Results on Real Hardware, Reported Against a Baseline

We have run optimization and machine learning problems on annealing, gate-model and entropy quantum hardware, and on quantum-inspired methods. Each result names the baseline it was measured against and what was assumed.

Earth observation satellite schedulingD-Wave quantum annealing
23.5% more high-priority targets, 39% faster than a classical solver (Gurobi)

Every constraint met in all 100 test cases of 1,000 targets. The classical solver (Gurobi) met every constraint in 68 of them and simulated annealing in none.

Baselines: a classical solver (Gurobi) and simulated annealing on the same instances. Peer-reviewed, IEEE TQCEBT 2024.

Battery storage siting, IEEE 118-busIBM Heron, hybrid gate-model
22% lower cost at 75 qubits

Run on quantum hardware from 24 to 156 qubits. The edge peaks at 75 qubits and narrows as noise grows. At smaller sizes on a simulator, the hybrid method was 1.84x to 4.6x faster than the same classical search started at random.

Baseline: the same classical search without the quantum step. At 75 qubits the hybrid run finished within 16.5% of the MILP optimum, against 49.3% without the quantum step.

Microgrid resilience dispatch, 14-bus systemQCi Dirac-3 entropy computing
1.4% maximum customers unserved across 11 outage scenarios

All four microgrids self-sustaining.

Baseline: piecewise MILP (HiGHS) on the same instance. Hardware runs, July 2026.

Hyperspectral image labelingIBM Brisbane, quantum SVM
90% accuracy from about 20 reference pixels per class

Classes include submerged seagrass and aircraft on the ground. Accuracy rose from 60% to 90% on hardware.

Hardware figures only; simulator figures are reported separately. Peer-reviewed, SPIE Defense + Commercial Sensing 2025.

Drone swarm planning with obstaclesQuantum-inspired
30 to 55% better than the classical method

Baseline: the classical algorithm on the same scenarios, with obstacles added. Delivered project.

Workforce travel, traffic management company
About 8 miles saved per employee per day

Across roughly 2,500 employees. Baseline: the client's existing home-to-job assignment.

Publications

Earlier Papers and White Papers

Patents: three filed, covering the problem-to-solver translation layer, hyperspectral labeling and grid simulation architecture.

Get Started with Planck

Optimized decisions that move the business forward, powered by quantum and AI.

Talk to Planck
Talk to Planck Book 15 minutes with the founder