Industry · Exploratory
Quantum Computing for Insurance
Insurers run some of the heaviest Monte Carlo simulations in the economy, and they are also exposed to a new kind of risk if quantum computers break encryption. Quantum computing could one day speed up risk calculations, but today's work is exploratory and the security angle is more urgent.
Maturity
Exploratory; research on risk simulation and security preparedness
Most credible quantum workload
Monte Carlo estimation for risk and liability valuation (amplitude estimation)
Hardware needed
Fault-tolerant machines with thousands of logical qubits
Strongest classical competition
GPU Monte Carlo, proxy models, and machine learning surrogates
Immediate task
Assess post-quantum cryptography and cyber-underwriting exposure
Realistic horizon
2030s for any production use of quantum simulation
Overview
Insurance is a business of estimating uncertain future losses. Underwriters price policies from statistical models, actuaries project liabilities decades ahead, and reinsurers aggregate catastrophe risk across the globe. All of that rests on simulation: running a model thousands or millions of times with random inputs and averaging the results. The more complex the product and the tighter the regulation, the more computing the simulations demand, so insurers are watching quantum computing for its promise of faster sampling.
There is a second link that is easy to miss. Insurers hold long-dated liabilities and large archives of sensitive personal data, and they underwrite cyber risk. A future quantum computer that breaks public-key encryption would affect their own systems and the risk they insure, which makes quantum preparedness part of both operations and underwriting, and gives the sector a reason to follow the technology that goes beyond computing speed.
Why this industry is a candidate for quantum computing
Regulation makes simulation central. European Solvency II, for example, sets capital requirements based on a one-year value at risk at the 99.5 percent level, which insurers estimate with large Monte Carlo runs, often nested, where each outer scenario needs inner simulations to value liabilities. Accuracy improves only as one over the square root of the number of paths, so extra precision is expensive.
Quantum amplitude estimation offers a proven quadratic improvement for this kind of estimation, in the same way it does for banks. The catch is the hardware needed and the effort to encode actuarial models into quantum circuits. Insurers are therefore in a research and preparation phase, with the main immediate action being to understand the security implications.
Key facts and numbers
| Item | Detail |
|---|---|
| Solvency II capital standard | Solvency capital is calibrated to a 99.5 percent one-year value at risk |
| Monte Carlo convergence | Classical error shrinks as 1 over the square root of N; quantum amplitude estimation shrinks as 1 over N |
| Derivative pricing estimate | Around 8,000 logical qubits for a useful advantage in a related financial task (2021 study) |
| Quantum risk analysis | Woerner and Egger (2019) showed amplitude estimation applied to value at risk and related measures |
| RSA-2048 break estimate | Under one million noisy physical qubits and under a week (Gidney, 2025) |
| Post-quantum standards | NIST FIPS 203, 204, and 205 published in August 2024 |
Where quantum computing could help
Catastrophe and portfolio risk simulation
The problem: Insurers and reinsurers estimate losses from hurricanes, earthquakes, floods, and pandemics by simulating events across portfolios of exposures. Tail estimates, which matter most for capital, need very many samples.
The quantum approach: Quantum amplitude estimation and quantum mean estimation replace the sampling loop with a quadratic speedup, once the loss model is encoded as a circuit.
Where things stand: Resource studies from finance suggest thousands of logical qubits and very fast logical gates for a useful gain, and loading catastrophe models, which are large and data-heavy, is a significant obstacle. Today's work consists of toy models and feasibility studies.
Related algorithms: Quantum Amplitude Estimation, Quantum Monte Carlo Integration, Quantum Mean Estimation
Life and annuity liability valuation
The problem: Valuing long-dated liabilities with embedded options requires nested Monte Carlo simulations over economic scenarios and policyholder behavior. These runs take hours or days on clusters.
The quantum approach: Quantum algorithms for Monte Carlo and for stochastic differential equations aim to speed up the inner and outer loops.
Where things stand: Industry practice has moved to proxy models and machine learning surrogates that avoid most of the nested simulation, and these set a strong classical baseline. Quantum approaches are theoretical for now.
Related algorithms: Quantum Amplitude Estimation, Quantum Approximate Counting
Fraud detection and claims analytics
The problem: Detecting fraudulent claims, scoring risk, and automating claims triage are classification and anomaly detection problems on imbalanced data.
The quantum approach: Quantum kernels, variational classifiers, and quantum anomaly detection have been tested on small data sets.
Where things stand: Studies report accuracy similar to classical models on small samples. Production fraud systems use gradient boosting and deep learning that are cheap and well understood, and a quantum advantage has not been shown.
Related algorithms: Quantum Kernel Estimation, Quantum Anomaly Detection, Quantum Support Vector Machine
Quantum risk: cyber underwriting and data protection
The problem: Insurers store decades of sensitive data and sell cyber coverage. If public-key encryption fails systemically, losses could be correlated across many policyholders, and data stolen today could be decrypted later.
The quantum approach: Insurers migrate their own systems to post-quantum cryptography, ask clients about their migration plans as part of underwriting, and model systemic scenarios.
Where things stand: Industry bodies and supervisors have begun to discuss quantum risk, and the migration timelines set by governments between about 2030 and 2035 frame the exposure. This is the most concrete quantum topic for insurers today.
Related algorithms: Shor's Algorithm, BB84 Quantum Key Distribution
Who is working on it
- Large insurers and reinsurers, including Allianz, Munich Re, Zurich, and AXA, which have publicly discussed exploratory quantum programs or participation in consortia.
- Banks and asset managers whose financial algorithms, such as amplitude estimation for risk, are directly transferable to insurance.
- Researchers who developed the foundations, including the authors of Quantum Risk Analysis at IBM and groups in academia studying actuarial applications.
- Supervisors and standards bodies, including European authorities and national regulators, which are issuing guidance on post-quantum migration in finance.
- Cloud and quantum vendors such as IBM and Quantinuum, which offer access and consulting for pilots.
Realistic timeline
| Horizon | What to expect |
|---|---|
| Now to 2030 | Post-quantum migration planning, inclusion of quantum risk in underwriting questionnaires, and small research projects on risk simulation. |
| 2030 to 2035 | Early fault-tolerant machines enable small amplitude estimation experiments. Regulators and standards bodies tighten requirements for cryptography. |
| 2035 and later | If large fault-tolerant machines arrive, quantum-accelerated Monte Carlo could appear in actuarial and risk systems. Insurers whose clients migrated late face higher cyber losses. |
How to judge quantum claims in this sector
Insurance has a distinctive way of testing technology claims, because actuaries must justify models to regulators. Any quantum method used in a capital calculation would need validation, documentation, and explainability, and a speedup alone would not be enough. That requirement favors methods that produce the same numbers as classical models, such as amplitude estimation of the same expectation, over opaque learned models.
A fair test of a quantum risk method compares it with the best classical alternative, which today is often a proxy model or a GPU-accelerated simulation. The relevant metric is accuracy for a fixed compute budget and wall-clock time, including the cost of encoding the model. A result that works only on a toy portfolio, or only against a naive simulator, says little about production value.
What stands in the way
- Hardware: useful quadratic speedups need large fault-tolerant machines.
- Data and models: actuarial models are large and rely on external data that are costly to load into quantum states.
- Regulation: capital models require validation, auditability, and stability that complicate adoption of new methods.
- Classical alternatives: proxy models and machine learning cut the cost of simulation without new hardware.
What organizations in this sector should do now
- Inventory cryptography and long-lived data, and put post-quantum migration on the technology roadmap.
- Add quantum readiness questions to cyber underwriting and monitor industry guidance on systemic quantum risk.
- Keep a small research effort on amplitude estimation applied to a representative model, benchmarked against current methods.
- Educate actuarial and risk teams about what quantum can and cannot do, so that vendor claims can be evaluated.
Myths versus reality
- Myth: quantum computers will make insurance pricing instantaneous. Reality: the proven gain is a quadratic speedup in sampling, and it needs hardware that does not exist yet.
- Myth: quantum matters to insurers only in the future. Reality: the cryptographic risk requires action and affects underwriting now.
- Myth: quantum machine learning will transform fraud detection. Reality: classical models perform well, and quantum advantage is unproven.
The bottom line
For insurers, quantum computing has two faces. Computational speedups for risk simulation are a long-term possibility worth tracking, and the security and underwriting implications of quantum risk are a near-term operational matter. The sensible stance is to prepare for the second now and to follow the first without committing capital.
Frequently asked questions
How could quantum computing help insurers?
By speeding up Monte Carlo simulations used for risk, capital, and liability valuation, using amplitude estimation, once fault-tolerant hardware exists.
Is any insurer using quantum computing in production?
No. Insurers run exploratory projects and research collaborations, and production systems remain classical.
What is quantum risk for insurers?
The possibility that quantum computers break public-key cryptography, exposing stored data and creating correlated cyber losses. It affects both their own systems and the policies they underwrite.
Will quantum machine learning improve fraud detection?
Not shown. Classical models are accurate and cheap, and quantum models have matched them only on small data sets.
What should insurers do now?
Plan the post-quantum migration, incorporate quantum risk into cyber underwriting, and run small benchmarked research projects.
Keep exploring
- Algorithm: Quantum Amplitude Estimation
- Algorithm: Quantum Monte Carlo Integration
- Algorithm: Quantum Mean Estimation
- Algorithm: Shor's Algorithm
- Hardware: IBM Heron (r1, r2, and r3)
- Company: IBM
- Company: Quantinuum
- Research: Case Study: Post-Quantum Cryptography Migration in a Large Financial Institution
- Industry: Quantum Computing for Finance
- Industry: Quantum Computing for Cybersecurity
- Industry: Quantum Computing for Climate and the Environment
This page is written for educational purposes and summarizes public announcements, published research, and published resource estimates (last reviewed September 30, 2026). Vendor and customer claims have not all been independently verified, timelines are estimates, and nothing here is investment or business advice.