Banks are not, as a rule, in the earthquake business. They are in the business of pricing risk, which is adjacent. Still, the operational job of telling the ground when it is about to move has historically belonged to seismologists, governments, and a handful of specialist insurers. Bank of Montreal would like to change that, or at least to share the work.
In an interview with Bloomberg published on 1 May, Kristin Milchanowski, BMO’s recently appointed Chief AI and Quantum Officer, said the bank had filed a provisional patent on a quantum algorithm intended to help forecast earthquakes. The same team, she said, is using artificial intelligence to dispatch mobile banking units to communities affected by wildfires, including those that swept through parts of Los Angeles last year. It is an unusual portfolio for a Tier 1 Canadian lender, and it is meant to be.
From back office to backstop
BMO formalised its bet last month with the launch of the BMO Institute for Applied Artificial Intelligence & Quantum, announced on 9 April. The Institute, an enterprise-wide centre of excellence, is intended to consolidate the bank’s research, governance, and applied work in two technologies that have, until recently, sat at opposite ends of the maturity curve. AI is everywhere in financial services, integrated into fraud detection, credit scoring, and customer service. Quantum computing, by contrast, is still mostly a research line, valuable in theory and largely unproven in deployed banking workflows.
Milchanowski, who spent the previous eighteen months as the bank’s chief AI and data officer, leads the Institute as its founding director. Days after its launch, BMO announced partnerships with Quantum Industry Canada and the Chicago Quantum Exchange, two of the more established quantum policy and research bodies in North America. On paper, this is corporate housekeeping. In practice, it is a deliberate signal: BMO wants to be read as a quantum-curious bank, not just an AI-fluent one.
The provisional patent is, for now, the most concrete artefact of that ambition. Quantum algorithms are well-suited, in principle, to problems involving high-dimensional optimisation and combinatorial search, the kinds of computations that overwhelm classical hardware as data sets grow. Seismic forecasting, which relies on enormous volumes of geophysical signal data and models that are notoriously difficult to fit, is one such problem.
Milchanowski did not, in the published interview, claim that BMO had cracked it. The patent is provisional, the algorithm has not been independently benchmarked, and useful quantum hardware capable of running such workloads at scale does not yet exist outside research labs. What the filing represents is intent, and a small bet on optionality. If quantum advantage in this domain ever materialises, BMO will own a piece of it.
There is also a more immediate commercial logic. Better catastrophe modelling has direct applications in insurance, mortgage portfolios, and infrastructure lending, all areas where Canadian banks have meaningful exposure and where climate-driven loss patterns are forcing a rethink of underwriting. Banks hold massive balance sheets tied to real estate and infrastructure; accurately predicting where earthquakes or wildfires will strike can reduce unexpected losses. For BMO, which has significant operations in both Canada and the United States, including California, the ability to forecast seismic events could be a game-changer for its mortgage and commercial loan portfolios.
The science of earthquake prediction is notoriously difficult. Unlike weather forecasting, which benefits from dense sensor networks and well-understood physics, earthquakes involve complex interactions of tectonic plates, stress accumulation, and fracture mechanics that are still poorly understood. Classical machine learning models have struggled to produce reliable forecasts, partly because the data is sparse and the signal-to-noise ratio is low. Quantum computers, which can represent and process exponentially more states than classical bits, offer a potential way to simulate these interactions at a granular level that classical supercomputers cannot achieve. However, current quantum hardware is still in the noisy intermediate-scale era, with limited qubit counts and high error rates, so the actual implementation of such an algorithm remains a long-term ambition.
If quantum is a long-dated option, the bank’s AI work is already in production. Milchanowski told Bloomberg that BMO is using AI models to identify communities cut off by wildfire and to route mobile branch units, essentially banks-on-wheels, into them. The Los Angeles fires of early 2025 left tens of thousands of residents without access to physical banking; AI dispatch, in the bank’s telling, helps reduce the lag between displacement and service restoration. This is not a glamorous use case. It will not feature in earnings calls. But it is the kind of work that distinguishes AI-as-marketing from AI-as-operations, and it is consistent with a broader pattern at BMO, which has been quieter than some peers about its AI roadmap and more visible about its applied projects.
The timing also matters. Bloomberg reported earlier this month that the wider financial industry is bifurcating in its quantum stance. Goldman Sachs has scaled back parts of its quantum research effort, while JPMorgan continues to invest. BMO’s announcement positions it on the investment side of that line, although on a smaller scale. Other major banks have also dabbled in quantum: HSBC has explored quantum for pricing derivatives, and Wells Fargo has a quantum computing research group. However, BMO’s focus on physical catastrophe modeling rather than purely financial optimization sets it apart and aligns with its broader climate risk strategy.
Whether any of this produces returns within the typical investor horizon is another question. Quantum hardware capable of solving real-world problems faster than classical alternatives is, by the consensus estimate of researchers, several years away. BMO is essentially arguing that the risks of being late outweigh the costs of being early. It is a defensible position, particularly for a bank with substantial climate exposure and a reputational interest in showing it takes the technology seriously. It is also, by definition, a bet that may not pay for years, if it pays at all.
For now, the most novel thing the bank can show for it is a piece of paper at the US Patent and Trademark Office and a fleet of mobile branches arriving where the smoke has cleared. Yet that piece of paper represents more than just a patent filing; it is a signal to the market that BMO is willing to invest in frontier technologies to solve fundamental problems. The Institute’s launch also cements a governance structure that will allow the bank to scale its AI and quantum efforts as the technology matures. In an era where climate change is increasing the frequency and severity of natural disasters, financial institutions must evolve their risk assessment tools. BMO’s bet on quantum earthquake forecasting may seem far-fetched today, but it could become a core competitive advantage in the coming decades.