At a glance:

Ontario’s new Data Centre Playbook will be open to a 30-day public comment period. There’s time for the public to ensure future AI data centres are operated in the public interest. The draft playbook falls short in these key ways:

  • It doesn’t guarantee municipalities the right to turn down a data centre proposal because it gives the provincial government the final say. It also fails to put limits on big lobbyists’ power to manipulate municipalities into saying yes to data centres.
  • It doesn’t guarantee full transparency about who owns the data centre, what it will be used for, or how many resources (like electricity and water) it will consume.
  • It doesn’t establish binding limits on the use of fossil fuels as behind-the-meter power sources, cooling practices, waste disposal, or noise—problems that are plaguing communities that already have data centres.
  • It doesn’t do enough to ensure that the economic benefits of AI data centres are captured by Ontario workers, businesses and communities.

This analysis identifies ways to improve Ontario’s data centre playbook and provides tools for the public to judge whether an AI data centre in your community is right for you.

Who benefits from AI data centres in Ontario?

Ontario says it wants to attract the “best” data centres. But best for whom?

On August 13, the Ontario government released a draft of its Data Centre Playbook alongside a 30-day public comment period through the Environmental Registry of Ontario (ERO) and Ontario Regulatory Registry (ORR). The playbook presents data centres as a cornerstone of the future economy, arguing they can strengthen Canadian digital sovereignty, create jobs and generate tax revenue. Ontario is positioning itself as a discerning host, promising to only select projects that offer the greatest economic benefits while meeting the highest environmental standards.

The rapid expansion of hyperscale data centres elsewhere has shown Canadians what can happen when governments prioritize development without establishing adequate safeguards. In the United States, electricity prices have risen more than 200 per cent for some residents. The use of behind-the-meter gas turbines has raised concerns about increased emissions and poor air quality. Constant low-frequency noise has disrupted sleep and quality of life. Community involvement has been minimal. Developers have lobbied local officials, used non-disclosure agreements to sidestep public consultation, and targeted rural jurisdictions that have less political power to resist when they come looking for cheap land, electricity and water.

Premier Doug Ford has dismissed growing public opposition to data centres as NIMBYism and misinformed social media politics. The province announced the playbook as a good faith commitment to ensure Ontarians capture the benefits of data centres while bearing the fewest possible costs. The playbook’s existence is itself a notable step. It acknowledges that the current policy vacuum, and the near total absence of minimum guidelines, cannot continue. 

As data centre proposals move through municipalities across the country, all levels of government will increasingly face pressure to establish common standards for how these facilities are approved and regulated. So, what would a public interest data centre policy framework require? Drawing on our research into the data centre boom and its impacts, we assess the Ontario draft against four basic thresholds.  

1. Does it require full transparency?

A framework meets this threshold if the public can find out, by default and before approval, who owns the data centre, what it will be used for, and the types and quantities of resources it will consume. This information should be publicly available throughout the facility’s operation and should not require a freedom of information request.

Ontario’s playbook describes two internal assessments (the Strategic Priority Assessment and the System Impact Assessment) that the government will use to evaluate proposed projects. However, it does not specify whether or when the results will be made public. Nor does it mention requiring developers to disclose their ownership structures, publish records of meetings between proponents (or the lobbyists representing them) and government, or prevent non-disclosure agreements (NDAs) from shielding important financial or environmental information from public scrutiny.

This transparency gap is particularly significant given the playbook’s emphasis on achieving digital sovereignty. It is important to note that a Canadian-owned data centre on Canadian soil does not, on its own, guarantee digital sovereignty. Canadian data can still be exposed to U.S. law under the CLOUD Act, depending on whether the provider also services U.S. clients or relies on U.S. software. Canadian ownership tells us very little about what kinds of data a facility is processing or who may ultimately have access to it.

Assessment: As currently drafted, Ontario’s playbook does not meet the transparency threshold.

2. Does it establish real environmental limits, or just commit to using better technology?

A framework meets this threshold when it sets enforceable limits on environmental impacts and clear restrictions on how data centres meet their electricity needs. Governments are notorious for treating newer technologies as a substitute for actual limits. Closed-loop cooling, waste heat recovery and renewable energy sourcing may all be improvements over the alternatives, but none establishes a limit on how much water, electricity or other resources a data centre can consume.

Ontario’s playbook gets one important piece right: it requires data centres to pay the full cost of their electricity, including generation and transmission upgrades, rather than passing those costs onto other ratepayers. But it says much less about how data centres will meet their electricity needs when grid capacity is constrained, or how they will maintain continuous operations when grid power is unavailable.

For facilities designed to operate 24 hours a day, seven days a week, backup power is a central operational requirement. A data centre can pay the full cost of connecting to the grid while still relying on gas, diesel or other carbon-intensive backup sources to provide additional or emergency power. Even limited fossil-fuel use can produce significant emissions when multiplied across hyperscale facilities. If backup systems are regularly tested or used during periods of peak demand, they become a source of emissions that runs counter to Ontario’s clean electricity narrative.

If Ontario wants to promote its clean and reliable grid as a cornerstone of sustainable data centre development, its playbook should also set limits on behind-the-meter generation and fossil-fuel backup. For example, after a four-year moratorium, Ireland now requires data centres to meet 80 per cent of their annual electricity demand from additional renewable energy sources. 

Meanwhile, the operational impacts of data centres are brushed over through technological solutionism. The playbook presents closed-loop and direct-to-chip cooling as clear environmental improvements, emphasizing that they can reduce water consumption to “less than a golf course”. New cooling technologies can reduce one environmental impact while creating or shifting others. For example, closed-loop cooling uses significantly more energy and relies on chemicals such as glycol and anti-corrosives that require management and periodic discharge. In Cheyenne, Wyoming, wastewater from a closed-loop system at a Meta facility was discharged into the city’s water system, forcing the municipality to adopt stronger regulations for its disposal.

The same logic applies to noise. The playbook’s appendix refers to “noise-reduction engineering” but does not establish clear decibel thresholds, setbacks from residential areas, or standards calibrated for continuous operation. Quieter equipment and noise mitigation are important (yet costly) design choices, but a framework should establish a minimum standard that the public—and local governments—can use to hold developers accountable.

Assessment: Ontario’s playbook falls short on this threshold. While the draft sets clear rules on electricity cost recovery, it does not establish binding limits on the use of fossil fuels as behind-the-meter power sources, cooling practices, waste disposal, or noise. 

3. Can communities hosting these projects actually say no?

A framework meets this threshold when communities and local governments have a meaningful role in deciding whether a data centre proceeds, not simply a role in negotiating the terms of a project that has already been prioritized by higher levels of government. Community benefit agreements are good, but they are not a substitute for the ability to refuse a project.

Ontario’s playbook raises questions about who gets to make these decisions. Under the Protect Ontario by Securing Affordable Energy for Generations Act, the province has given itself final authority over which large-load projects, including data centres, can connect to the electricity grid. The playbook says these projects will be assessed through a “whole-of-government” process that weighs grid considerations alongside economic development, community investment and public confidence.

But what does this mean for municipalities? If the province has the final say over whether a data centre can connect to the grid, what meaningful decision-making power remains with the communities that will host it? The playbook does not answer this question.

Municipalities are already facing significant pressure when trying to slow or stop data centre development. Hamilton’s recent experience, where city council reversed course on a proposed moratorium amid lobbying and the threat of an Ontario Land Tribunal appeal, demonstrates how difficult it can be for municipalities to protect the public interest when faced with well-resourced lobbyists.

Assessment: Ontario’s playbook does not currently clear this threshold. If anything, it risks centralizing more decision-making power within the provincial government without clearly defining or protecting the role of municipalities in determining whether and how these projects proceed in their communities. It also fails to ensure strong checks and balances against big lobbyists.

4. Are the proposed economic benefits legitimate, or just plausible?

A framework meets this threshold when jobs, tax revenue, local procurement, compute access and supply chain participation come with binding commitments, specific allocations and reporting requirements. It should not simply assume these benefits will follow from large investments.

Ontario’s playbook treats community benefit agreements as central to project approval, which is a good starting point. But a community benefits agreement (CBA) only provides material benefits when communities have enough leverage to negotiate them, and the previous threshold shows how limited that leverage can be. The playbook’s other economic promises have a similar problem in that they are described as desired outcomes rather than secured through specific mechanisms.

For example, Pillar 1 of the Strategic Priority Assessment identifies helping Ontario businesses become “more productive and innovative, including through compute for Ontario SMEs [small and medium enterprises]” as a measure of success. Yet the playbook does not yet explain how this would work. There is no requirement for data centres to allocate compute capacity to Ontario SMEs, no pricing commitment and no mechanism for access. Without these measures, “access to compute” assumes that building data centres in Ontario will somehow translate into affordable compute for Ontario businesses. It’s not a policy commitment.

The approach to supply chains is similar. Pillar 1 calls for data centres to “support Ontario and Canada’s domestic supply chains and local supplier participation where feasible.” But “where feasible” leaves the commitment largely to the discretion of developers. More importantly, much of the material and equipment required to build and operate data centres, including servers, chips and other specialized hardware, is sourced through global supply chains. Canada has limited domestic capacity to produce many of these components. It is, therefore, difficult to know how much of this investment in data centres will actually remain in Ontario or Canada.

Furthermore, the 17,000 new jobs annually cited in the draft is not broken down by role, wage tier or permanence. It is unclear how many of these are long-term operational jobs or whether they will offset losses elsewhere in the labour market as AI adoption accelerates the potential for automation.

Assessment: Ontario’s playbook does not currently meet this threshold. It identifies economic benefits that could result from data centre development, but does not do enough to ensure that those benefits are captured by Ontario workers, businesses and communities.

What kind of AI economy is Ontario choosing to build? 

The four gaps identified in this analysis illustrate how Ontario is approaching data centre regulation through a technical and economic lens. The draft is primarily designed to make data centre development possible, not to ensure the AI economy underpinning the data centre boom is serving the public interest.

The playbook is framed almost entirely in economic terms, emphasizing $122 billion in projected growth, new jobs, and the goal of making Ontario the most competitive economy in the G7. Yet this does not account for the jobs AI may displace, or the  workers whose jobs are deskilled, surveilled or degraded by the AI systems these data centres are built to train and operate.

This narrow focus on data centres as an economic investment opportunity ignores how these projects are increasingly experienced by the communities that host them. Data centres have come to represent something much larger than a piece of infrastructure or a utility demand. For communities opposing these projects, they can represent the material reality of the AI economy: workplace disruption, the concentration of economic power, the erosion of privacy, and the growing role of private technology companies in public institutions and everyday life. These consequences are top of mind for many Canadians, and have yet to be adequately addressed.

At the moment, governments have failed to demonstrate that they can responsibly manage the electricity, water and economic impacts of data centres, let alone the larger questions about what that data centre makes possible. The growing movement opposing data centres is calling for more information and transparency—to ensure a well-informed public, and their governments. 

If governments want to build public confidence in AI, public interest policies and thresholds governing the infrastructure that enables it need to move alongside public interest policies governing how AI is developed and deployed. This requires all levels of government to ensure data centre policies are matched by stronger protections for workers, robust public education, personal privacy, information integrity, public sector capacity, environmental stewardship, and other areas where AI is already creating social risks.

A sustainable data centre strategy cannot be separated from the question of what kind of AI economy Ontario is choosing to build, and who that economy is meant to serve. And it’s the public that will pay the price.

About the author

Rachel Pettigrew

Rachel Pettigrew (she/they) is a research assistant at the Canadian Centre for Policy Alternatives. Drawing on a degree in sociology and a Master of Social Work, their work combines policy research and advocacy to advance the interests of workers, communities and the environment amid political, economic and technological change.