How researchers get access — queue, allocation, pricing posture

Who actually gets cycles, and when

Ireland Quantum 100 is a sovereign 100-physical-qubit superconducting transmon machine being commissioned in Co. Tipperary. The delivery window runs Q3 2026 site fit-out, Q4 2026 cryostat install, Q1 2027 first-light single-qubit characterisation, and Q2 2027 multi-qubit availability with the first external workloads on the system. That schedule dictates everything about quantum access in Ireland: there is no general-purpose queue in 2026, and there will not be one until single- and two-qubit gate calibration is stable, readout is characterised across the full chip, and we have a credible noise model for users to compile against.

When external access opens in Q2 2027, the cohort is climate-first by design. That is not a marketing line — it is how the allocation is structured. The first wave of quantum researcher access is reserved for workloads where a 100-qubit, NISQ-era transmon machine can plausibly contribute scientific value: variational quantum eigensolver (VQE) work on small-molecule catalysts relevant to carbon capture, photovoltaic absorber screening, lithium- and sodium-ion cathode chemistries, climate-relevant protein folding fragments, and grid-optimisation problems that decompose cleanly to QAOA.

The allocation model

Ireland Quantum allocation is split into three tracks. The proportions will be tuned as the system stabilises, but the structure is fixed:

• Climate priority track — the majority of shot budget. Open to academic groups, national labs, and industrial R&D teams running workloads with a defensible climate-science thesis. Allocation is by panel review, not first-come-first-served. Proposals are scored on scientific merit, fit to a 100-qubit transmon device with heavy-hex connectivity, and whether the team can actually use the output (i.e. they have classical pre/post-processing in place).
• Calibration and benchmarking track — reserved time for randomised benchmarking, cycle benchmarking, mirror circuits, quantum volume runs, and surface-code distance-3 and distance-5 experiments as the error-correction roadmap progresses. This is partly internal, partly open to research groups working on characterisation methods.
• Sovereign and educational track — Irish university groups, Tyndall-adjacent research, and a small teaching allocation. The point is to make sure Irish-domiciled researchers do not have to apply through a foreign cloud provider to touch hardware sitting in Tipperary.

There is no pay-to-skip lane. Industrial users on the climate track are reviewed against the same scientific criteria as academic groups. They will pay; that is separate from queue position.

Queue mechanics and the realities of a transmon device

People who have only used cloud-hosted quantum backends sometimes underestimate how much the device's physical state dictates the queue. A superconducting transmon chip running at sub-15 mK in a dilution refrigerator drifts. Qubit frequencies move. Two-qubit gate fidelities degrade between calibration windows. T1 and T2 vary qubit-to-qubit and day-to-day. The queue is not a simple FIFO over jobs — it is a scheduler that has to interleave:

• Automated calibration passes (single-qubit randomised benchmarking, two-qubit cross-resonance or tunable-coupler tune-up depending on the gate set, readout discrimination retraining)
• User circuits compiled against the most recent calibration snapshot
• Periodic full-chip characterisation that invalidates in-flight compilations and forces recompilation

Practically, a user submitting an OpenQASM 3 circuit through Qiskit, PennyLane, or Cirq via our access layer gets a job ID, a position in the climate-track queue, and a calibration-snapshot identifier. If their circuit was transpiled against snapshot N and the device has moved to snapshot N+2 by the time it reaches the front of the queue, the system either re-transpiles automatically (the default for variational workloads) or holds and notifies (the default for benchmarking work where reproducibility matters more than throughput).

Pricing posture

We are not publishing a price list on this page, and I am not going to invent one. What I will commit to is the posture:

• Academic climate work is subsidised. If your group is doing genuine carbon-capture catalyst VQE or photovoltaic discovery and you have peer-reviewed output as a credible endpoint, the price should not be the thing that stops you.
• Industrial climate work is priced to recover cost, not to extract maximum margin during the early-access period. The system is sovereign infrastructure; it is not a hyperscaler SKU.
• Non-climate industrial work — finance, logistics outside grid optimisation, generic optimisation demos — is deprioritised in 2027 and priced to reflect that it is consuming shot budget that could go to a climate workload. This is deliberate.
• Pricing is per-shot and per-circuit-depth-class, not per wall-clock hour. A shallow 20-qubit VQE ansatz and a deep 80-qubit QAOA do not cost the same, and we will not pretend they do.

What you need to bring

Quantum researcher access works when the user arrives with a workload that fits the hardware. For a 100-qubit transmon with heavy-hex topology, that means: a problem that maps onto nearest-neighbour connectivity without catastrophic SWAP overhead, a circuit depth that respects the coherence budget (rough rule: keep two-qubit gate count per qubit below the T2-over-gate-time ratio with margin), and a classical wrapper that can absorb shot noise. VQE and QAOA fit. Hamiltonian simulation of small chemistry systems fits. Shor's algorithm at any cryptographically interesting scale does not, and will not on this generation of device — please do not apply with that.

The application itself asks for the Hamiltonian or cost function, the ansatz, an estimate of shots required, the classical optimiser, and what success looks like. We would rather have a tight 12-qubit problem that produces a publishable result than a hand-wavy 90-qubit demo.

Next steps

Through the rest of 2026 we are publishing the calibration-snapshot format, the OpenQASM 3 dialect notes, the heavy-hex coupling map, and the first version of the access application. Climate-priority track applications open ahead of Q2 2027 multi-qubit availability so that the first cohort is queued and compiled before the device is ready,