Industries
Nine industries, each written for its buyer: the decision they make today, five to eight applications with their status, and the measured results under the applications they prove.
1 · Semiconductors & advanced materials
Written for Process-integration and materials R&D leads at a fab or a materials supplier: the photolithography, precursor and packaging-materials groups.
The decision Which formulation or process variant goes to the next wafer lot.
- EUV photoresist candidate ranking Live
- Precursor and dopant chemistry (ALD/CVD precursors, dopant activation) Ready
- Low-k and high-k dielectric and interconnect materials Ready
- Defect and yield analysis across a tool chain Mapped
- Fab scheduling and lot dispatch Platform network
- Lithography process-window exploration Mapped
2 · Energy & utilities
Written for Transmission and distribution planning, system operations and asset management at a TSO, DSO or utility.
The decision Which reinforcement to commit capital to, which units to commit, how the network behaves when a line trips.
- Transmission expansion planning Live
- Contingency and cascading-failure analysis Live
- Unit commitment and economic dispatch Platform network
- Distribution-grid reconfiguration Platform network
- Renewable integration and storage siting Mapped
- Battery, catalyst and capture chemistry Ready
3 · Chemicals & process manufacturing
Written for Plant scheduling, supply-chain planning and formulation R&D at a specialty or bulk chemicals maker.
The decision Which production plan to run this shift; which formulation to make next.
- Production scheduling across grades and changeovers Platform network
- Nesting and job-shop scheduling (sheet, batch, multi-machine) Platform network
- Formulation and additive candidate ranking Live
- Catalyst and process-chemistry screening Ready
- Revenue management and order acceptance Platform network
- Predictive maintenance on cable and rotating assets Platform network
4 · Mobility & automotive
Written for ADAS and autonomy safety and validation, traffic operations, fleet and logistics planning, materials for batteries and lightweighting.
The decision Can this controller be released; how to time signals across a city; how to route a fleet.
- Autonomous-driving safety validation Live
- Traffic-signal coordination Platform network
- Vehicle routing and fleet assignment, including drone delivery Platform network
- Rail timetable recovery and rolling-stock rotation Platform network
- Battery and lightweight-materials candidate ranking Live
- CFD acceleration for aerodynamics Roadmap
5 · Telecom
Written for RAN engineering and network planning at a mobile operator; security architecture for QKD-ready links.
The decision How to get more capacity from spectrum already owned; where to place QKD hardware; which traffic is anomalous.
- Carrier-aggregation optimisation Platform network
- Optical network routing and wavelength assignment Platform network
- QKD hardware placement Platform network
- Network traffic anomaly detection Platform network
- Where throughput is lost across the whole network Mapped
- Customer segmentation from CRM data Platform network
6 · Pharma & life sciences
Written for Discovery chemistry, structural biology and formulation science: long-acting injectables and excipients.
The decision Which pocket to target before the assay campaign; which carrier or excipient to take to the bench.
- Allosteric site prediction from the structure file Live
- Formulation and excipient candidate ranking Live
- Drug-candidate and fragment energetics Ready
- Protein–ligand interaction energetics Ready
- Clinical and hospital shift scheduling Platform network
- Medical-imaging classification Platform network
7 · Finance
Written for Financial-crime and fraud analytics, market and portfolio risk, treasury.
The decision Which entities move together before a loss lands; which hedging or index-tracking book to hold.
- Correlated-risk and fraud-ring detection Live
- Credit-card fraud and anomaly detection Platform network
- Index tracking and cardinality-constrained selection Platform network
- FX hedging optimisation Platform network
- Derivative pricing and portfolio risk simulation Platform network
- Regulatory-capital optimisation Platform network
8 · Security & critical infrastructure
Written for The CISO office, the crypto-agility programme, resilience and continuity planning.
The decision Which cipher to migrate first and by when; which assets fail together under attack.
- Post-quantum migration planning Assessment
- Correlated-failure and blast-radius analysis Mapped
- QKD network design Platform network
- Network-traffic anomaly detection Platform network
- Water and utility anomaly detection Platform network
- Key-size assurance Assessment
9 · Aerospace & sovereign programmes
Written for Aerodynamics and mission-engineering groups; national quantum programmes that own a 5 to 25-qubit machine.
The decision When quantum CFD pays off; what to run on the installed machine now; how to schedule traffic, gates and missions.
- CFD acceleration roadmap Roadmap
- Air traffic flow management and multi-aircraft routing Platform network
- Airline gate allocation and mission scheduling Platform network
- Satellite placement and launch-vehicle trajectory Mapped
- Device characterisation for the installed machine Live
- Materials for airframes and propulsion Live
Proof points
One measured result per industry, under the application it proves, with its baseline and the platform it ran on.
1 · Semiconductors & advanced materials
Hardware result · award, 2nd place
EUV photoresist candidate screening
4/4 ranked correctly
candidate formulations ranked in the correct order on a quantum processor, energy differences within 0.3% of reference
Decision Which photoresist formulation goes to wafer testing
What is modelled An industrial tin-oxo photoresist cluster and its counter-anion variants, as a 40-qubit quantum chemistry model.
Operational outcome Ranked candidate formulations before synthesis · Fewer wafer runs on candidates that will not win · Model training in minutes instead of 173 GPU-hours
Problem size 40-qubit chemistry model · ranking run on 156 qubits · 120 qubits in a single job
Platform IBM Heron, 156 qubits
Baseline Exact reference energies for every candidate
Published reference Published device runs on the same photoresist chemistry: 14 and 22 qubits. arXiv:2607.23988, 2026
Merlin 4/4 candidates ranked correctly on hardware · 40-qubit chemistry model · training 173 GPU-hours → minutes
Baseline Exact reference energies; the published method’s training cost
2 · Energy & utilities
Research result
Transmission expansion planning
1,048,576 plans · optimum proven
the chosen expansion plan proven optimal over every one of 2^20 candidate build plans
Decision Which network reinforcement plan to commit capital to
What is modelled 20 candidate reinforcements, so 2^20 build plans, each with the network loading it produces.
Operational outcome A proven-optimal plan, not a sampled one · The cost of a reduced-order model quantified: 9% worse on the same QPU · Network response simulated to 8 time steps on 156 qubits
Problem size 20 binary investment decisions · 1,048,576 plans · 156-qubit network response
Platform CPU search; network response on a 156-qubit processor and a photonic QPU
Baseline Exhaustive classical search over all plans
Published reference Unit commitment at 25,755 variables and 48,939 constraints ran feasibly on a photonic QPU, with a commercial MILP solver ahead at short runtimes; generation scheduling at 26 generators × 24 periods on 36 trapped-ion qubits, no optimality proof. arXiv:2505.00145; platform-network results, 2025–26
Merlin Optimum proven over 1,048,576 plans · reduced-order model 9% worse on the same QPU
Baseline Exhaustive classical search; the full model on the same machine
3 · Chemicals & process manufacturing
Customer deployment · platform network
Production scheduling
6 h → 1 min
production schedule across 100+ product grades, recalculated in 1–10 minutes instead of 2–6 hours, in daily use
Decision Which production plan to run this shift
What is modelled 100+ product grades with changeovers, due dates and capacity, as a constrained scheduling model.
Operational outcome Re-planning on demand when conditions change · Verification included in the run time · In daily production use
Problem size 100+ grades · one plant · minutes per re-plan
Platform Classical optimisation solvers, delivered through our platform network
Baseline The plant’s prior manual and spreadsheet planning
Published reference No published production-scheduling result on quantum hardware at industrial scale; visible results are challenge prototypes. arXiv:2411.10303
Merlin 6 h → 1 min across 100+ grades, in production
Baseline The plant’s prior manual planning
4 · Mobility & automotive
Hardware result · customer proof of concept
Autonomous-driving safety validation · traffic signal coordination
95.6% safe episodes
with a disturbance model measured on a 128-qubit processor, against 39.5% with the standard white-noise model, same controller
Decision Can the autonomy controller be released
What is modelled The disturbance the controller faces, as a 128-qubit model; the controller evaluated in closed-loop simulation.
Operational outcome +56 points in safe-episode rate at the same operating point · Evidence a safety reviewer can sign off · City signal coordination: 20% less waiting time (proof of concept, simulation)
Problem size 128-qubit disturbance model · one controller · city-scale signal network
Platform A 128-qubit processor; closed-loop evaluation in simulation
Baseline White-noise disturbance model, same controller, same operating point
Published reference Nonlinear flow on superconducting hardware reaches 15 time steps; vehicle routing on device reaches 3 nodes. IBM Heron R3, 2026; arXiv:2504.01567; arXiv:2509.11469
Merlin 95.6% vs 39.5% safe episodes · 20% less waiting at signals (PoC)
Baseline White-noise disturbance model; conventional signal timing
5 · Telecom
Customer proof of concept · platform network
Carrier aggregation optimisation
+10% downlink
average downlink speed on a live network, no added hardware; secondary-cell data volume up to +50%
Decision How to raise capacity on spectrum already owned
What is modelled Carrier-aggregation settings across the cells of a live network, as a combinatorial optimisation.
Operational outcome +10% average downlink speed · No new radio hardware · Measured on the live network, not in simulation
Problem size Live network · cell-level settings · one operator
Platform Quantum-inspired solver, on the network operator’s live network, delivered through our platform network
Baseline The network operator’s existing carrier-aggregation settings
Published reference Published telecom figures in the field state a percentage with no baseline and no date. public vendor and operator announcements, undated
Merlin +10% average downlink on a live network, no added hardware
Baseline The network operator’s existing settings
6 · Pharma & life sciences
Research result
Allosteric site prediction
AUC 0.89
on an independent biological benchmark, for allosteric sites read from the structure file alone; site energetics within 1.4% of reference
Decision Which pocket to target before the assay
What is modelled Site energetics of the protein, from the structure file alone, with no assay data.
Operational outcome Ranked candidate sites before the assay · Site energetics within 1.4% of reference · Fewer assays on sites that will not bind
Problem size Whole-protein structure files · per-site energetics
Platform Quantum hardware plus simulation
Baseline Exact reference energies; an independent biological benchmark set
Published reference Peptides of 14–16 residues folded on 46–61 trapped-ion qubits reach the classical reference energy on some instances. arXiv:2604.26861
Merlin AUC 0.89 · site energetics within 1.4% of reference
Baseline Exact references; an independent biological benchmark
7 · Finance
Research result
Correlated-risk detection (fraud rings)
72% hidden exposure
of the correlated exposure in a 14-entity ring is invisible to a pairwise risk model, measured on a 14-qubit model of the ring
Decision Which entities move together before a loss lands
What is modelled 14 entities and the links between them, as a 14-qubit model whose joint behaviour is measured directly.
Operational outcome Hidden-correlation detection beyond pairwise models · Ranked exposure across the ring · A method ready for a pilot on client transaction data
Problem size 14 entities · 14 qubits · exact references
Platform Quantum hardware plus simulation, checked against exact references
Baseline A pairwise correlation model of the same ring
Published reference Portfolio optimisation benchmark instances are solved to proven optimality by MIP solvers in seconds; the one notable hardware result (corporate-debt fills, 2025) is unreproduced. arXiv:2509.17876
Merlin 72% of correlated exposure invisible to pairwise models, measured
Baseline A pairwise correlation model of the same ring
8 · Security & critical infrastructure
Readiness assessment
Post-quantum migration planning
2030–31
central estimate of the risk window per cipher, range 2028–34: a dated window to budget the migration against
Decision Which cipher to migrate first, and by when
What is modelled Per-cipher break timelines from published resource estimates, and the dependency graph of the assets that rely on each cipher.
Operational outcome A dated risk window per cipher · A migration order across the estate · A budget timeline the board can plan against
Problem size Every cipher in the estate · the asset dependency graph
Platform Our dated assessment, not a measurement
Baseline Published resource estimates and error-rate requirements
Published reference Lattice attacks on hardware run at dimension 7 against classical sieving beyond 180; the 2024 polynomial-time claim was withdrawn by its author. arXiv:2006.14057; arXiv:2604.27153
Merlin A dated window per cipher, central 2030–31
Baseline Published resource estimates
9 · Aerospace & sovereign programmes
Roadmap
CFD acceleration roadmap · on-premise quantum
~93 logical qubits
the projected crossover for the flow solver, from a measured cost law; the same circuits run at fixed depth on 5–25-qubit national machines today
Decision When quantum CFD pays off, and what runs on the installed machine today
What is modelled The cost of the flow-transport solver against its classical equivalent, as a measured scaling law.
Operational outcome A dated crossover for quantum CFD · Fixed circuit depth of three entangling layers from 5 to 156 qubits · Runs on the 5–25-qubit machine a national programme already owns
Problem size 5–156 qubits · 7 quantum platforms · heavy compute inside the customer perimeter
Platform A forward projection; the circuits measured on seven quantum platforms
Baseline The measured cost law of the classical solver; the installed machine’s size
Published reference Most application methods need about 100 physical qubits before returning anything; national programmes own 5–25-qubit machines. published resource requirements across the application layer
Merlin Fixed-depth circuits on the installed machine · crossover projected at ~93 logical qubits
Baseline The measured cost law; the installed machine’s size
The detail, sector by sector
Fifteen sectors, one decision shape: a resist screen, a grid plan, a berth schedule, a protein, a payment graph are each a coupled system read for what happens when one part moves. Here is where that lands sector by sector, and honestly how far each has been taken.
5
sectors with a result already run on quantum hardware
15
sectors mapped to a quantum use case
7
quantum hardware platforms our methods run on
Beyond the work we run ourselves, the home page collects results delivered through Merlin’s platform network — chemicals production scheduling, city traffic-signal coordination and telecom carrier-aggregation tuning, each under its industry with its published conditions.
Walkthroughs
Nine guided walks, each on its own page, each ending with what to send us. They run in the browser and need no key.
network planners and utilities
Grid and critical infrastructure
7 steps
terminal and fleet planning
Ports, maritime and logistics
7 steps
security and risk programmes
Cybersecurity and resilience
8 steps
programme offices with a machine installed
National quantum programmes
7 steps
formulation and excipient selection
Formulation screening
8 steps
materials and semiconductor design
Materials and resist screening
8 steps
contact materials and capture chemistry
Device and diagnostics
8 steps
commercial partners and distributors
The commercial briefing
10 steps
Explore a sector
Three sectors have a full walkthrough you can click through, end to end.
critical infrastructure
Grid and networks
A ten-line grid reinforcement case solved to the certified optimum, defensible line by line. A simplified degree-two encoding of it came out about nine per cent worse than the full model: the cost of the simplification, not a classical failure.
maritime and logistics
Berth and fleet plans
Whether the plan your scheduling tool produces still matches what you are actually paid on — answered in about a second, and worth having either way.
national programmes
Sovereign capability
Runs on the small quantum machine a national programme already owns, and the heavy computation never leaves your network.
Questions buyers actually ask
Six questions buyers actually ask before they commit, and the straight answer we give to each. Four of them are already backed by a result on hardware.
| the question | what comes back | state |
|---|---|---|
| Which of our problems is quantum actually worth it for? | A verdict on your own system in about a second, with the cost of the best classical route printed beside it. Free, no key, and it says classical when that is the answer. | flown |
| Will our network hold when part of it goes down? | The leftover spectrum after the contingency and a reinforcement plan. A ten-line transmission-expansion case returned the certified optimum with zero gap on photonic hardware; a simplified degree-two encoding of the same case came out nine per cent worse than the full model on the same machine. | flown |
| Can a quantum machine give us a result we can act on today? | A ranking of your candidates with the smallest difference it can resolve stated beside it, and an abstention when two are too close to separate. The same methods run on superconducting, neutral-atom, trapped-ion and photonic machines: IBM Heron, IBM Nighthawk, QuEra Aquila, AQT IBEX, Rigetti Cepheus, QCi Dirac-3 and IonQ. | flown |
| Does any of this reach our sector? | The map below: fifteen sectors, one constrained problem class. The industry label is not the boundary; the shape of the question is. | flown |
| Can it run on the machine our own programme owns? | The unit is six sites, so it runs on a 5-to-25-qubit machine rather than waiting for a hundred. The heavy computation can stay inside your network. | computed |
| What is past the edge of this? | Not a line we have drawn. The method is derive-first: collapse the problem by algebra, and the search that everything else has to do disappears. Where a lane is undone it is undone, not closed. | open |
The map
| sector | the decision | what you give it | what comes back | state |
|---|---|---|---|---|
| Advanced materials | Which design variant to take forward | Two to twelve candidate structures, or an active space from the package you already run | A ranking with the smallest difference it can resolve, or an abstention | flown |
| Pharmaceutical formulation | Which carrier or excipient to take to the bench | A candidate set, as files or active spaces | The ordering and the resolving floor between adjacent candidates | flown |
| Life sciences | Which residues carry the allosteric signal | A structure file. PDB or mmCIF. | A ranked residue list from structure alone | flown |
| Energy networks and critical infrastructure | Which reinforcement plan survives the contingency | A network file. MATPOWER, with in-service status honoured. | The leftover spectrum after the contingency, and a plan | flown |
| Aerospace and defence | Where the onset sits on a fine grid | A field problem | A field solve, and the cost of the best classical route beside it | computed |
| Automotive and mobility | Where the safe operating boundary actually is | A control problem and its envelope | A certified boundary, and where a coarse surrogate misplaces it | computed |
| Telecommunications | Where throughput is actually lost across the network | A network as an edge list | The real-frequency response of the network | mapped |
| Financial services | Which entities move together before the loss lands | A transaction graph with timestamps | A response feature on the entity graph | mapped |
| Cybersecurity | Which parts of a network fail together under attack | A dependency or asset graph | The response of the network to a disturbance, and where it concentrates | open |
| Maritime and logistics | Which routing or berth plan holds under disruption | A route network, or a schedule as a landscape | A plan, and whether the classical route already settles it | mapped |
| Manufacturing and industrial operations | Which process setting or material change holds the spec across the line | A process landscape, or the candidate chemistry behind it | A ranking, and whether the model you are already solving preserves the true objective | mapped |
| Healthcare delivery | Which pathway or allocation holds when demand moves | A network of units, flows and dependencies | Where the response concentrates, and what the classical route costs to find it | open |
| Climate and sustainability | Which intervention survives the scenario you are worried about | A network, or the chemistry behind a capture or storage candidate | A ranked set of interventions, or the contingency spectrum behind a plan | mapped |
| Space and frontier systems | Where a design or a schedule stops holding | A field problem, a structure, or a constrained plan | The solve, and the cost of the best classical route beside it | open |
| Semiconductors | Which process or material variant holds the spec | Candidate chemistry, or a process landscape | A ranking, and the absorption band line by line where that is the descriptor | flown |
Where each one stands
| Advanced materials flown | One screen of four formulations (the bare core and three counter-anion variants), ranked within 0.3%. The three counter-anion variants ranked on 120 qubits in one job, measured order equal to exact order. Award winner, 2nd place worldwide, GIC 2026 Advanced Materials track (Mitsubishi Chemical Group × AIST). |
| Pharmaceutical formulation flown | Live with a formulation partner on long-acting injectables. |
| Life sciences flown | No assay required. We do not lead the independent benchmark on the current method. |
| Energy networks and critical infrastructure flown | A transmission-expansion case with ten candidate lines, held at its true degree on photonic hardware, returned the certified optimum with zero gap. A simplified degree-two encoding of the same case dropped a candidate line and came back nine per cent worse than the full model, on the same machine in the same session: the cost of that simplification, not a failure of classical methods. |
| Aerospace and defence computed | A field solve to 4e-5 at 4,096 points, with the cost of the best classical route beside it. Ready for hardware. |
| Automotive and mobility computed | A shallow surrogate misplaces the boundary by 32 per cent. That gap is the product. |
| Telecommunications mapped | The same object as the grid work. The mapping is written. |
| Financial services mapped | The same object as the grid work, read on a transaction graph instead of a network. The mapping is written and the lane is open. |
| Cybersecurity open | The question has the shape we read: a coupled system under disturbance. Open for a first engagement. |
| Maritime and logistics mapped | The same constrained landscape as the grid work, on the photonic lane that already carries it. The mapping is written. |
| Manufacturing and industrial operations mapped | The same object as the grid work: a coupled system read for what happens when one part moves. The mapping is written. |
| Healthcare delivery open | The question has the shape we read: a network of units and flows under demand. Open for a first engagement. |
| Climate and sustainability mapped | Two routes reach this: the flown grid lane for networks, and the flown materials lane for capture and storage chemistry. |
| Space and frontier systems open | The field-solve lane reaches this directly. Open for a first engagement. |
| Semiconductors flown | The resist work sits here as well as in materials. |
Four stages of engagement in every sector:liveon quantum hardware today,readyfor hardware with the problem already solved,mappedand ready to start, oropenfor a first engagement. The industry label is not the boundary; the shape of the question is.