BMO Reports Quantum Utility In Machine Learning Application Using IBM Quantum Computers
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

BMO says it used IBM quantum computers to demonstrate a machine-learning feature-selection method developed with Democritus University of Thrace. The bank reports that the method produced results comparable to conventional techniques using about one-tenth to one-hundredth as many data samples, but the announcement describes research progress, not a deployed banking product.

BMO says it has demonstrated quantum utility in a machine-learning application using IBM quantum computers, in research with Democritus University of Thrace that selected useful dataset variables with fewer samples than conventional techniques. The bank reported comparable result quality with roughly 10 to 100 times fewer data samples, a finding it says could inform future financial-services applications, though no customer-facing use or commercial performance was announced.

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

The work adapts a framework called Proxy Quantum Games for feature selection, the process of identifying which variables in a dataset are most useful to a machine-learning model. BMO describes the method as treating each feature as a “player” in a strategic game: features receive higher scores when they move with the model’s target and lower scores when they provide little information or duplicate other features.

BMO said the approach was run on IBM quantum computers, including the IBM Quantum Nighthawk r2 processor, and reviewed by IBM’s quantum research team. The bank reported that the method achieved results of similar quality to conventional feature-selection techniques while using one to two orders of magnitude fewer samples. The announcement does not provide the dataset sizes, specific benchmark results, or details needed to independently assess that comparison.

BMO identified client modeling, risk assessment and detection, and personalized banking as possible future applications. It also stressed that application development remains at an early stage. The announcement describes a research demonstration; it does not say the technique is being used in production banking systems or has improved decisions for clients.

At a glance
announcementWhen: Announced today; the source does not pr…
The developmentBMO reported a quantum-computing demonstration for machine-learning feature selection, conducted with IBM quantum computers and Democritus University of Thrace.
Quantum computing · Briefing

BMO reports a quantum research milestone for machine learning

BMO says IBM quantum computers ran a feature-selection method developed with Democritus University of Thrace. The result is an early research demonstration, not a banking product in use.

10–100×
fewer data samples reported
2025
BMO joined the IBM Quantum Network
9th
BMO’s reported 2026 Quantum Index Finance rank among financial institutions

How the method works

Feature selection identifies which dataset variables are most useful to a machine-learning model.

1
Treat features as players

The Proxy Quantum Games framework scores each dataset feature.

2
Score useful signals

Features that move with the model’s target score higher; uninformative or duplicate features score lower.

3
Compare with conventional methods

BMO reports similar result quality using roughly one-tenth to one-hundredth as many samples.

What this could mean for banking

BMO named these as areas for possible future exploration.

Potential use

Client modeling

A possible way to identify useful signals in complex financial data.

Potential use

Risk assessment and detection

A research direction for evaluating risk related to financial services.

Potential use

Personalized banking

An area BMO says could be explored in future applications.

Research progress and proof still needed

The announcement describes a narrow comparison for feature selection, not improved banking outcomes.

Reported
  • Run on IBM quantum computers, including the IBM Quantum Nighthawk r2 processor.
  • Reviewed by IBM’s quantum research team.
  • Comparable quality with fewer samples, according to BMO.
Still pending
  • Dataset sizes and detailed benchmark results.
  • Independent assessment across varied tasks and datasets.
  • Evidence on runtime, cost, reliability, or operational banking results.
“While we remain in the early stages of application development, these projects show how quantum computing can help solve business problems in new ways.”— Kristin Milchanowski, BMO’s chief AI and quantum officer
The takeaway
BMO reports a promising sample-efficiency result, but customers should view it as early research until detailed benchmarks and real-world outcomes are available.

A Data-Efficiency Test for Banking AI

The result matters because machine-learning projects can depend on large datasets, while banking decisions often involve complex information and strict requirements for reliable risk evaluation. If the reported sample-efficiency advantage holds up across relevant tasks, a quantum approach could offer another way to identify useful signals in data. That would be a research benefit, not proof that quantum computers currently outperform conventional systems across banking workloads.

For now, the reported comparison is narrow: it concerns feature selection and sample use, not a demonstrated improvement in loan decisions, fraud detection, customer outcomes, or operating costs. BMO’s examples describe areas where the method might eventually be explored. The distinction matters to readers assessing whether this represents an immediate change to banking services or an early technical milestone.

The announcement also illustrates how financial institutions are testing quantum computing alongside existing AI research. BMO framed the project as part of a longer effort to find practical applications. Its significance will depend on whether subsequent work reproduces the reported performance on varied datasets and shows that the method is useful under real business constraints.

Products Worth Considering

Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W

Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

As an affiliate, we earn on qualifying purchases.

BMO’s Quantum Research Program

BMO said the project was developed with Democritus University of Thrace and used IBM quantum systems. The bank joined the IBM Quantum Network in February 2025 and says it is investing in quantum research, workforce development and applications. It also named partnerships with the Chicago Quantum Exchange and Quantum Industry Canada as part of its broader work in the field.

The bank’s announcement places the demonstration within its Institute for Applied Artificial Intelligence & Quantum, an enterprise-wide centre focused on responsible AI application and governance as well as quantum capabilities. Those institutional details provide context for the research effort but do not independently establish the method’s performance or readiness for commercial use.

BMO also cited external rankings: it said it placed ninth among financial institutions in the 2026 Quantum Index Finance and ranked highest among Canadian commercial banks in that index. The announcement further cited recognition from Evident AI for AI use-case outcomes and responsible AI processes. These rankings concern institutional readiness and AI practices; they are separate from validation of this quantum feature-selection result.

“While we remain in the early stages of application development, these projects show how quantum computing can help solve business problems in new ways.”

— Kristin Milchanowski, BMO’s chief AI and quantum officer

Benchmark Details Still Pending

BMO has not provided the underlying benchmark data, the number and type of datasets tested, or the precise measure used to judge whether the results matched conventional techniques. The announcement also does not specify how many samples each approach used in the experiments or whether the reported reduction applies consistently across different tasks. Without those details, the reported one-to-two-orders-of-magnitude difference cannot be independently evaluated from the announcement alone.

It is also unclear how the method compares with conventional approaches in overall computing time, cost, reliability, or performance on operational banking data. BMO did not announce a deployment timetable, a production system, or measured customer or business outcomes. Its stated banking uses remain potential applications, not confirmed results.

Evidence Needed for Deployment

The next meaningful steps would be publication or release of more detailed evaluation results, followed by testing across a broader range of datasets and machine-learning tasks. Such evidence could clarify how the sample savings were measured, whether the results can be reproduced, and what trade-offs arise in runtime and system resources. The announcement does not specify when any further results will be released.

BMO says it intends to continue research into responsible quantum applications, but it has not named a specific product launch or banking deployment connected to this work. Until further technical details or operational results are available, the development is best understood as a reported research demonstration that may guide future experiments, rather than a change to services customers can use now.

Key Questions

What did BMO demonstrate?

BMO reported using IBM quantum computers to run a machine-learning feature-selection method developed with Democritus University of Thrace. The method ranks which dataset variables may be useful to a model.

What does BMO mean by using fewer samples?

BMO said the method produced results similar in quality to conventional feature-selection techniques while using roughly one-tenth to one-hundredth as many data samples. The announcement does not include the detailed benchmark data needed to independently assess that comparison.

Is the technology already being used for banking customers?

The announcement does not say the method is in production or being used in customer-facing services. BMO described client modeling, risk assessment and personalized banking as possible future applications.

Which quantum computers were involved?

BMO said the work used IBM quantum computers, including the IBM Quantum Nighthawk r2 processor. IBM’s quantum research team reviewed the work, according to BMO.

Source: thequantuminsider.com

You May Also Like

How Wearable Computing Is Moving Beyond Fitness

AIThis post was created with the assistance of artificial intelligence (AI).Wearable computing…

Japan suspected of near $30bn interventions in thin Golden Week trading

Japanese authorities likely intervened in currency markets with over $28.8 billion during Golden Week, raising concerns about market stability and policy stance.

Jay Forrester filed the first practical computer RAM patent 75 years ago this week — his Magnetic Core Memory patent would be granted five years later

Celebrating 75 years since MIT engineer Jay Forrester filed the first practical computer RAM patent, a milestone in digital computing history.

Inside Anthropic: METR’s Study Of AI And The Pace Of Development

METR estimates Anthropic’s AI capability progress accelerated 1.9x, while most of its most informative evidence remains redacted.