Climate applications — the workloads the machine is built for

Why the machine is sized for climate, not for benchmarks

Most quantum hardware roadmaps are written against abstract benchmarks — quantum volume, randomised benchmarking, cross-entropy. Ireland Quantum 100 is being built against a different question: what is the smallest physically-honest machine that can produce useful intermediates for climate science before fault-tolerance arrives? A 100-qubit superconducting transmon array, in a heavy-hex topology, sitting at sub-15 mK in a dilution refrigerator, is not going to factor RSA-2048 or simulate a full enzyme active-site to chemical accuracy. It is, however, large enough to run variational quantum eigensolver (VQE) workloads on small-to-mid molecules, quantum approximate optimisation (QAOA) on grid and routing problems, and hybrid kernel methods on climate-finance data — all of which are real bottlenecks in current climate workflows. That is the design target. Quantum climate computing, framed honestly, is about getting classical-quantum hybrid pipelines into the hands of Irish and European researchers a year or two earlier than they would otherwise see them.

Carbon-capture chemistry: small molecules, large impact

The first priority workload is direct-air-capture and point-source-capture sorbent chemistry. The relevant physics is the binding energy of CO₂ to amine groups, metal-organic frameworks (MOFs), and emerging covalent-organic-framework (COF) candidates. Classical density functional theory (DFT) gets you a useful starting point but systematically mishandles strong electron correlation in transition-metal binding sites — exactly where the most interesting MOF candidates live (Cu-, Fe-, Mg-centred frameworks).

On a 100-qubit transmon machine, the realistic play is active-space VQE: hand the bulk of the molecule to classical CCSD(T), carve out the 8–20 strongly-correlated orbitals around the binding site, and solve that active space on the quantum device using a unitary coupled-cluster ansatz (UCCSD) or a hardware-efficient ansatz with symmetry-preserving entanglers. With heavy-hex connectivity the natural mapping is Jordan-Wigner with parity-check reduction, or Bravyi-Kitaev where qubit count becomes the binding constraint. The output is not a publication-ready energy — it is a corrected correlation energy that feeds back into the classical embedding. That is how quantum climate research actually delivers value in the pre-fault-tolerant era.

Photovoltaic discovery and battery materials

The second cohort is materials. Perovskite stability under humidity and UV is, at root, an excited-state problem: defect-mediated charge recombination is governed by spin-orbit-coupled excited-state surfaces that classical TDDFT handles poorly. Quantum subspace expansion methods on top of a VQE ground state give access to low-lying excited states without needing the full quantum-phase-estimation circuits that 100 qubits cannot yet support with adequate depth.

For battery materials, the interesting chemistry is in cathode redox couples — Li-rich layered oxides, Na-ion polyanionic frameworks, and the solid-electrolyte-interphase reactions that determine cycle life. The quantum contribution is again narrow and specific: correlated treatment of the transition-metal d-manifold during charge/discharge. A useful rule of thumb for the Ireland Quantum 100 envelope: any active space we can express in fewer than ~40 logical qubits, with circuit depth bounded by what error-mitigation (zero-noise extrapolation, probabilistic error cancellation, Clifford data regression) can recover from a ~10⁻³ two-qubit gate-error regime, is in scope. Anything larger is research-grade only — and we will say so.

Grid optimisation and climate finance

Not every climate workload is chemistry. Ireland's grid is moving towards 80% renewables by 2030 with growing constraint-management complexity: when wind generation in the west exceeds interconnector capacity, the operator pays to curtail. The underlying mathematical problem — unit commitment with stochastic renewable inputs and N-1 contingency constraints — is a mixed-integer programme that grows badly with grid size.

QAOA and quantum-annealing-inspired warm-starts are not magic; the published results so far show modest-to-no advantage over good classical heuristics on most instances. But the operator-level question is whether a hybrid solver can get within tolerance faster than CPLEX or Gurobi on specific instance classes — chiefly highly-symmetric or highly-degenerate instances where classical branch-and-bound stalls. That is an empirical question and the machine exists, in part, to answer it on real Irish grid data rather than on synthetic benchmarks.

Climate finance — pricing physical-risk exposure across a portfolio under correlated climate scenarios — is structurally similar: a high-dimensional sampling problem where quantum-enhanced Monte Carlo, via amplitude estimation, has a theoretical quadratic speed-up. At 100 physical qubits without fault tolerance the speed-up is not realised in practice. What is realised is a credible parallel-development environment so that when fault tolerance arrives — Ireland Quantum's surface-code roadmap targets first logical qubits in the 1,000-physical-qubit successor machine — the algorithms are already in production.

Climate-relevant protein folding and biological carbon

The fourth workload class is biological: carbonic anhydrase variants for engineered CO₂ hydration, RuBisCO efficiency mutations for crop carbon-fixation, and methane monooxygenase for atmospheric methane drawdown. These are large proteins; full-structure quantum simulation is decades away. The honest near-term contribution of quantum climate science is to the active site — the metal cofactor and its first coordination sphere — embedded in a classical QM/MM scheme. Zinc in carbonic anhydrase, magnesium in RuBisCO, the di-iron centre in MMO: each is a ~20-30 orbital active-space problem that is right at the edge of what a 100-qubit transmon machine with aggressive error mitigation can address.

The IMPT loop: from quantum result to offset stack

This is where the project closes the loop. IMPT.io's offset stack already evaluates carbon-removal supplier candidates against durability, additionality, and cost-per-tonne. Sorbent and biological candidates that come out of Ireland Quantum 100's chemistry workloads — improved MOF binding energies, more stable amine variants, better-characterised enzymatic pathways — feed directly into that supplier evaluation as physically-grounded performance priors rather than vendor-supplied marketing numbers. That is the integration argument for climate qu