The 12-month delivery plan — month-by-month

Why a twelve-month window is the right honest number

The Ireland Quantum 100 plan is not a research-grade desktop dilution fridge being plugged in by a postdoc. It is a 100-physical-qubit superconducting transmon system installed in a purpose-fitted shielded hall in Co. Tipperary, with a documented path from a cold empty room to a multi-qubit machine accepting external climate-science workloads. Twelve months is what that actually takes when you are honest about civils, cryogenics commissioning, RF tune-up, and the long tail of two-qubit gate calibration. Anything shorter is marketing. Anything longer means you have not committed to a design freeze.

This page walks the four quarters of the ireland quantum plan in the order the work has to happen, because in superconducting hardware you cannot parallelise away the physics. The dilution refrigerator must be cold and stable before you mount a chip. The chip must be characterised at single-qubit level before you tune two-qubit gates. And you cannot run a useful chemistry circuit until your readout error and crosstalk budget are bounded.

Q3 2026 — Site fit-out, civils, and the cryogenic envelope

The first three months are the least glamorous and the most unforgiving. The Annerpark facility in Clonmel is being prepared as a Class-controlled environment with three coupled engineering problems running concurrently: vibration isolation, electromagnetic shielding, and thermal-load management for the compressor plant.

The cryostat itself will sit on a seismic mass decoupled from the building slab. A dilution refrigerator targeting a base temperature below 15 mK is exquisitely sensitive to mechanical noise in the 1–100 Hz band — pulse-tube cryocooler vibration alone is enough to dephase qubits if it couples into the mixing chamber stage. The civils work in this quarter is therefore not generic data-centre fit-out; it is a foundation-level decision about how the pulse-tube heads are mounted relative to the sample stage.

Running in parallel:

• Mu-metal and aluminium magnetic shielding design, targeting sub-microtesla residual fields at the chip plane to protect Josephson junction coherence.
• RF cable plant — coax runs from room-temperature electronics down through the 4 K, still, cold-plate, and mixing-chamber stages, with attenuation and filtering budgeted per stage to suppress thermal photons on the drive lines.
• Helium recovery and compressor siting, because uncaptured helium boil-off is both an operational cost and an environmental one we are not willing to wear on a climate-priority machine.
• Power conditioning and a UPS regime sized for an orderly cryostat warm-up, not just a server reboot.

Q4 2026 — Cryostat install, wiring, and first cooldown

This is the quarter where the dilution refrigerator is mounted, plumbed, and brought to base temperature for the first time, with a dummy load before any quantum chip is risked. Initial cooldown of a large-format dilution fridge typically takes 36–72 hours from room temperature to base, and the first run is a diagnostic exercise: confirming base temperature stability, checking that every thermometer on every plate reads sensibly, and quantifying the standing heat load on the mixing chamber.

Wiring up a 100-qubit system is a non-trivial integer problem. Each transmon needs at least one drive line, the readout resonators are multiplexed per feedline, and flux-tunable couplers add further bias lines. You are looking at several hundred coaxial lines threaded through the temperature stages with calibrated attenuators, infrared filters, and circulators on the output chain to protect the quantum-limited amplifiers. Every line is a potential heat leak and a potential noise channel. The wiring plan is fixed in this quarter; revisiting it later means a full warm-up cycle, and a warm-up plus cooldown costs you a working week.

The chip package itself — heavy-hex topology, with the connectivity pattern IBM popularised because it suppresses frequency-collision and spectator errors in surface-code-friendly layouts — is mounted at the end of the quarter. Heavy-hex is a deliberate choice: lower average connectivity than a square lattice, but a far cleaner path to fault-tolerant surface-code embedding once we extend the device generation.

Q1 2027 — First light, single-qubit characterisation, and the calibration stack

"First light" on a superconducting machine means the first coherent Rabi oscillation observed on a single qubit. It is the moment the project transitions from a cryogenics build to a quantum computer. The work in this quarter is methodical:

• Resonator spectroscopy to find each readout resonator and confirm the dispersive shift.
• Qubit spectroscopy, T1 (energy relaxation) and T2 (dephasing) measurements across the device, with the working target for transmons in this generation being T1 and T2 in the high tens to low hundreds of microseconds.
• Single-qubit randomised benchmarking to drive gate error below the 10⁻³ threshold across the lattice.
• Readout calibration, including discrimination of the |0⟩ and |1⟩ states with assignment fidelity high enough that mid-circuit measurement is meaningful.

By the end of Q1 2027 the control stack — arbitrary waveform generators, the FPGA-based control system, and the calibration software that automates the above — must be running unattended overnight calibration cycles. A 100-qubit machine cannot be hand-tuned. The calibration graph itself is software infrastructure, and it is being built in parallel through the earlier quarters so it is ready when the chip is.

Q2 2027 — Two-qubit gates, multi-qubit circuits, and customer access

The final quarter of the twelve-month quantum facility delivery is where the device becomes useful. Two-qubit gate calibration on a heavy-hex lattice — typically a CZ or echoed cross-resonance variant depending on the coupler design — is an iterative optimisation against leakage, ZZ crosstalk, and spectator errors on neighbouring qubits. The realistic working target is two-qubit gate fidelity in the 99%+ range for the best edges, with a known distribution across the device that we will publish honestly rather than quoting only the best pair.

Once multi-qubit circuits run reliably, access opens to the priority cohort. The first workloads are climate-relevant variational chemistry: VQE on small active spaces for carbon-capture amine and metal-organic-framework candidates, qubit-efficient encodings for photovoltaic absorber screening, and ansatz exploration for battery cathode materials. These are the exact circuits where a 100-qubit nois