Banks and Quantum Computing: What JPMorgan, HSBC and Others Have Actually Shown
Why banks are early adopters
Banks live on math problems: pricing, risk, scheduling, fraud checks and security. They also have the budgets and the patience to run long research programs. So it is no surprise that finance teams are among the most active customers of quantum computing companies. The exciting part is that real hardware results are now appearing in the open. The honest part is that most of them are early demonstrations, announced by the companies involved. For the wider picture see quantum computing in finance and quantum use cases.
Case 1: JPMorgan and certified randomness (March 2025)
JPMorgan Chase worked with Quantinuum, Oak Ridge and Argonne national laboratories and the University of Texas at Austin on a protocol called certified randomness. According to JPMorgan's write-up and the Nature paper, they used Quantinuum's 56-qubit System Model H2, a trapped-ion machine, as an untrusted server. A classical computer sent it challenge circuits one at a time. Each circuit ran in about two seconds on the quantum computer, versus roughly 100 seconds to simulate on the largest supercomputer, as reported. The outputs were checked and certified at least 71,313 bits of entropy.
Why does a bank care about random numbers? Encryption keys, simulations and audits all depend on randomness you can trust. What is impressive here is that the guarantee comes from physics plus a hard-to-fake speed test, not from trusting the hardware maker.
The limits are stated openly by the authors. Verification is heavy: four supercomputers, including Frontier, were used, and it is run only occasionally. The guarantee holds against an adversary only a bounded factor stronger than today's biggest supercomputers. JPMorgan presents it as a first demonstration of a possible cryptographic building block, not a deployed service. (Quantinuum Ltd is the real company; the QNT memecoin has no link to it, see the difference explained.)
Case 2: HSBC and IBM on bond trading (September 2025)
On 25 September 2025 HSBC announced a trial with IBM on European corporate bond trading. The task was to predict the chance that a customer's request for a quote would be filled at the quoted price. HSBC reported that a hybrid workflow, using IBM Heron processors alongside classical computers, improved that prediction by up to 34 percent compared with common classical techniques used in the industry. HSBC called it a world-first for bond trading.
Here is how to read that sentence like a skeptic who still loves the field:
- "Up to" means the best case, not the typical case.
- The classical baseline was not named in the announcement. A 34 percent gain over a simple model is very different from a 34 percent gain over the best model a bank has.
- It was a trial on production-scale data, not a live trading system. The announcement does not describe it as deployed.
- It was announced by the bank and its partner. The announcement itself does not point to peer review.
Even so, this is the kind of result that gets quantum teams funding. Prediction tasks on noisy market data are a plausible place for odd new tools to help, and the bank is right to keep testing.
Case 3: JPMorgan, Argonne and Quantinuum on optimization (May 2024)
In a Science Advances paper, the same group reported evidence of a speedup for the Quantum Approximate Optimization Algorithm (QAOA) on a problem called LABS. The scaling evidence came from ideal, noiseless simulation on a supercomputer, while hardware runs on Quantinuum's H1 and H2 were small. Error detection cut the impact of errors by up to 65 percent, as reported. JPMorgan itself calls it a step toward quantum advantage, not advantage achieved. More on that in our portfolio optimization guide.
Other banks in the picture
Commentary on the sector names Goldman Sachs, Barclays and Credit Agricole alongside JPMorgan and HSBC as banks with quantum teams. Goldman Sachs researchers, together with IBM, have published detailed estimates of what hardware derivative pricing would need, covered in our Monte Carlo guide. The Bank for International Settlements has also published a quantum readiness roadmap (July 2025), which (BIS Papers No 158) argues that institutions should start now with awareness and an inventory of their cryptography, and plan a phased move to quantum-safe encryption. That security side is arguably the most urgent finance story, see harvest now, decrypt later.
A simple scorecard
| Pilot | What is solid | What is not shown yet |
|---|---|---|
| JPMorgan certified randomness | Peer reviewed in Nature, real 56-qubit hardware | Cheap verification, a deployed service |
| HSBC and IBM bond fills | Real hardware, production-scale data | Named baselines, live results, independent checks |
| JPMorgan QAOA scaling | Peer reviewed, hardware plus simulation | Speedup on noisy hardware at scale |
The optimistic, honest takeaway
It is a great sign that serious banks put their names on quantum experiments, and that some results appear in top journals. The field moves by steps: first physics demonstrations, then hybrid tools that help a little, then fault-tolerant machines that change what is possible (see error correction). Right now we are between step one and step two. Treat bank press releases as promising signals, ask what the baseline was, and watch for results that repeat. Nothing here is financial advice, and bank pilots say nothing about the price or future of any token.
Sources and further reading
- JPMorganChase: certified randomness
- HSBC: quantum-enabled algorithmic trading with IBM
- The Quantum Insider: HSBC and IBM
- JPMorganChase: QAOA speedup research
- PostQuantum.com: quantum computing in finance
Reported as of 2026-10-09. Pilot results are mostly announced by the companies involved. This is an educational overview, not financial advice. The QNT memecoin is independent of Quantinuum Ltd and of every bank, lab or company named here.
Frequently asked questions
Has a bank used a quantum computer in live trading?
Not as far as the sources here show. HSBC's September 2025 bond trial used production-scale data, but the announcement does not describe live deployment.
What did JPMorgan show with Quantinuum?
Certified randomness on a 56-qubit trapped-ion machine, published in Nature in March 2025, with at least 71,313 certified bits reported. Verification needed supercomputers.
Does the 34 percent HSBC figure mean quantum beats classical trading?
No. It is an up-to figure for one prediction task against unnamed common classical techniques, announced by the bank and IBM.
Does this say anything about the QNT token?
No. QNT is an independent memecoin with no link to Quantinuum Ltd, JPMorgan, HSBC or any lab. This is not financial advice.
Keep reading
- Quantum Computing in Finance: What Could Change?
How might quantum computing affect banking, trading and risk? A careful beginner guide to the possible uses, the limits, and the link to crypto security. - Quantum Monte Carlo: Risk, Derivative Pricing and the Hardware Gap
How quantum amplitude estimation could speed up the Monte Carlo simulations banks use for pricing and risk, what Goldman Sachs estimated it would take, and why it is a longer-term story. - Quantum Portfolio Optimization: A Reality Check With Real Optimism
Portfolio optimization is the most-hyped quantum finance idea. Here is what QAOA, annealing and benchmarking efforts like QOBLIB really show in 2026. - Where Quantum Finance Claims Outrun the Evidence: A Hype Checklist
A friendly checklist for reading quantum finance, logistics and optimization headlines: baselines, simulation versus hardware, up-to numbers, and why quantum news is not token news.
All Quantum industry, people and AI guides | Back to top | Search the site
Main pages: Quantum computing explained | Quantum and crypto | Companies | Quantum news | Glossary