Feodora Chiosea/iStockPhoto / Getty Images
Before war in the Middle East and the possibility of a global energy crisis became the focus of investor worries, there was the AI scare trade.
New tools from Anthropic’s large language model (LLM), Claude, and a viral report from independent firm Citrini Research in February catalyzed investor worries about the effects of the spread of artificial intelligence technology, hammering the shares of U.S. software firms. The S&P North American Technology Software Index U9BF has shed about a quarter of its value since Jan. 1.
While the disruption narrative has centred on software, professional investors say they are watching AI’s impact on a much broader range of sectors.
“AI is used in everything from my self-driving car to my diagnostics for my radiological equipment,” said Paul MacDonald, president and co-chief investment officer at Harvest ETFs. “We’re just at that component of really starting to see rapid adoption across industries.”
The Globe and Mail spoke with investors, and analyzed earnings calls, management commentary and company reports from the recent Canadian earnings season to understand how they are assessing the potential impact of AI.
Josh Sheluk, portfolio manager and chief investment officer at Verecan Capital Management, said that even those at the cutting edge can’t predict winners and losers, but some company metrics may provide indications of the technology’s effects. Revenue per employee, for instance, may help to show if companies are able to use AI to reduce headcount without sacrificing revenue, and he said he would be paying close attention to profit margins.
“Even if you can’t derive revenue from AI, if you can save on costs from AI that should start to show up on the margin side of things,” he said. He cautioned, however, that markets may start pricing in trends before they show up in company numbers.
Congratulations, you survived the bursting of the AI bubble
In its annual report, CIBC said it had saved approximately 1.2 million hours in the first quarter of 2026 through “AI implementations,” and that an AI client agent had boosted conversion rates for savings accounts by 44 per cent. A BMO investor presentation in late March forecast what it said were around $10-million in expected savings over five years, and about $70-million in revenue, from benefits attributed to AI.
However, even companies that claim benefits could be exposed to potential risks from technologies outside their control.
Charles Brindamour, CEO of property and casualty insurance provider Intact Financial Corp. IFC-T, said in the company’s most recent earnings call that AI models had generated more than $200-million in recurring benefits, and that the use of AI in software engineering had “increased our output by close to 20 per cent per dollar of investment in less than 24 months.”
But in response to a question from Jefferies analyst John Aikens, Mr. Brindamour acknowledged there’s a risk of sales disruptions posed by the increased use of LLMs. He said the company wanted to insure that, particularly in retail insurance, LLMs pointed people to its products first.
“Large language models will certainly have an impact on our ability to capture traffic and shopping in the digital channel,” he said.
While companies consistently tout the potential for AI to enhance their labour productivity, a Telus Corp. T-T management discussion and analysis warned of reputational risks if increased automation leads to layoffs and fewer employment opportunities. New legislation and regulatory requirements around AI may also lead to operating restrictions and create additional costs related to compliance, the discussion said.
Other companies warn that the deployment of AI may not be a simple process. In its 2025 annual report, West Fraser Timber Co. Ltd. WFG-T noted that AI adoption requires specialized skills and said that its workforce may not yet be equipped to support its effective use, leading to additional expenses from training, hiring and organizational changes.
“Competition for specialized AI talent may also increase costs or limit our ability to implement AI initiatives as planned,” it said.
Mr. Sheluk said these risks underscored the difficulty of using a single metric like revenue per employee. “You’re able to cut a bunch of your workforce, but if your technology costs go up by the same degree because now you have a whole bunch of AI infrastructure that you need to support, then that might not actually make your business more profitable or your stock more investable.”
Mr. Sheluk said there’s more at play. “It’s not just revenue, it’s not just profit margins, it’s not just costs. You’re going to have to look at all those things in concert,” he said.