Banks could treat quantum compute capacity as a deterrent in global financial conflict, like nuclear early-warning systems, but for real-time risk asymmetry.
I recently caught up with the Bundesverband Deutscher Banken on the future of quantum computing for banking in Germany. For context, German banks play an outsized role in economic orchestration, not just capital allocation. They:
- fund 80%+ of corporate debt.
- enable 90%+ of export financing.
- guide ESG capital flows (Germany leads OECD in green bond issuance).
As such, banks are not passive technology adopters. They are critical economic infrastructure. Quantum computing (QC) will not be a tool for operational efficiency alone — it is a potential sovereign capability lever, with both upside and systemic risk.
- Part I: Widely known opportunities and risks
- Part II: Six Fringe Quantum Frontiers for Banking
- Part III: Three real QC disruptions
- Part IV: What’s so special about Q-ABM?!
Part Ia: Quantum Computing Opportunities in Banking — The Usual.
| Category | Quantum Opportunity | Product Examples |
|---|---|---|
| Risk Analytics & Portfolio Optimization | Quantum-enhanced Monte Carlo simulations drastically reduce run time and improve precision. | Dynamic credit risk rebalancing; real-time stress tests; regulatory simulations (Basel IV compliance). |
| Pricing and Hedging of Complex Derivatives | Quantum algorithms (e.g. QAOA, HHL) can outperform classical solvers in high-dimensional spaces. | Structured product pricing; exotic option hedging; faster VAR scenario generation. |
| Secure Finance and Data Sovereignty | Quantum Key Distribution (QKD) and post-quantum cryptography guard client confidentiality — critical for export-sensitive industrial clients. | Quantum-secure correspondent banking; encrypted trade finance rails. |
| SME Lending and Sustainability | QC applied to ESG scoring and SME credit models may close the data asymmetry gap in Mittelstand finance. | Quantum-enhanced sustainability-linked loans for mid-cap industrials. |
Specifically for Germany, there are three strategic control opportunities:
- Sovereign Compute: Banks could co-own or partner on quantum hardware with EU cloud or HPC initiatives (GAIA-X, JUPITER, etc.).
- Capital Markets Edge: If quantum-native banks gain 5–10ms faster trade execution or better pricing, Frankfurt’s status in global capital markets improves.
- Green Finance: Quantum machine learning can optimize asset allocations under climate uncertainty (e.g., stochastic climate scenarios + carbon pricing volatilities).
Part Ib: Quantum Computing in Banking — The Known Risks
Technological Dependence and Timing Risk
- Vendor Lock-in: US or Chinese quantum cloud providers could introduce a dependency comparable to SWIFT or semiconductors.
- Hype Cycle Timing: If banks overcommit before fault-tolerant QC is production-ready (~2030s), CapEx is stranded.
Cryptographic Collapse
- Shor’s Algorithm can break RSA-2048 in polynomial time. Once a 10,000 logical qubit machine is operational (~mid-2030s), today’s financial archives (legal agreements, keys, client records) become vulnerable to “harvest-now-decrypt-later” attacks.
- Exposure: Core banking systems, interbank networks, digital identities.
Job Displacement
- Quantitative analysts using classical methods.
- Operations and reconciliation roles — QC could resolve matching problems with near certainty.
- Cybersecurity teams reliant on legacy encryption protocols.
Secondary Effects
- Monetary Policy: If central banks use quantum models for inflation targeting, it could affect interest rate volatility — impacting bank balance sheet strategy.
- Geo-Financial Warfare: Quantum advantage could tilt competitive edges in FX, trade finance, or sovereign debt modeling — areas where German banks have strong export dependency.
Part II: Six Fringe Quantum Frontiers for Banking: Where It Gets Weird — and Useful
1. Quantum Advantage in Macro Forecasting Under Radical Uncertainty
Banks model financial systems using assumptions of ergodicity, stationarity, or convexity. But the real economy, particularly in the face of climate shocks, demographic collapse, and geopolitical dislocations, is none of those.
- Use quantum annealers to simulate non-linear macro scenarios where thousands of interdependent variables (energy, water, trade flows, labor shifts) create wicked problem clusters.
- Extend from “stress tests” to “complexity tests” — how fragile is the entire Eurozone lending architecture under a 5-sigma event?
This is not for quarterly earnings calls. It’s for systemic capital reallocation under deep uncertainty.
“Ex ante, agents cannot know what is the best thing to do because the outcomes of planned actions cannot be known, to any measurable extent.” — David Tuckett
2. Quantum-Enhanced Sovereign Wealth Optimization
Germany doesn’t have a sovereign wealth fund (yet), but its banks effectively act as national allocators through development finance, SME credit, and ESG capital structuring.
- Combine quantum machine learning with public climate-economic datasets to run counterfactual climate investment strategies — e.g., what if we overfund battery production in the wrong chemistry? Or back the wrong grid model?
This pushes quantum into a counterfactual optimizer — a tool not to predict but to rehearse regrets at national scale.
3. Quantum-Simulated Legal and Regulatory Arbitration
Compliance costs consume 10–20% of large bank operational budgets. But most regulations are logically contradictory under dynamic cross-jurisdictional contexts (e.g., GDPR vs. KYC transparency).
- Use quantum logic solvers (Grover variants or quantum SAT solvers) to simulate future regulatory collisions (e.g., between digital identity portability, crypto custody, and sanctions regimes).
- Banks could preemptively rearchitect systems to hedge against regulation-as-risk.
This isn’t compliance. It’s regulatory arbitrage at compute scale.
4. Quantum Markets as Synthetic Control Systems
Imagine designing entire synthetic economies to rehearse incentives — not in spreadsheets, but in living simulations with policy agents, energy shocks, and behavioral feedback.
- Use quantum agent-based modeling (Q-ABM) to simulate how financial networks respond to tail events (e.g., digital euro launch, global cyber shock).
- Treat these models as wind tunnels for policy — then sell the insights as products to ministries, multilaterals, or insurers.
If central banks can issue synthetic CBDCs, why can’t banks run synthetic economies?
5. Quantum Finance as Game-Theoretic Deterrence
“Data symmetry is the new deterrence.” — Ram Charan
- Banks could treat quantum compute capacity itself as a deterrent in global financial conflict. Like nuclear early-warning systems, but for real-time risk asymmetry.
- A bank (or country) with real-time quantum-powered visibility into cross-border capital flows, debt spirals, or trade coercion gains a geo-financial edge — not by acting, but by knowing first.
This reframes quantum not as a tool, but as a power posture. Especially important for Germany if quantum supremacy emerges outside NATO.
6. Quantum Ethics for Capital Allocation
At the extreme edges, quantum computing forces us to ask:
- What does it mean to model a future that is inherently unknowable?
- If quantum systems simulate counterfactuals, who chooses which futures are worth exploring?
- Can a bank become an epistemic institution — not just a capital allocator, but a “choice architect” for society’s trajectory?
This aligns with my general investment approach: don’t just invest in technologies, invest in trajectories.
Part III: Why Bother?
Most quantum encryption discussions are inflated. It’s yet another technology that can be bought by a vendor. it will come with a steep learning curve, but that was likewise the case with the “Internet”, “Cloud Computing”, “Java Software Development” (after many years of COBOL and FORTRAN). But where it is disruptive — and not merely incremental — has less to do with cryptographic performance, and more with the architecture of trust in finance, statecraft, and industrial coordination.
1. QC Disruption: Re-wiring what ‘trust’ means.
Classical encryption rests on assumptions of computational difficulty. Quantum encryption (e.g., Quantum Key Distribution, or QKD) rests on physical laws (no-cloning theorem, entanglement, collapse on measurement). That’s not just a new supplier — that’s a change in the epistemic basis of digital trust.
- In classical systems, keys are secrets. In quantum, keys are detected if eavesdropped.
- That makes the communication link itself a trust actor. Not just the sender/receiver.
- It’s the difference between locking a door (and hoping the lock holds) vs. the door screaming if touched.
Implication: Banks, states, and exchanges can verify if they’re being watched — a form of adversary detection not possible in classical systems.
2. QC Disruption: Displacing global trust intermediaries.
Trust in Swift, VisaNet, CHIPS, Bloomberg, Clearstream, etc. is ultimately a trust in institutional processes (compliance, access control, insurance, etc.). PKI (Public Key Infrastructure) is layered over geopolitical arrangements. It assumes adversaries can’t solve discrete logs or factor large primes.
With quantum-secured networks, particularly when combined with satellite QKD:
- Trust can be infrastructure-native and state-aligned.
- China already operates space-based QKD via Micius satellite.
- EU’s IRIS² and NATO allies are exploring QKD over terrestrial fiber and orbital relays.
So if you control the quantum channel, you control the strategic trust layer. This is disruptive in kind, not just in degree.
3. QC Disruption: Rendering legacy archives vulnerable (retroactive collapse of confidentiality).
Even if quantum encryption doesn’t replace current standards, quantum decryption threatens everything that was encrypted — and stored. This is the “harvest-now, decrypt-later” threat: Adversaries (state or criminal) are hoarding encrypted communications, trade contracts, medical records, SWIFT logs. Once quantum computers hit ~10,000 logical qubits (10–15 years by optimistic projections), much of it may be decrypted.
So the disruption isn’t future-facing, it’s past-shattering.
For a bank, that means:
- Legal contracts and records become retroactively insecure.
- Exposure to liability and counterparty risk across all historical digital interactions.
This breaks the time structure of trust, not just the algorithm.
PART IV: Q-ABM — Why It’s Special
Most “quantum modeling” today is still rooted in physics and chemistry — solving linear algebraic systems (e.g. Schrödinger equation, Hamiltonians) faster using quantum hardware. Useful, yes — but it doesn’t shift the way we think about systems. Q-ABM (Quantum Agent-Based Modeling) is qualitatively different. It’s not just quantum computing accelerating existing models. It’s quantum changing what it means to model decentralized behavior under uncertainty:
- It Models Strategic Agents — Not Just States.
- It Embeds Decision-Making Under Ambiguity — Not Just Probability.
- It Enables “Moral Uncertainty” and Value Pluralism in Simulations.
- It Collapses Paths Only When Observed — Just Like Real Markets.
Q-ABM is not just modeling behavior. It is modeling how complex adaptive agents think about other agents, including their trust, anticipation, and memory. It’s suited for central bank policy modeling, climate-finance adaptation, financial contagion under regulatory arbitrage, and even AI-human incentive alignment. Where classical simulations run faster, Q-ABM simulates deeper.
1. It Models Strategic Agents — Not Just States.
Regular quantum modeling solves state evolution. You give it initial conditions and it evolves a wavefunction or matrix. Q-ABM models agents with goals, memory, imperfect knowledge, and interactions:
- Think of a bank, a regulator, a trader, and an energy grid operator — all interacting under asymmetric information.
- These agents aren’t governed by global equilibrium equations. They adapt, explore, bluff, or panic.
In classical ABM, one must explicitly simulate every trajectory. Complexity explodes exponentially with more agents and decisions. In Q-ABM:
- Superposition encodes many potential agent states at once.
- Entanglement allows correlation structures between agents (e.g. “if the regulator signals hawkishness, the trader’s probability of shorting increases”).
This allows simulation of not just what will happen, but how belief systems co-evolve — like modeling expectations as quantum amplitudes.
2. It Embeds Decision-Making Under Ambiguity — Not Just Probability.
Classical ABM assumes uncertainty is statistical. But real-world systems face ambiguity: not knowing what’s possible, not just how likely. Q-ABM can encode ambiguous decision trees using quantum logic (e.g. qutrits or indefinite causal order). This means:
- Agents can maintain superposed intentions (undecided until triggered).
- You can model scenarios where outcomes are dependent on the order of actions — which classical models cannot easily encode.
For example, a sovereign default simulation evaluates whether the IMF should intervene before or after China restructures bilateral loans changes the path. In Q-ABM, the entire tree of causal sequences can be encoded and evolved simultaneously.
3. It Enables “Moral Uncertainty” and Value Pluralism in Simulations.
In many bank/government systems, there’s no single utility function:
- The central bank wants stability.
- The finance ministry wants growth.
- Citizens want fairness.
- Firms want ROI.
Q-ABM allows agents to operate on incommensurable value sets. You can encode multi-dimensional “ethical amplitudes”, while also simulating agents whose priorities change over time or react to other values. This is essential for simulating:
- CBDC rollout under privacy-vs-KYC tradeoffs.
- ESG capital flows with dynamic stakeholder weights.
- Public trust under AI-assisted governance.
4. It Collapses Paths Only When Observed — Just Like Real Markets.
In classical ABM, every path is traced deterministically or probabilistically, and you must pick a scenario set and run them independently. In Q-ABM paths are superposed until you “observe” the simulation — i.e., choose a measurement basis. This mimics real-world decision inertia:
- Until a policy is announced or a trade is executed, the system exists in a cloud of potentialities.
- Q-ABM lets you compute across that cloud without collapsing it until needed.
This is computationally radical.
BONUS: Startups and People
You made it through another long-format post of mine. Or at least scrolled to the bottom.
Startups:
- SandboxAQ is developing Large Quantitative Models (LQMs) that integrate quantum computing techniques to simulate complex systems. Their LQMs aim to model real-world systems, which may include agent-based dynamics, particularly in fields like finance and healthcare. They raised $150 million in funding with backing from Google and Nvidia, highlighting significant industry interest.
- Multiverse Computing is headquartered in Spain. They provide quantum and quantum-inspired software solutions across various sectors, including finance and energy. Their platform, Singularity, allows users to apply quantum algorithms to complex problems, which can encompass agent-based modeling scenarios. Publicly they talk about collaborations with the Bank of Canada and BASF to explore quantum computing applications in financial modeling and materials science. But there is more 😉
- MIT Media Lab is developing the Quantum Urbanizable Booster (QUB) platform aimed at accelerating the simulation of humanized agent-based models using real-world data. It leverages advanced quantum computing techniques to enhance urban planning simulations, directly aligning with Q-ABM principles.
- Agentics Foundation (Open Source) is developing a quantum-inspired task scheduling system designed for multi-agent environments. The framework utilizes quantum-inspired algorithms to optimize task allocation among agents, a core aspect of agent-based modeling.
People:
- Bob Coecke is Chief Scientist at Quantinuum and formerly Professor at the University of Oxford. He pioneered categorical quantum mechanics and diagrammatic reasoning, providing foundational frameworks for integrating quantum computing with agent-based systems.
- Giuseppe Carleo is Professor at Ecole Polytechnique Federale de Lausanne (EPFL). He developed neural network quantum states, bridging machine learning with quantum physics, which can be applied to simulate complex agent-based systems.
- Jingbo Wang is Professor and Head of Physics Department at the University of Western Australia. She researches quantum walks and their applications in analyzing complex networks, relevant to agent-based modeling.
- Vlatko Vedral is Professor at the University of Oxford. He is known for work in quantum information theory and quantum entanglement, providing theoretical underpinnings that can inform Q-ABM frameworks.