Hydro-Québec, the largest utility in Canada, has taken an ambitious path to adopt artificial intelligence (AI) in all its load forecasting practices. This effort will help increase the reliability of our grid and avoid future blackouts, especially as demand patterns are more erratic than ever before. Through its work with the city, the utility company was able to identify the weaknesses of its conventional forecasting methods. These models inaccurately account for rapidly increasing energy needs during peak days. Hydro-Québec powers up on cutting-edge artificial intelligence technologies to bridge the divide between predicted and real-life energy use. This effort will hopefully contribute to a more efficient, reliable and resilient energy grid.
On an average day, real world usage shows that legacy models and AI models produce fairly comparable forecasts. On those extreme days, like on May 22, 2024 during a multi-day heatwave, the differences are clear. During this stretch, one of Hydro-Québec’s legacy models failed to predict a bizarre surge in demand. Due to this mistake, the company was required to implement extensive 1,500 megawatt (MW) corrections. By comparison, the AI model successfully predicted the demand spike, demonstrating the model’s ability to provide more accurate predictions.
Hydro-Québec is committed to increasing its own in-house AI capacity. They are rolling out a Continuous Improvement Program for 2026-2027. The firm plans to take advantage of smart meter data. As part of this initiative, they’re creating a prototype for renewable energy forecasting.
Challenges of Traditional Forecasting Models
Hydro-Québec’s conventional load forecasting models have worked for the utility over the years. As energy use patterns become more variable, they’re becoming less effective. Aroon Vijaykar, an electric power industry expert and practitioner, puts it succinctly when he stresses that “load forecasting is key.” As he emphasizes, with the exciting new possibilities of AI come dangerous new challenges.
“Load forecasting has always been an important function. AI is both a challenge and an opportunity on both sides of that equation,” – Aroon Vijaykar.
As Vijaykar clarifies, the increasing expectation for precise forecasts only adds to the challenges of maintaining load balance. To state one of Mr. McNamee’s big themes AI can help fine tune operations efficiency and prevent huge blackouts from happening. According to a Hydro-Québec spokesperson, the real challenge would be in managing the data coming from over four million smart meters. Managing this data properly is necessary to improve forecasting accuracy and precision.
“Having to deal with data from more than four million smart meters is another ball game,” – Hydro-Québec’s spokesperson.
Sylvain Clermont, who was then one of Hydro-Québec’s forecasting team leads, explains this point in more detail.
“Then comes something totally out of the box – whether extreme weather or something else that you have never experienced in the past – and your model will be off,” – Sylvain Clermont.
The Role of AI in Modern Forecasting
AI’s entry into Hydro-Québec’s forecasting workflow is a significant turning point in the utility’s evolution in the way they manage loads. On the traffic side, the AI system provides real-time and 36-hour short-term traffic forecasts. It’s the first to offer hourly forecasts up to 10 to 12 days ahead, leveraging the most advanced meteorological data available. For even longer-range forecasts, Hydro-Québec uses historic normal weather values so that the long-term view of energy consumption trends can be seen more clearly.
Hydro-Québec is currently preparing for a more bottom-up, regional, approach that will kick off in 2028. To begin with, they anticipate being able to load forecast at over 350 substations. This unique approach seeks to leverage the capabilities of AI in order to enhance precision throughout its extensive service area.
“In the future, AI is expected to facilitate autonomous grid operations, optimise energy flows, and enable seamless integration of distributed generation and storage,” – Source: https://www.power-technology.com/features/key-themes-2025-what-data-centres-tariffs-and-grid-bottlenecks-mean-for-the-energy-transition/.
Additionally, Hydro-Québec understands that even though AI shows great potential in load forecasting, it is not a full substitute for classic methods. As David Adkins, Director of Policy and Strategy at GPI, highlights, the integration of AI-driven orchestration with traditional grid planning and human operational oversight will be key.
“AI complements but does not replace grid planning and forecasting. AI-driven orchestration works best alongside grid forecasting, market signals and human operational oversight,” – David Adkins.
Looking Ahead: The Future of Load Forecasting
As Hydro-Québec moves forward with its AI initiatives, the company remains focused on reducing discrepancies between forecasts and actual energy consumption. Ultimately, we want to improve system efficiency, taking the risk of costly major disruptions due to false positives off the table.
As Vijaykar emphasizes, utilities are naturally risk-averse when it comes to new technologies because there are major consequences if they try something and it doesn’t work. He points out that while tech companies can move quickly to innovate, utility companies require thorough testing before adopting new solutions.
“Utilities are understandably conservative about the system. Unlike the tech industry, which can afford to move fast and break things, for utilities one mistake can be catastrophic,” – Vijaykar.
Clermont strongly backs this up by pointing out the lengthy training process that is necessary for all AI systems. He’s confident as Hydro-Québec’s faith in AI grows, it will be able to eliminate older models altogether.
“When they get confident, Hydro-Québec will stop using the old model. It is a training question; the AI needs to be trained,” – Sylvain Clermont.
Hydro-Québec is excited to lead the charge on changing how we think about load forecasting. This change is happening at real transformational breakthrough times across a complicated energy future. Beyond increased accuracy, the incorporation of AI holds the key to a more robust resilience to unexpected challenges.