Quantum update — July 2026

A short update on what's new at qubit-lab.ch — no hype, just what we built.

qubit-lab.ch helps banks and other financial institutions understand where quantum computing matters, where it doesn't yet, and how to build practical capability — evidence, not promises. On the opportunity side, the Lean Quantum Exploration Process and our configurable Rapid Quantum Prototypes (RQPs) let you test quantum on your own use cases; on the risk side, PQC Mobilization and PQC Navigator prepare you for post-quantum security. The RQPs below make the opportunity side hands-on: your own data and requirements go into a simple Excel workbook, runs happen in simulation and on real quantum hardware, and every result is benchmarked against an exact classical referee. Workbook in, certified report out — no hand-built quantum circuits needed. You focus on tuning the circuit and its training; the RQPs do the rest.

New RQP: Index Replication

Replicate an index with a fraction of its names — demo case: the full S&P 500, all 500 positions in the objective — with fixed, excluded and bundled positions as first-class constraints. ▶ Watch the 3-minute demo

New RQP: Feature Selection

Which of your data columns actually carry signal? The newest RQP screens hundreds of candidate variables down to the set that matters — with group budgets as first-class constraints — and hands off directly to the QML RQP.

Hardware second opinion

Portfolio optimization now runs on IBM quantum hardware as well — same problem, real device, compared side by side with the simulation.

A certified referee for every result

Portfolio Optimization and Index Replication now benchmark against an exact classical solver. When we say “optimal”, it's proven — not claimed. The numbers are public — wins and losses: qubit-lab.ch/evidence. ▶ Portfolio Optimization demo

Gate before you train

The QML RQP now assesses your data before any training run: a fast spectral probe tells you whether there's structure a quantum model could even exploit — including an honest “don't bother” verdict before you spend a franc on training.

The agentic interface

Claude, ChatGPT & co. can now drive our RQPs directly: one plain-English mandate in, one certified report out. A full index-replication mandate, live, in 4 minutes: ▶ Agentic Quantum Finance

Beyond the tools

Awareness briefings for leadership and risk, the five-session Quantum Capability Kickstart for data, quant and risk teams, and structured advisory up to the decision paper — everything slots into our Lean Quantum Exploration Process: lean steps from first question to a defensible decision, no big-bang commitment. The map is on the freshly revamped qubit-lab.ch/services. And when the question is quantum risk rather than quantum opportunity: PQC Mobilization and PQC Navigator take you from first stakeholder alignment to a structured readiness assessment — including exactly the board-level strategy and roadmap FINMA now recommends having by mid-2027 — qubit-lab.ch/pqc-navigator.

The RQP suite today — five tools, all on your own Excel data

Portfolio Optimization — allocation under real mandate constraints. Demo case: 78 positions, 23 investment decisions on 23 qubits; QAOA reached 100% of the certified optimum — on an instance small enough that the referee can prove it, which is the point.

Index Replication — lean index tracking with certified rule pricing. Demo case: the S&P 500, 17 selection decisions on 17 qubits; minimum tracking error 4.473% proven, and the two mandate rules cost +0.33% and +0.34% per year.

Feature Selection — screens wide datasets down to the variables that carry signal, under group budgets — the front door to QML.

Quantum Classification (QML) — fraud, anomaly and credit scoring prototypes. On a quantum-native benchmark the 6-qubit classifier beat the best classical baseline 0.89 to 0.71 test F1 — on real power-price data the same instrument recommended classical, and the report says so.

Quantum Monte Carlo Lab — derivatives-pricing convergence studies across European, digital, basket and barrier payoffs.

A two-page overview of the RQP suite with worked examples is available on request — write to daniel.hug@qubit-lab.ch.

Worth your attention

FINMA Guidance 05/2026 on quantum computing (Aufsichtsmitteilung, July 9) — the Swiss regulator surveyed 60 financial institutions and now recommends a PQC roadmap — board-adopted strategy, milestones, target dates — by mid-2027 at the latest (§3.1). Today, only 8% have one. That leaves a twelve-month runway for the other 92%, and “harvest now, decrypt later” doesn't wait for roadmaps.

“The Fourier Wall” (arXiv:2607.15815) — a quantum ML model is, at its core, a Fourier series, and a plain trigonometric model on the same frequencies often matches it. We liked the question so much that our QML data assessment now trains exactly that classical twin as referee — an independent implementation inspired by the paper.

Crédit Agricole CIB & Pasqal (announcement, June 30) — expanded partnership with a stated goal of quantum in production for portfolio optimization and counterparty credit risk as early as 2028. Banks announce partnerships all the time; a written production date for named use cases is new — and 2028 is refreshingly sober.

Goldman Sachs & JPMorgan — the quantum split (Bloomberg, April 26) — Goldman scaled back its quantum research after concluding practical advantage in portfolio optimization is still far out; JPMorgan keeps 50+ specialists on it across business lines. Two smart houses, same facts, opposite conclusions — both can be right, which is exactly why testing your own use cases beats waiting for consensus.

If any of this is worth a conversation: get in touch — happy to show the tools live.

Daniel

Daniel Hug · qubit-lab.ch · daniel.hug@qubit-lab.ch