This report compares the six flagship interactive quantum demonstrations built for the IBM Quantum engagement page. Each demo addresses a distinct enterprise question and demonstrates a different quantum computational pattern — optimization, chemistry, simulation, advantage, security, and operations. All six share a common lifecycle (Define → Model → Classical Baseline → Quantum → Hybridize → Validate → Interpret), a shared ClaimsRegistry component that exposes hover-enabled provenance for every quantitative metric, and a consistent credibility posture that distinguishes synthetic demo data from published research and classical benchmarks.
Two demos are grounded in published research (Molecular Discovery references the IBM / Cleveland Clinic / RIKEN Gordon Bell finalist workflow; Advantage & Trust frames the beyond-classical question), two are business simulations of current enterprise problems (Portfolio Optimizer, Logistics Optimizer), one models an emerging research domain (Materials & Energy), and one addresses the "Act Now" problem that exists today regardless of quantum hardware maturity (Quantum-Safe Defender). No demo claims universal quantum advantage; every comparison is framed as hybrid candidate versus classical benchmark.
How can we construct a portfolio when thousands of combinations, constraints and competing objectives make optimization increasingly difficult?
Constrained combinatorial optimization over a synthetic 50-security universe (500 in full production). Three execution paths: classical baseline (heuristic/MILP, CPLEX-like), quantum sampling (binary encoding → quantum circuit → QPU/simulator), and hybrid refinement (quantum samples + classical local search + constraint validation). Dynamically-calculated search space via exact binomial coefficient C(50, maxAssets).
Universe of 500 securities (50 selected for demonstration). User-adjustable constraints: target return, max volatility, max assets, max sector exposure. Side-by-side classical benchmark vs hybrid candidate results table with sector breakdown. Never claims quantum universally wins — framed as HYBRID CANDIDATE vs CLASSICAL BENCHMARK.
React + useState; useRunner hook (3 steps, 1100ms). Synthetic per-security return/volatility assumptions. Bitstring sample generation for quantum sampling visualization. ClaimsRegistry integration on search-space metric and three result metrics (expected return, estimated risk, objective score) in both columns.
Can quantum computing help calculate properties of molecular systems that become extraordinarily difficult to represent accurately using classical computation?
Quantum-centric supercomputing for molecular science. Protein-ligand system (PX-42 receptor, 12,000+ atoms, ~30,000 orbitals). Fragmentation architecture: full system → fragmentation → easy regions to classical HPC, difficult electronic regions to QPU → quantum samples → HPC reconstruction → energy estimate. Live CPU/GPU/QPU/HPC utilization dashboard across 6 pipeline steps.
Three drug candidates (A/B/C) ranked by binding energy (kcal/mol, lower = stronger). References real IBM / Cleveland Clinic / RIKEN quantum-centric workflow at 12,635 atoms — a 2026 Gordon Bell Prize finalist. Includes explicit credibility box: quantum does not currently replace state-of-the-art classical computational chemistry.
React + useState; useRunner (6 steps, 850ms). Synthetic binding-energy model. Per-step resource utilization table. ClaimsRegistry on 2 system metrics, 3 candidate energy values, and 1 research-context claim. Confidence mix: RESEARCH (system + Gordon Bell) and SYNTHETIC (energies).
Can quantum computers model the quantum mechanics of industrial materials and energy systems?
Quantum simulation of molten-salt fusion-blanket materials. User selects salt formulation (LiF-BeF₂, LiF-NaF-KF, or experimental), operating temperature (500–900°C), and lithium concentration (0–100%). 6-step hybrid workflow: classical MD → identify difficult state → quantum circuit → QPU sampling → HPC processing → material prediction. AI Materials Agent proposes next configurations — the AI + HPC + Quantum loop.
Four material-property predictions per run: relative stability, tritium extraction, corrosion proxy, calculation confidence. All predictions flagged EXPERIMENTAL — not validated against physical experiments. Forward-looking AI-proposed next experiments (candidate configurations) with AI + HPC + QPU loop diagram.
React + useState; useRunner (6 steps, 800ms). Outcome matrix keyed by salt formulation. Tone-coded property values (green/amber/red). ClaimsRegistry on all 4 property predictions. Confidence: SYNTHETIC.
How do we know a quantum computer has done something genuinely beyond classical computation — and how do we trust an answer the classical computer cannot reproduce?
70-logical-qubit challenge circuit at high depth — complexity beyond practical exact classical simulation. Classical simulation: memory exceeds practical limit, runtime infeasible. Quantum execution: estimated ~15 min (accelerated playback in demo). Trust framework with syndrome extraction, fidelity bounds, and 4 validation checks (computation, fidelity evidence, error thresholds, classical tractability). Interactive advantage race showing advantage must be continually re-tested as classical algorithms improve.
The intellectual anchor demo. Frames the central question of beyond-classical computation and verifiability. Trust framework pipeline: logical encoding → error syndromes → noise characterization → statistical tests → fidelity bounds → trusted result. Advantage race with controls to improve classical algorithm, quantum hardware, or error correction.
React + useState; useRunner (5 steps, 700ms). Live syndrome-extraction visualization (40-cell grid). Race state with three improvement buttons. ClaimsRegistry on qubit count, time estimate, and all 4 validation checks. Confidence: RESEARCH + BENCHMARK (classical tractability).
What enterprise quantum problem requires action even before large fault-tolerant quantum computers exist?
Cryptographic discovery, CBOM (Cryptography Bill of Materials) generation, Harvest-Now/Decrypt-Later simulation, and post-quantum remediation for a synthetic enterprise (GlobalBank). Four-tab workflow: Discover (scan repositories/apps/certs/network), CBOM (generate crypto inventory table), HNDL (simulate capture → store → future decrypt with adjustable confidentiality lifetime), Remediate (migrate RSA-2048 to PQC with crypto-agility abstraction).
Synthetic enterprise: 437 apps, 12,400 servers, 8,700 certs, 412 APIs, 17 BUs. Discovery KPIs: 27,482 crypto assets, 8,341 certificates, 312 libraries, 1,284 quantum-vulnerable, 73 critical risks. Algorithm distribution (RSA/ECDSA/AES/SHA/ML-KEM). CBOM table with 6 application records showing algorithm, key, data sensitivity, exposure, lifetime, risk. Explicitly framed as the "Act Now" track — a business problem that exists today.
React + useState; useRunner (4 steps, 600ms). Tab-based sub-step navigation. CBOM table with risk-tone coloring. HNDL lifetime slider (1–25y) driving dynamic year projection. ClaimsRegistry on 5 discovery KPIs. Confidence: SYNTHETIC. All data explicitly labeled simulated — no production security data.
How should a company allocate scarce resources across many competing objectives and constraints?
Multi-objective supply-chain optimization. 6 distribution centers, 40 vehicles, 250 orders, 400 delivery constraints. Objectives: minimize distance, cost, late deliveries, carbon; maximize service level and utilization. Toggleable operational disruptions (fuel +20%, driver availability −15%, DC-3 offline, priority hospital order). 8-step hybrid algorithm: operational data → constraint model → classical preprocessing → quantum sampling → candidate schedules → classical refinement → feasibility check → dispatch plan. Network map with re-routed dispatch visualization.
Side-by-side comparison: classical heuristic vs quantum-hybrid candidate across cost, late orders, and miles. Results dynamically perturbed by active disruption count. Explicitly framed as DEMO RESULT — does not establish quantum advantage; exists to explain combinatorial optimization in business terms.
React + useState; useRunner (8 steps, 500ms). SVG network map with 6 DC nodes and dynamic route coloring. Disruption toggle state. Perturbation model scaling metrics by active-count. ClaimsRegistry on 3 metrics × 2 blocks (6 hoverable instances, 3 unique claims). Confidence: SYNTHETIC.
| Confidence Level | Count | Description |
|---|---|---|
| SYNTHETIC | 19 | Generated by the demo from a synthetic model. No real-world or production data. |
| RESEARCH | 8 | Based on published research (IBM / Cleveland Clinic / RIKEN, Gordon Bell Prize finalist work). |
| BENCHMARK | 1 | Classical benchmark — the basis of the beyond-classical claim in the Advantage & Trust demo. |
useRunner hook that animates step-by-step execution.ClaimMetric component exposing a registered claim (ID, source, methodology, confidence, date) on hover — 28 registered claims across the six demos, with a ClaimsAvailableChip in each demo header indicating the hoverable count.