in ²⁸Si) donor — STM-placed phosphorus atoms; nuclear qubits gated through a shared electron's hyperfine coupling; no foundry path.
• Colour-centre / defect spin (NV, SiV, SnV, T centre) defect — Optically addressable spin in a solid host; network node with spin–photon interface rather than a processor qubit.
• Majorana parity (InAs–Pb tetron) majorana — Gate-defined nanowire device; only single-wire parity readout demonstrated; topological protection contested.
• rf-SQUID flux qubit (annealer) fluxq — Analogue-Hamiltonian carrier; 4,400+ qubits with 20-way Zephyr coupling; not a gate-model qubit.

2 · Encoding

• Bare physical qubit (no encoding) enc_bare — Computational subspace of an anharmonic oscillator; leakage to |2⟩ is the price.
• Hyperfine / clock-state qubit enc_hf — Ground-state hyperfine (or nuclear-spin) levels of atoms and ions; second-to-hour coherence.
• Optical narrow-line qubit (ion S–D, atom clock) enc_opt — Qubit on an optical transition to a metastable level: the S₁/₂–D₅/₂ quadrupole line of an alkaline-earth ion (⁴⁰Ca⁺ at 729 nm, ⁸⁸Sr⁺ at 674 nm) or the ¹S₀–³P₀ c
• Ground–Rydberg analogue qubit enc_gr — The two computational states are a ground level and a Rydberg level: the native encoding of an analogue Rydberg machine, whose limit is the Rydberg lifetime (~1
• Metastable ('omg') erasure encoding enc_omg — Qubit in metastable manifold so that decay leaves the subspace detectably; ions (proposal) and Yb atoms (shown).
• Dual-rail (erasure) encoding enc_dualrail — One excitation in two modes/transmons/cavities; loss leaves the codespace → detected as erasure (80–90% of gate errors).
• Cat-code encoding (biased noise) enc_cat — Two-photon-dissipation-stabilised coherent states; bit-flips exponentially suppressed, phase-flips grow ∝ n̄.
• GKP grid encoding enc_gkp — Grid states in a bosonic (microwave cavity) or optical mode; corrects small shifts; needs ~10 dB squeezing.
• Exchange-only / singlet–triplet spin encoding enc_eo — Encoded spin qubits controlled purely by baseband exchange pulses; leakage 0.015%/Clifford (AEON).
• Time-bin / path photonic encoding enc_timebin — Photonic dual-rail in time or path; loss is the error and it is heralded.
• Fermion-parity encoding (tetron) enc_parity — Qubit in joint parity of two Majorana wires; X-measurement lifetime 1000× shorter than Z in 2025 data.
• Single-spin (Loss–DiVincenzo) / nuclear-spin encoding enc_spin_ld — Bare spin-1/2 of an electron, hole or nucleus; needs microwave/EDSR for 1Q.

3 · Gate mechanism

• Tunable-coupler CZ / iSWAP g_tc — Flux-tunable coupler switches ZZ/exchange on and off; 30–70 ns gates.
• Cross-resonance (fixed-frequency, all-microwave) g_cr — Microwave-only entangling gate on fixed-frequency transmons; 180–500 ns; superseded by tunable couplers in IBM Heron.
• Rydberg-blockade CZ g_ryd — Global Rydberg pulses entangle neighbouring atoms; 270 ns; fast for a natural carrier.
• Analogue Rydberg Hamiltonian evolution g_rydanalog — Global laser fields drive a programmable Ising/XY Hamiltonian on the whole array at once — no discrete entangling gate, no circuit; the geometry of the tweezer
• Mølmer–Sørensen / light-shift laser gate g_ms — Laser-driven spin-motion coupling; 70 µs (Helios) to 550–880 µs (IonQ Forte chains); 1.6 µs record.
• Electronic microwave gate (ions, laser-free: near-field or static gradient) g_elec — Chip currents create microwave gradients; no lasers for gates, no ground-state cooling; 120–226 µs.
• Exchange gate (spins; incl. shuttled-spin CZ) g_exch — Voltage-pulsed exchange J (10–90 MHz); 58–500 ns; ~80% of error is calibration/control (HRL).
• Linear-optical fusion (heralded) g_fusion — Probabilistic Bell measurement on photons (50%, 75% boosted); failure is heralded → erasure.
• Programmable linear-optical interferometer (no entangling primitive) g_lointer — A mesh of beamsplitters and phase shifters, or a time-multiplexed fibre loop, that scatters photons or squeezed light for sampling; it implements no entangling
• Ancilla-mediated bosonic gates (cat CX, dual-rail CZ, beamsplitter) g_bos — Transmon ancilla, or a differentially driven DC-SQUID beamsplitter coupler between modes; ~500 ns.
• Measurement-based Majorana gate g_mbq — Braiding by sequences of parity measurements — nothing demonstrated; single-wire readout only.
• Quantum annealing (analogue evolution) g_anneal — Coherent quenches 3.6–27 ns on 1,222–5,627 qubits; not a gate.
• CV Gaussian gates + GKP-assisted non-Gaussian operations g_cv — Beamsplitters, squeezers and homodyne feed-forward on optical modes at 1 MHz clock (Aurora).
• Microwave / optical spin gates (defect centres, 1Q spins) g_mwspin — ESR/EDSR-driven rotations and hyperfine-conditional gates; < 0.1% error on NV electron+nuclear registers.
• Bias-preserving cat–cat CNOT g_catcnot — Required by every cat-qubit resource estimate; proposed since 2019 (Guillaud–Mirrahimi) and 2020 (Puri et al.), the first unitary scheme in July 2026; no demons
• Multi-photon resource-state generator (RSG; 6-ring etc.) src_resource — Fusion-based FT needs 24–168-photon encoded resource states at high rate; largest fused deterministic-emitter graph states are 8 photons at < 1 Hz.

4 · Connectivity / transport

• Static nearest-neighbour lattice cx_nn — Heavy-hex (degree 2.3) or square lattice (degree 3.5–3.6) with on-chip couplers; ZZ crosstalk is the tax.
• Long-range on-chip couplers (c-couplers, mm-scale) cx_lr — Resonator/coupler links beyond NN; enables degree-6 qLDPC layouts; 2 mm CZ 99.81% (IQM); IBM Loon components without numbers.
• Ion shuttling (QCCD, junctions, grid traps) cx_qccd — Ions moved between zones at m/s with sub-quantum heating; transport + cooling dominate runtime (55 ms per full layer on Helios).
• Ion-chain motional bus (all-to-all in chain) cx_bus — Shared motional modes give all-to-all within one chain; gates slow with chain length (median 672 µs on Forte's 30-ion chain).
• Atom transport by AOD tweezers (zoned architecture) cx_aod — Coherence-preserving moves of 100s µm in 0.4–1.6 ms at ~99.95%; storage/entangling/readout zones.
• Continuous atom reloading (reservoir + optical conveyor belt) cx_reload — A continuously fed atom reservoir (optical-lattice conveyor belts, a cavity-enhanced lattice or a molasses-stopped beam) from which tweezers extract new atoms i
• Spin shuttling (conveyor mode) cx_shuttle — Spins carried 10 µm in < 200 ns at 99.5%; gates between moving spins 98.86%; gives fabricated carriers transport connectivity.
• Photonic switching / routing (electro-optic, feed-forward) cx_switch — Every switch costs loss: 0.19 dB/MZI at system scale (Aurora), 100 mdB for the best in-line element (PsiQuantum BTO); FT needs ~7 mdB.
• Crossbar shared-line control (spins) cx_crossbar — Shared plunger/barrier lines: 23 lines for 16 dots, T = 6√g − 1 scaling; no coherent qubit operation shown.

5 · Control

• Room-temperature electronics + per-qubit coax/flex ct_rt — One rf line per qubit through the fridge; the I/O wall: > 4,000 rf lines (KIDE), 1,121 qubits (Condor).
• Vertical (out-of-plane) signal delivery — VIO, coaxmon, 3D wiring ct_vio — Room-temperature electronics reach every qubit through wiring that leaves the chip plane — coaxial pins, through-substrate vias, a signal-routing interposer — i
• Cryo-CMOS controller (4 K / mK) ct_cryocmos — CMOS ASICs at 4 K (or 7 mK) generating pulses next to the qubits; 2–23 mW/qubit at 4 K; 20 nW/MHz per cell at mK.
• SFQ digital control (4 K / mK) ct_sfq — Single-flux-quantum (superconducting-electronics) digital circuits either at the mK stage — pulse trains drive qubits from a flip-chip (1Q > 99 %, 99.9 % peak;
• On-chip flux-DAC multiplexing ct_fluxdac — Annealer-heritage SFQ flux DACs: ~200–300 bias lines for 10⁴ qubits; applied to fluxonium (bump-bonded MCM).
• Laser + AOD/SLM optical control (atoms) ct_laser — SLM-generated tweezer arrays (12,000 sites), AOD moves, Rydberg lasers; 33 W for 18,225 tweezers; the binding constraint above 10⁴ sites is the AOD time–bandwid
• PIC-generated tweezers / integrated optics for atoms ct_pic_trap — Trap generation moved onto a photonic chip (4 atoms, 27.5 s lifetimes); fabricated-world optics for a natural carrier.
• Laser control with integrated photonics (ions) ct_ionlaser — In-vacuum integrated photonics at room temperature (cryogenic traps at 4–10 K exist): Waveguides in the trap deliver all wavelengths; > 99.3% 2Q (ETH). Helios (
• Free-space laser addressing of ions (AOM/AOD beams) ct_ionaod — Individual and global gate beams steered onto a chain by acousto-optic modulators or deflectors through bulk optics above the trap — the most common ion control
• Chip-integrated microwave control (ions) ct_ionmw — Electronic signal sources replace gate lasers: ~200 sources for 1,000 ions (WISE); 1Q 1.5×10⁻⁷ without shielding.
• Baseband electrical control (spins, Majorana) ct_base — Voltage pulses on gates; digital, dense; the calibration burden shows up as coherent error.
• Electro-optic drive + feed-forward electronics (room temperature) ct_eo — MHz–GHz switching of BTO/TFLN modulators conditioned on detector outcomes within one clock cycle.

6 · Readout

• Dispersive microwave readout (+TWPA, Purcell) ro_disp — Resonator shift read through a parametric amplifier; 240 ns pulse at 99.94% (IQM); readout-induced leakage is the hidden cost.
• Fast unconditional / dissipative qubit reset ro_reset — Returning a qubit to |0⟩ in ~100–500 ns by coupling it to a lossy resonator or by a measurement-free dissipative drive — a per-cycle cost term of every surface-
• Flux readout via QFP / SQUID shift registers (annealers) ro_fluxro — The annealer's own readout: the final flux state of each qubit is latched by a quantum-flux-parametron and clocked out through on-chip shift registers to a hand
• Fluorescence state detection (ions, defects) ro_fluor — State-dependent scattering counted by PMT/SNSPD; 11 µs at 99.93% record; SPAM 4.8×10⁻⁴ on Helios.
• Fluorescence imaging of atom arrays ro_img — Camera imaging, 0.5–1 ms typical; non-destructive with 0.24% loss; fast neutral-Yb imaging emerging (17.6 µs, 99.89%).
• Spin-to-charge conversion + rf reflectometry ro_s2c — Pauli/energy-selective tunnelling sensed by a gate-based single-electron box (SEB) or an SET; 99.2% in < 6 µs; 99.9% at 100 µs.
• Single-photon detection (SNSPD / TES) ro_spd — Destructive by nature; 93.4% wafer-scale median to 99.73% on a hero device, 3 ps jitter, 1–4 K; thousands of channels needed.
• Homodyne quadrature detection (CV) ro_homodyne — Measurement of a field quadrature against a shared local oscillator — the computational measurement of continuous-variable and GKP machines (Aurora); photon cou
• rf quantum-capacitance parity readout ro_qcap — Interferometric parity readout of a Majorana wire; 1% assignment error, SNR 1 in 3.6 µs.
• Mid-circuit erasure check ro_erasure — Ancilla-based detection of leakage/loss without disturbing the code: 384 ns (transmon), 1.8 µs (cavity), 20 µs (Yb atoms).

7 · Code

• Rotated surface code (+ yoked variants) code_surface — 2D nearest-neighbour code; threshold 0.94%; ~650–1,500 physical per logical at 10⁻¹² (yokes cut ~1,500 to 600–800).
• Colour code (transversal Cliffords) code_color — Triangular 2D code with transversal H/S/CNOT; ~1.9× the surface-code footprint; hosts distillation and injection.
• Non-local qLDPC codes (bivariate bicycle, 'gross') code_qldpc — The surface code is itself a qLDPC code with 2D-local checks and vanishing rate; this node is the non-local, constant-rate family. 'Gross' = IBM's [[144,12,12]]
• High-rate concatenated codes with transversal gates code_highrate — Iceberg [[k+2,k,2]], [[80,48,4]], tesseract colour code [[16,6,4]] (subsystem variant [[16,4,2,4]] at Quantinuum): many logical qubits per physical block; needs
• Erasure-adapted codes code_erasure — Surface/loss-tolerant codes fed with erasure flags: threshold 0.94% → 4.15%; dual-rail simulation Λ≈27 (idealised).
• Bosonic concatenation (repetition-cat, LDPC-cat, GKP+qLDPC) code_bosonic — Outer code exploits inner bias/shift correction; Ocelot's two points are not a distance scan; LDPC-cat assumes 0.1% phase-flip (measured ~10%).
• Fusion-based fault tolerance code_fusion — Resource states + fusions; loss threshold 2.7%/photon (boosted 6-ring), 17.4% only for a {7,4}-encoded resource state of ~168 photons; no demonstration.
• Algorithmic FT / transversal architectures code_aft — Constant syndrome rounds per logical gate via transversal gates + correlated decoding; amortises slow cycles (> 10× spacetime).
• Magic-state factory (cultivation / distillation / code switching) code_magic — Non-Clifford resource; cultivation reaches 10⁻⁹ T-error at ~cost of a CNOT (theory), 0.9999 shown at 8% acceptance.
• Error mitigation in the code slot (ZNE, PEC) — not a code code_mitig — Zero-noise extrapolation, probabilistic error cancellation and tensor-network post-processing: what actually occupies the code slot of a utility-scale NISQ mach
• Error-detection codes ([[4,2,2]], d=2 surface, iceberg, spacetime) code_detect — Codes that flag an error and discard the run instead of correcting it — the [[4,2,2]] code, the distance-2 surface code, spacetime codes: the first rung of the

8 · Decoder

• MWPM / Sparse Blossom (+ correlated matching) dec_mwpm — Matching decoder; < 1 µs/round at d=17 on one core (throughput), 0.8 µs FPGA latency at d=13 (Micro Blossom); Λ=2.04 on Willow vs 2.14 neural.
• Neural decoders (AlphaQubit 2, CNN, transformers) dec_nn — Learned decoders make ~6% fewer errors than a tensor-network decoder and ~30% fewer than correlated matching on Sycamore data, and 20–29% fewer than matching in
• Relay-BP for qLDPC (FPGA) dec_relaybp — Belief propagation with relays: 24 ns/iteration, average < 1 µs per cycle, ~100× lower LER than BP+OSD.
• FPGA real-time decoders (LCD, Deltaflow) dec_fpga — Local-clustering < 1 µs/round to d=17; Riverlane + Rigetti real-time loop 9.6 µs (6.5 µs decode + 3.1 µs communication, 2×2 stability patch, arXiv:2410.05202) v
• GPU decoding via NVQLink dec_gpu — 3.84 µs round trip; Helios+GH200 BP+OSD median 67 µs (5.4× LER gain); platform-agnostic layer.
• Correlated / loss-aware decoding (transversal, atom loss) dec_corr — Decodes across transversal gates and uses loss flags; 1.73× QEC gain on 448-atom processor.
• In-loop RL calibration / control steering dec_rl — Reinforcement learning tunes > 1,000 control parameters during QEC: ~20% extra suppression, 3.5× drift robustness.
• Cryogenic / on-chip decoder (cryo-CMOS, SFQ) dec_cryo — Designs only: NISQ+ (≤ 20 ns, SFQ), QECOOL (2.8 µW), Pinball/CryoZip (4 K predecoders) — no fabricated decoder chip.

9 · Interconnect

• Multi-chip modules / l-couplers (same cryostat) ic_mcm — Chiplet tiling (Rigetti 12×9 q, 99.1%) and metre-scale l-couplers between modules (IBM Cockatoo 2027 [R]).
• Multi-die packaging in one module (flip-chip, bump-bonded control die) ic_multidie — A qubit die flip-chipped to a wiring or control die inside one package — Sycamore, Zuchongzhi, Ocelot, the SFQ flip-chip module: the interconnect below the modu
• Cryogenic signal fan-out for spin arrays (router die, 3D stacked wiring) ic_fanout — A millikelvin router (Intel's Pando Tree fanning out to 64 channels), 3D stacked cryogenic wiring or on-die routing that takes a spin array from tens to thousan
• Cryogenic microwave link between refrigerators ic_cryolink — 30 m superconducting waveguide at < 50 mK; Bell fidelity 80.4% at 12.5 kHz; 0.55–0.65 dB total loss.
• Ion–photon link ic_ionphoton — Remote Bell pairs at 9.7–250 s⁻¹ with 94–97% fidelity; teleported CZ 86%; needs ≥ 10⁴ s⁻¹.
• Atom–photon cavity interface ic_atomcavity — Tweezer atoms coupled to cavities: ~90% generation-to-detection atom–photon efficiency (MPQ); no inter-module rates yet.
• Spin–photon solid-state link (SiV, NV, T centre) ic_spinphoton — Entanglement over 25–35 km deployed fibre at 0.02–1 Hz with F 0.53–0.86; two distinct inter-cryostat teleported-CNOT results — post-selected on T centres (Photo
• Microwave–optical transducer (useful efficiency) ic_transducer — Best: η_tot 15%, N_add 0.16 (EO); useful regime needs η > 1/2 and N_add ≪ 1; ~3 orders of magnitude short (IBM analysis).
• Fibre links between photonic modules ic_fibre — Chip-to-chip Bell 99.72% over 42 m (loss excluded); Aurora inter-rack fibre.

10 · Manufacturing

• 300 mm CMOS foundry (spins, cryo-CMOS, superconducting wiring) fab_cmos — Intel D1 EUV (24k devices/wafer, 96% tune-up), imec (2Q > 99%), GF 22FDX (1,024 dots), ST FD-SOI; GF now a multi-modality foundry ($375 M LOI).
• Superconducting-qubit lithography (Al/AlOx/Al junctions, Nb wiring, 300 mm) fab_sc — IBM Albany 300 mm / Anderon foundry; Google Santa Barbara; Rigetti ABAA trimming (97.4% frequency-targeting success); OQC 500-qubit wafer-scale package.
• 3D machined superconducting cavities fab_3d — Machined aluminium cavities reach Q > 0.5×10⁹ (10 ms photon lifetime, Yale 2013); niobium accelerator cavities with no qubit attached hold photons for 0.5–2 s a
• Ion-trap fabrication (surface-electrode chips; machined 3D blade traps) fab_trap — MEMS/CMOS-fab traps: Helios 1,228 electrodes / 273 signals; Oxford Ionics chips from standard semiconductor fabs; SkyWater in-house at IonQ.
• Photonic IC foundry (SiN, BTO, TFLN, SNSPD on 300 mm) fab_pic — GlobalFoundries Fab 8 for PsiQuantum (SiN 0.5 dB/m multimode but 1.8(2) dB/m single-mode, BTO switches, on-chip SNSPD at 93.4% median); TFLN/SiN for Xanadu; als
• Optical / mechanical assembly (lasers, vacuum, objectives) fab_optics — Room-temperature systems: Pasqal Orion 3 kW / 2,500 kg; continuous reload hardware (2 optical-lattice conveyors, 300,000 atoms/s into tweezers → 30,000 initiali
• Bulk-optics and fibre assembly (photonic samplers) fab_bulk — Table-top optics and fibre loops assembled by hand — Jiuzhang, Borealis, PT-2 — as opposed to a photonic-IC foundry; it is how every beyond-classical sampler wa
• III-V MBE heterostructures (InAs–Pb wires, QD sources) fab_mbe — In-house Microsoft growth for tetrons (no external replication); GaAs quantum-dot photon sources (Quandela).
• STM hydrogen-resist lithography (donors) fab_stm — Atom-precise placement of P donors (~3 nm); bespoke, serial, no foundry route.
• Diamond growth / implantation (NV, SiV, SnV) fab_diamond — SnV nanocavities (QuTech, PRX 2026): 327 characterised at room temperature on two chips, coupled centres in 7 of 9 cooled, two measured above cooperativity one