Avoiding gridlock: how Canada can use AI to expand and improve its hydroelectric future

Hélène Deschamp-Marquis

Borden Ladner Gervais, Montreal

hdeschampsMarquis@blg.com

Matt Saunders

Borden Ladner Gervais, Montreal

msaunders@blg.com

Cléa Jullien, Associate

Borden Ladner Gervais, Montreal

cjullien@blg.com

Introduction

The rapid evolution of AI will dramatically affect Canada’s energy sector. Hydroelectric operators are no exception to this rule, particularly considering the benefits foreseen in the development of large data centres across the country to meet the surge in energy consumption resulting from the use of Gen AI. Although Canada has an opportunity to become a leader in the growth of these centres, along with the use of novel AI systems to improve hydroelectric operations, legal risks abound. This brief article reviews the current state of data centre development, legal and regulatory considerations, opportunities and risks for operators, and a forward-looking conclusion on how Canadian innovation – coupled with strong systems of governance – can take advantage of this dynamic shift in the technology landscape to ensure its grid does not get locked.

Current state of play

There is a global surge in data centre development to support the advancement of AI, cloud services, and cryptocurrency mining operations. Today, there are approximately 11,000 data centres registered around the world, 252 of which are in Canada. This rapid expansion comes with a significant increase in energy consumption.[1] According to the International Energy Agency, data centres and their data transmission networks accounted for around one per cent of global electricity consumption in 2022,[2] or in the range of 240 to 340 terawatt-hours, which is equivalent to 71 per cent of Canada’s electricity generation that same year.[3] In Canada, Hydro Québec anticipates that high-tech AI data centres will boost the demand for hydroelectricity by 14 per cent from 2023 to 2032.[4] Globally, AI energy consumption could double by the end of 2026.[5]

Legal and regulatory summary

The proliferation of data centres presents significant energy, real estate, and corporate legal challenges that Canadian developers, landowners, operators, and investors will need to address, as demonstrated in the following examples.

Energy

Applying for and negotiating the necessary agreements (eg, electrical utility agreements, transmission service agreements and permitting licences) is complicated. It also involves the transmission infrastructure to facilitate interconnection to the electricity grid, particularly as data centres consume substantial amounts of energy and require reliable power sources to operate continuously.[6]

Real estate

Site selection is a crucial factor for data centre success, driven by latency and electrical transmission requirements, land acquisition costs, local zoning laws, and land use regulations.[7]

Corporate

Structuring transactions and entities, and conducting appropriate due diligence, also requires consideration of multiple pieces of legislation, including tax, securities, privacy, and competition law.[8]

As securing data is paramount in the digital age, AI and data centres will play vital roles. An integrated approach, encompassing physical, technical, and operational security, is required. Particular attention will need to be paid to the legal regimes that apply to the collection, use and disclosure of personal information stored in these data centres. Internationally, the European Union is leading the way in the passing of regulatory requirements for AI and cybersecurity compliance – the recently passed AI Act being a prime example.[9]

Canadian lawmakers, however, continue to struggle in advancing federal legislation aimed at regulating cybersecurity for critical infrastructure and AI. Two bills addressing these issues, C-26 and C-27, were tabled more than two-and-a-half years ago by the federal government, had been reviewed and debated, and were close to being passed. However, with the recent prorogation of parliament, both proposed laws are now ‘dead’, and a new federal government will have to start the process all over again.[10]

Nevertheless, provincial governments have taken up the cause and are advancing their own regional laws to regulate cybersecurity and AI. This is particularly the case in Ontario, with its Bill 194 to regulate AI and cybersecurity in the public sector,[11] and in Québec, with its Law 25 aimed at revolutionising privacy, cybersecurity and AI law across the province.[12]

Opportunities and risks for the development of data centres in Canada

The rapid expansion of AI has significantly increased the demand for data centres in Canada, which in turn, has resulted in several new opportunities. These include economic growth, technological advancement, and renewable energy integration.

Investing in data centres can stimulate economic growth by attracting global tech companies and fostering innovation. The Canadian government’s consideration of up to CAD15bn in incentives to encourage domestic pension funds to invest in AI data centres powered by green energy emphasises the potential economic benefits of this sector.[13]

Canada’s abundant renewable energy resources, such as hydroelectric, wind, and solar power, offer a strategic advantage for powering data centres sustainably. Provinces such as Québec, with access to low-cost, renewable hydroelectric power, are particularly attractive locations for data centre investments. Leveraging these resources can reduce the carbon footprint of data operations and position Canada as a leader in green technology.[14]

However, data centres are not without operational challenges. These include energy consumption and grid strain, and their environmental impact. Data centres are energy-intensive, and their proliferation could strain Canada’s electricity grid. If all the data centre projects currently under review proceed, they would account for 14 per cent of Canada’s total power needs by 2030, potentially challenging grid reliability and increasing energy costs.[15] The substantial energy requirements of data centres may lead to increased greenhouse gas emissions, especially if fossil fuels are employed to meet energy demands. Without intervention, data centre emissions could rival those of major global industries, exacerbating climate change.[16]

Risks and opportunities for operators: reducing the chance of a ‘HAL’ moment

The development and deployment of AI for use by Canadian hydroelectric operators create significant avenues for efficiency and improvements to the grid. However, several risks arise when machines start taking over from humans.

For example, there is interest as to how AI can predict the timing of equipment maintenance, which in turn, can significantly reduce legal and reputational risks from services being down. When advanced machine learning (ML) systems can ‘see’ and process massive amounts of data points as they relate to part installation and shelf-life, these tools can better detect equipment that is at risk of failure, identify the parts that need replacing, and automate the lifecycle, greatly reducing the risk of breakdowns and outages.[17] Similar tools and strategies can also be deployed to improve cybersecurity preparedness and maintenance with autonomous, real-time protection against threats like zero-day vulnerabilities.[18]

Worker safety is also likely to see improvements, due to enhanced visibility over staff in remote sites or performing potentially hazardous tasks. AI can either take over the dangerous roles (eg, technical repairs completed through AI-enhanced robotics and human oversight), or enhanced surveillance technology can help better monitor the workers engaged in risky activities on the grid.[19] There are, of course, various workplace and labour regulations which will have to be considered given the invasiveness of such surveillance over employees.[20]

Operators and governments looking to advance AI investment into the hydroelectric sector, however, must also consider potential pushback from residents who may challenge the development of data centres in their locality due to the potential for major rate hikes to cover the surge in energy demand.[21] These concerns have already become a reality in the US where the federal energy regulator refused an application from one of the world’s largest cloud services provider to purchase power from a Pennsylvania nuclear power plant on the grounds that it would raise customer rates, and threaten the reliability of the local grid.[22]

Conclusion: a future powering more than chatbots

It is not science fiction to consider a future where AI reduces legal risks and results in the application of novel ideas to improve and redistribute hydroelectric energy more efficiently, particularly through data centre development. Consider the following three examples:

AI could enable ‘self-healing’ hydroelectric grids

Testing has already begun in the US through the Grid Modernization Lab Consortium, where the implementation of ML algorithms – in real-time – to identifying issues leading to the redirection of energy flows, has resulted in the cutting of outage times by 40 per cent.[23]

AI-microgrids

These are independent energy systems that operate alongside the main grid, allowing for improved resilience in the case of natural disasters or cyberattacks. Their development is expected to increase by 30 per cent by 2030, especially in developing nations scaling up renewable energy sources.[24]

EV charging schedules

AI could improve the charging schedules of electric vehicle (EV) stations, allowing more opportunities for vehicle-to-grid (V2G) technologies to flourish. In this way, EVs will not only benefit from being on the road with no emissions, but they will also leverage AI to control the flow of energy from stationed EVs back to the grid when additional electricity is needed most.[25]

Insurance policies for operators will also need to evolve, particularly when it comes to addressing cyber risks. New AI systems deployed across Canada’s hydroelectric grid could bring about significant efficiencies to operators and their organisations (while also strengthening the scope of visibility against cyber risks), resulting in a reduction in premiums. The policies are also likely to benefit from the insurance industry’s deployment of AI tools to improve modelling and predictive analytics on disasters affecting the distribution of electricity to Canadians, leading to improved mitigation strategies.[26]

Finally, Canada, has an opportunity to marry its investment in the development of data centres and the power grid needed to keep their lights on, with a renewed approach to mining the rare earth metals essential for their construction. Instead of focusing solely on the buildings that will house these centres, as well as how to ensure that there is sufficient grid capacity once they are plugged in, the extraction of the raw materials necessary for the components of data centres should be foremost in mind for government officials and regulators. Facilitating their extraction, while limiting risks to the environment and surrounding communities, could turn Canada into a ‘one-stop-shop’ for the building, housing, and operating of data centres which will be indispensable to sustaining the future AI revolution.[27]