Canada’s biggest banks continue to rank among the world’s leading lenders on artificial intelligence, but the next test is whether they can turn billions of dollars of investment into measurable financial returns.

TD Bank returned to the top 10 in Evident AI’s latest ranking of banks’ AI capabilities, climbing three places, while RBC maintained its third-place position, behind JPMorgan and Capital One, for the third consecutive year. 

London-based AI benchmarking firm Evident assesses 50 banks worldwide annually on AI talent, innovation, leadership and transparency. Eight of this year’s top 10 banks are headquartered in North America, up from six when the index launched in 2023. 

Daniel Shackleford Capel, Evident’s managing director of banking, said North American banks have benefited from their scale, proximity to major hyperscalers and an early start on AI investment, pointing to TD’s acquisition of Layer 6 and RBC’s founding of in-house research lab Borealis.

The remaining major Canadian banks had mixed results. BMO ranked 20th, down one position from last year, while a sharp improvement in CIBC’s talent metric helped bump it up one spot to 21st. Scotiabank slipped one place to 30th overall. National Bank was not included in Evident’s ranking. 

“Sometimes what on paper looks like a [ranking] decrease is not so much that some banks are doing less, but that others might be moving even faster,” Shackleford Capel said. “Everyone has moved really quickly.”

Canadian banks’ strong showing comes as AI adoption across the sector accelerates. The average bank’s score in Evident’s index increased 26 per cent from a year earlier, nearly three times the average pace of improvement between 2023 and 2025.

Banks are also placing more emphasis on financial targets. Twelve of the 50 surveyed lenders now disclose either realized or projected returns from their AI investments, up from eight last year, with Canadian banks being among the more willing to share numbers around those investments, Shackleford Capel said.

Three of the five Canadian banks in the index have publicly disclosed these targets. At TD’s September 2025 investor day, the bank targeted $200 million in “annualized revenue productivity” and $100 million in annualized cost savings from AI over the medium term. In the first nine months of fiscal 2026, TD generated $195 million from AI and later raised its medium-term target to $500 million across revenue gains and cost savings, according to Evident.

RBC said at its March 2025 investor day that it expects AI to generate between $700 million and $1 billion in enterprise value by fiscal 2027. The bank later clarified in its fourth quarter’s earnings call last year that the target excludes investments in areas including data, computing infrastructure, proprietary models, governance and talent. Evident said RBC remained on track toward that goal through its third quarter. BMO, meanwhile, said in March it expects to generate more than $1 billion from AI by fiscal 2030.

Those figures are difficult to compare, however, because banks define AI differently, Shackleford Capel cautioned. While some bank measures are closely tied to revenue or expenses, others include cost avoidance or risk reduction, and some subtract the cost of AI investment while others don’t, he said.

Most banks remain far from showing how AI investments directly impact their bottom line. Evident analyzed over 1,100 AI use cases banks disclosed since 2021 and found that 87 per cent reported only basic metrics—such as reach, work volume or speed improvements. Only 12 per cent shared operational metrics like efficiency gains or resolution rates, while barely one per cent reported direct financial outcomes like cost savings or added revenue.

Quantifying returns from AI is “tricky” because large projects rarely involve AI alone, Shackleford Capel said. Lenders may be simultaneously redesigning processes and upgrading technology, making it difficult to isolate how much of a company’s improvement came specifically from AI. Banks also differ in how much they are willing to disclose publicly, he added.

Market conditions can further muddy the picture: an AI tool could help investment bankers pitch deals more efficiently, for example, without generating much additional revenue during a weak deal-making period, Shackleford Capel said.

So far, increased AI use has not led to widespread bank layoffs some have feared, with Canadian banks maintaining or improving their ranking positions on AI talent hiring this year, according to Evident.