WEF's "Net-Positive AI" Plan Aims to Save Power Grids from AI Energy Crunch

"By 2035, global data centre electricity use could exceed 1,200 terawatt-hours (TWh), up from 420 TWh in 2024. Without strategic intervention, AI could become a hidden contributor to system stress and climate risk", the report stated.

The report discusses the concept of "Net-Positive AI Energy," a state where the energy savings and efficiencies generated by AI applications exceed the electricity consumed during their development and operation.

The report also talks about the Jevons paradox - "As AI becomes more accessible and its use expands, a key challenge will be ensuring efficiency gains create real value rather than triggering even more AI use that cancels out the energy and resource savings"

This approach aims to address the "Jevons paradox," a phenomenon in which efficiency gains in technology often lead to increased usage, thereby negating any original savings. By shifting focus from raw computational growth to an impact-first paradigm, the Forum argues that AI can become a strategic asset rather than a liability, enhancing grid resilience and driving down operational costs across the global economy.

The report also identifies several "hidden drivers" of AI's environmental footprint, specifically highlighting the issue of "dark data", unused information stored in energy-hungry servers, and the technical inefficiencies of current model training. To combat these, the framework outlines three primary action drivers: designing for efficiency, deploying for impact, and shaping demand wisely. Currently, only 10% of AI use cases reflect a "demand-shaping" approach, which involves scheduling workloads during periods of low grid stress or high renewable availability. Closing this gap is essential to preventing a "net-positive divide," where only technologically advanced regions reap the benefits of the AI revolution while others face increased digital inequality and resource scarcity.

Real-world evidence suggests that the transition is already feasible. The report cites successful deployments where AI reduced data center cooling energy by 40% and improved industrial maintenance, saving millions in operational costs. In the UK, predictive maintenance models have allowed energy companies to detect equipment degradation early, preventing costly outages and optimizing labor. Furthermore, AI-enabled forecasting is helping grids integrate variable renewable sources like wind and solar with over 90% accuracy.

Achieving a net-positive future will require unprecedented collaboration between technology providers, governments, and industry leaders. The Forum emphasizes that transparency is the first step, advocating for standardized "sustainability labels" for AI models and public dashboards to track energy consumption. As the global community navigates this transition, the report concludes that the goal is not to slow down AI development, but to "design a better engine"--ensuring that the intelligence of the future is powered by a sustainable and equitable energy foundation.

- ANI