Expert explains why treating AI as cost‑cutting project will be ‘lose‑lose situation’ for employers
The world’s biggest tech firms are racing ahead with their massive AI investments, funnelling hundreds of billions into AI data centres and chips while trimming tens of thousands of jobs.
March saw the highest number of tech worker layoffs in at least two years, while leaders attribute the cuts to AI efficiencies.
It was reported this week that companies such as Alphabet, Meta, Amazon and Microsoft are spending an unprecedented US$674 billion on AI‑heavy capital spending for 2026, with the Wall Street Journal calling out the leaders for “trading their people for more chips.”
At the same time, Nvidia vice-president of applied deep learning Bryan Catanzaro told Axios his team’s compute cost is “far beyond the costs of the employees.”
For business leaders in Canada being asked to find room in their budgets for AI, the challenge is seeing the real story through the headlines as they plan to invest without gutting the workforce they need to make the technology pay off.
Budgeting for AI: cost-cutters versus value creators
Ajay Agrawal, professor of entrepreneurship at the University of Toronto’s Rotman School of Management and a researcher on the economics of AI, says a main issue is that the value of automation is being misunderstood.
He points to a basic principle of cost-cutting versus increasing revenue as an example of what is being lost in the conversation.
“The benefit that you’re trying to get from AI is limited to, effectively, the value of saving the cost that you would spend on [laid-off] people,” he says.
“But that’s the total value that you’re getting from your AI, in comparison to doubling or tripling your sales” if you kept those people but used AI creatively to innovate and expand the organization’s offerings.
For Agrawal, how organizations are currently using AI can be separated into two categories: cost cutters and value creators; employers gravitate toward the first category because the savings are easier to see. But that focus can become self‑fulfilling.
“They can see the benefit quicker,” Agrawal says.
“The problem with it is because they get focused on cost-cutting, the way they use AI becomes very focused on delivering that outcome, and then they miss the much bigger opportunities of increasing the overall size of the pie.”
In contrast, the more sustainable value‑creator lens treats AI as a way to “supercharge” existing employees, says Agrawal, and grow the pie – which in turn supports hiring and development rather than permanent budget contraction.
The illusion of higher revenue per employee
Yu‑Ping Chen, associate professor of management at Concordia University’s John Molson School of Business, warns that headline AI metrics can mislead actual budget‑setters who aren’t involved in what the Wall Street Journal calls the “game of chicken” among tech CEOs.
Chen points to revenue per employee as a classic example of a highly watched number that can rise for the wrong reasons.
“After increasing AI or investment in AI, the revenue per person is getting bigger... because companies lay off people,” Chen says.
“But this could be a very big issue in terms of budgeting.”
His concern is what happens a few years into that strategy, when that AI boomerang risk is particularly acute in specialized labour markets where rehiring takes time and money; cutting too deeply to hit short‑term efficiency targets can leave organizations with permanently constrained budgets, higher recruitment costs and an empty pipeline, just when AI‑driven growth opportunities arrive.
“Eventually, it will be a lose‑lose situation, because you push out the people who actually do the work, but at the same time, because you reduced your budget, there’s no way to do any extra work,” Chen says.
“Your budget is smaller, and in the long term, you will not have enough budget to rehire the required human capital.”
AI budgeting: where to protect and upskill employees
Chen says employers need to be more precise about where automation truly replaces work and where it should be treated as an assistive tool, striving for balance rather than simply reducing headcount and costs.
For example, he stresses that roles involving innovation and strategic judgment are hard to automate: “Product innovation or R&D can become AI-enhanced … but that kind of job cannot be replaced by AI.”
For employers and HR teams, this points to a different way of building labour budgets, Chen says: instead of short-sighted focus on how many roles to cut, there should be more investment in roles that will be AI‑enhanced rather than AI‑replaced.
This means new line items, not fewer – for training, upskilling and job redesign as well as software licences or cloud‑compute bills.
“More training, more investment towards the integration of AI [in] people’s work,” Chen says, “not just replacing people with AI, especially for important positions.”
Agrawal adds that reaping AI’s benefits is less about replacing tasks and more about orchestrating change around specific business outcomes. Rather than treating AI as a technology project, he recommends starting from clear performance goals and then backing into the talent plan.
This means HR is a necessary part of the equation.
“In other words, it’s not about technology, it’s about having a business target. Now you can’t achieve the business target without employing technology,” he says.
“Once we lay out the plan, then we work with the HR teams: ‘How are we going to reskill people in order to use the technology to deliver this outcome?’ Because, in many cases, the technology already exists — the tough part is the change management, and the change management is largely going to be in the hands of people who manage talent.”
