Energy management and innovation could be a key enabler and sustainable competitive edge for data centre development and related economic growth tied to smart technologies and artificial intelligence (AI). A new report sponsored by the MaRS Discovery District, a Toronto-based incubator for research, development and commercialization, models the potential for accommodating up to 3 gigawatts (3,000 megawatts) of additional data centre capacity in Ontario over the next 10 years, and explores how that could occur synergistically with other development demands.
Factoring in planned expansions to 2035, researchers conclude that Ontario’s electricity system could support up to 1.5 gigawatts (1,500 megawatts) of new data centre load plugging into the grid if it occurred in tandem with various conservation and demand management (CDM) strategies. Supplementary sources, such as on-site generation or district energy networks, would be needed to power data centre expansion beyond that.
The report outlines technologies and energy management practices that could play a role in both growth scenarios. It also makes the case for a coordinated, collaborative approach to facilitating data centre development to reduce rivalry for grid connections and local backlash against proposed projects. That includes a recommendation that the Ontario government undertake a data centre pilot project in a designated high-growth area to test and demonstrate the possibilities.
“We shouldn’t be viewing data centres as parasitic on the grid. They’re actually a grid asset. How we design them can make them more valuable as part of that,” Tyler Hamilton, senior director, climate, with MaRS, maintained during a recent webinar that delved into the report. “We want to put the policies and incentives in place today to get the kind of data centres we want in our communities over the long term — data centres that don’t increase the difficulty of managing supply and demand on the grid, but actually make it easier to manage.”
AI ambitions
Both the federal and Ontario governments have announced ambitions to foster AI-related opportunities in ways that strengthen Canadian autonomy. Last fall’s 2025 federal budget allocated nearly $926 million over five years for “sovereign public AI infrastructure” to underpin public and private research, and formulation of a Canadian AI strategy is now in progress. The 2026 Ontario budget likewise heralds a provincial AI plan, expected to be released this summer.
“That will support the scale-up of Ontario-based AI firms, expand access to sovereign compute and data resources and ensure the province has the digital and grid-ready energy infrastructure required for next-generation innovation,” the budget document states. “The strategy will also aim to increase adoption and deployment across communities and businesses in key sectors, with a focus on Ontario-made technology and talent.”
Authors of the MaRS report argue it is imperative to move quickly. Speaking alongside Hamilton, Kathleen Kauth, chief operating officer with the consulting firm, Mantle Climate (producer of the report), hypothesized that frenetic data centre development in the United States could be either a harbinger of relatively imminent trends here or a stealthy drain on Canadian resources and expertise.
“There are not going to be any major scientific innovation startup breakthroughs from here on out that don’t rely on heavy advanced compute. Canada needs access to advanced compute that we can control. We cannot lose our researchers to other countries because we don’t have access to advanced compute,” she asserted. “We need to innovate in a big way; we need to do it yesterday; we need to do it together; and it needs to be paired with massive urban development projects we’ve got planned over the next 10 years that all require hundreds of megawatts just for themselves.”
Proponents of energy management and innovation divide those objectives into short-term and long-term deliverables. Although many of the sustainable technologies envisioned to carry the bulk of future load are still embryonic or not yet economically competitive, in some ways they are an easier piece for strategists to place in the puzzle. Renewable energy, energy storage and AI energy management applications are largely categorized as eventual market dominators. The interim poses more risk for a precarious tilt to the expedient.
“It’s painful to see, but it’s true, this is on the back of natural gas in the short term,” Kauth acknowledged.
“We’re seeing a lot of inefficient natural gas generators that are being deployed as a kind of stopgap that are really bad for the environment,” Hamilton concurred. “The challenge is to not have these things become permanent.”
Efficiencies and offsets
Still, long waiting lists for gas turbines and the potential of cost-saving operational efficiencies spur uptake of a range of burgeoning and more common clean technologies. Hamilton cited: liquid immersion cooling; various forms of heat recovery, including converting it to electricity or directly into cooling; and examples of MaRS client companies with specialty focuses on reducing energy use through exclusive reliance on DC wiring and extracting water from fog.
“It will only work in certain environments, but data centres are testing this stuff out,” he advised, in reference to the water-from-fog endeavour.
As well, another MaRS client recently received an order from Microsoft for 300,000 tons of voluntary carbon dioxide removal offset credits over the next 10 years for its carbon mineralization technology — an example of offsetting short-term reliance on natural gas. Kauth commended many of the big tech players with “the most ambitious targets for positive climate outcomes in the world” for driving innovation in the sector, and theorized that could also work in favour of options like pairing data centre waste heat recovery with space heating for new developments.
“How do we pair this so that we only have to burn it once and then get compute and the heat — get double the productivity for the same amount of emissions?” she mused. “I do think there’s an appetite from these big data centre companies. They actually really do have strong governance internally on carbon emissions and they are willing to come to the table to talk about how to do that.”
Investment prospects
The Canadian government’s objective to draw more pension funds and institutional investors into the sphere might also align with those ambitions. The 2025 federal budget earmarked $1 billion over three years, beginning in 2026-27, to launch a proposed Venture and Growth Capital Catalyst Initiative through the Business Development Bank of Canada with an agenda to incentivize pension funds and institutional investors to provide more seed funding. For now, many of Canada’s richest pension funds — notably including Canada Pension Plan Investment Board (CPP Investments) — do have data centre strategies within their real estate and infrastructure portfolios, but much of that investment is occurring outside Canada.
Canada currently ranks fifth among global nations for data centre density, boasting about 330 facilities nationwide. With about 315 megawatts of existing data centre inventory and another 153 megawatts currently under construction the Greater Toronto Area is still far short of the 1 gigawatt baseline the MaRS report models for 2035. However, the assumption is based on a relatively modest increase in the 13 per cent annual growth rate of recent years.
Somewhat incongruously, data centres placed in the bottom half of the favourability ratings when CBRE Canada conducted its annual survey of Canadian lenders’ views on commercial real estate earlier this year. More than 20 per cent of respondents identified data centres as an asset class that caused concern, ranking it 13th among 22 property categories.
However, Joshua Sonshine, a senior vice president with CBRE Capital, stressed that’s not due to shaky fundamentals when he parsed out survey findings for attendees at the Real Capital conference in Toronto earlier this winter. Rather, lenders have faced obstacles in fulfilling their intentions for both data centres and life sciences facilities.
“Many lenders have had significant capital earmarked for these sectors that was not allocated in 2025 and is unlikely to be deployed in 2026 either. There is simply not enough construction or enough deal volume,” Sonshine said. “Lenders are coming to terms with those limitations in the Canadian market and their underwriting experience, but don’t get it twisted. If that product were available, lenders would be there.”
