Predicting the surge: How Cotality’s data powers the AI revolution in load forecasting

The failure of the "historic normal"

Traditional load forecasting involves matching current patterns to a mathematical model based on years of previous load curves. Sylvain Clermont, lead author of the UNECE Task Force on Digitalization in Energy, explains that while these models are excellent for regular patterns, they fail during "out of the box" events, such as extreme weather or the sudden shifts in demand seen during the pandemic.²

Real-world data from Hydro-Québec highlights this risk. During a heatwave on May 22, 2024, their legacy models failed to anticipate that the grid would not experience its typical load decrease. This required human operators to make "significant" manual corrections of 1,500MW. In contrast, their AI model, which had been in development since a 2018 proof of concept, successfully predicted the anomaly.

Bridging the gap

To achieve the accuracy demonstrated in the Hydro-Québec case study, AI requires more than just historical consumption logs; it needs real-time indicators of where and how a region is growing. While GlobalData reports that smart meter penetration has reached 70–90% across the US, China, and the EU, simply collecting data from "more than four million smart meters" is only half the battle.

The real transformation occurs when grid operators move from reactive legacy models to proactive orchestration using Cotality’s location and growth data. By feeding Cotality’s granular datasets into AI models, energy companies can address three critical modern challenges:

• Hyper-local substation planning: Cotality identifies future neighborhood developments in high-growth areas by leveraging ground-truth indicators. We track new construction through its entire lifecycle—from planning and development to recent completion. This allows utilities like Hydro-Québec, which aims to implement a "bottom-up" regional approach for over 350 substations by 2028, to predict "load hotspots" at individual nodes before the demand ever hits the wires.
• Managing the AI data center boom: As Aroon Vijaykar of Emerald AI points out, AI complicates load forecasting due to the immense demand generated by data centers. Cotality provides insight into these high-intensity developments by tracking permitting and construction timelines. This enables tools like Emerald AI’s Emerald Conductor to manage "grid flexibility" and scale non-critical workloads based on real-world development status, as seen in National Grid’s 2025 trials.³
• Adapting to renewable variability: David Adkins of National Grid notes that the "increasing penetration of renewables" introduces significant uncertainty. Cotality’s data identifies specific neighborhoods, retail centers, and other points of interest (POIs) undergoing rapid modernization. This ground-level context allows AI to perform the "real-time analysis of complex, multi-source data" that Adkins identifies as essential for grid stability.⁴

The path forward

The transition is not about replacing human oversight but enhancing it. As National Grid representatives emphasize, AI works best as part of a "broader flexibility and resilience toolkit"⁵. By leveraging Cotality’s growth and location data, energy companies are no longer just looking in the rear-view mirror; they are gaining the critical foresight needed to integrate intermittent renewables, manage the AI revolution, and avoid blackouts.

Grid stability shouldn't be a guessing game. Traditional models are failing to keep pace with 2026's volatile demands, but with property-level intelligence, utilities can predict hotspots before construction even begins.