1,450 qubits to break encryption, 7,500 to help finance: why the quantum threat may arrive first

Finance is racing toward quantum computing for the upside. New research suggests the downside could get there first.

Allianz researchers estimate that breaking today's encryption needs roughly 1,250 to 1,450 logical qubits. Leading financial applications need about 4,700 to 7,500. Their July paper is titled, fittingly, "Why the quantum threat could arrive before the reward."

These are estimates, and the summary page does not lay out the assumptions behind them. But the gap explains why a bank can be excited and worried about the same machine.

The reward: JPMorgan's London lab

In June, OQC, JPMorganChase and AMD announced a research collaboration around a dedicated Quantum-AI Data Centre that OQC is building in London.

• OQC provides its GENESIS quantum system and the platform
• AMD supports the AI and classical computing layers
• JPMorganChase will be the first dedicated user of the UK platform

It is expected to be fully operational within 12 months of the announcement.

The planned work includes portfolio optimization, quantum machine learning, AI models that improve quantum circuit performance, and quantum-enhanced AI for discovering financial algorithms. Workflows will be assessed for performance, scalability and reproducibility against financial-services standards.

The announcement does not claim quantum advantage. It describes research into near-term utility.

The reality check

Allianz separates what looks promising from what does not.

Quantum computing can accelerate Monte Carlo methods, mainly for derivative pricing. For portfolio optimization and financial machine learning, the authors see no consistent advantage over improved classical methods yet. Those are two of the main areas in JPMorgan's research plan, which is exactly why it is worth testing them under realistic conditions.

The most quoted trading result needs similar care. IBM says a trial with HSBC showed an up to 34% improvement in predicting whether a bond trade would be filled, compared with baseline methods.

Three things to keep in mind:

• "Up to" makes it a best-case figure
• It measures prediction, not profit
• IBM says more research is needed to reproduce it across market regimes

IBM also describes generating quantum features offline and reusing them. So quantum hardware is not sitting inside the high-speed trading loop.

The threat: an old recipe, waiting for a machine

The method for breaking today's public-key encryption dates back to 1994. What is missing is a machine large and reliable enough to run it.

From the same Allianz paper:

• 28% to 49%: expert odds that a machine capable of breaking current encryption exists within ten years
• After 2030: when NIST plans to phase out today's algorithms, and the EU's migration date for critical financial infrastructure
• About 35%: the share of large organizations that have completed a cryptographic inventory

Defensive standards have existed since 2024, so this is not a problem without a solution. It is a migration problem, and those tend to take years.

The timing is genuinely uncertain. Even so, data encrypted today may need to stay secret for years, which is why planning does not have to wait for the machine.

What this means for trading firms

Our reading: treat these as two separate decisions.

Quantum as a tool belongs in research. Watch the results from London and elsewhere, and ask whether they beat improved classical methods, on live-like data, after costs.

Quantum as a threat belongs in planning now. A few questions to start with:

1. Do you know where encryption is used across your systems, vendors and broker links?
2. Which data must stay confidential for ten years or more?
3. Could you swap algorithms without rebuilding everything?

The machine may be years away. The inventory does not have to be.

Which would you plan for first: the threat or the reward?

References

• Allianz: Cracking finance: Why the quantum threat could arrive before the reward, July 29, 2026
• JPMorganChase: OQC, JPMorganChase and AMD commence research collaboration, June 2026
• HPCwire: OQC, JPMorganChase and AMD to explore hybrid quantum-classical computing, June 4, 2026
• IBM: Quantum computing shows promising potential in finance, December 8, 2025

Qubit figures are logical qubits and are Allianz's estimates. FinRAG.tech has no connection to these firms. The interpretation is ours. This article is commentary, not investment or security advice.