AI presents opportunities and challenges in the energy landscape

Ginelle Greene-Dewasmes, Initiatives Lead, Artificial Intelligence and Energy, World Economic Forum and Thapelo Tladi, Lead, Energy Initiatives, World Economic Forum write on the WEF website about AI’s opportunities and challenges. What are your views?

AI’s energy dilemma: Challenges, opportunities, and a path forward

• The energy demand of data centres, including hyper-scale facilities and micro edge deployments, is projected to grow from 1% in 2022 to over 3% by 2030.
• AI is already helping companies reduce energy use by up to 60% in some instances. Key use cases include optimizing energy storage, battery efficiency, and smart grid management.
• Coordinated efforts are needed to enable sustainable AI adoption across industries. Key focus areas for action include regulation, financial incentives, technological innovation and market development.

While there have been numerous forecasts around the energy demands of artificial intelligence (AI) and the efficiency gains it will unlock, it is hard to predict these with certainty, given the rapidly evolving landscape.

A recently published white paper from the World Economic Forum titled Industries in the Intelligent Age – Artificial Intelligence’s Energy Paradox: Balancing Challenges and Opportunities, suggests four key interlinked areas for navigating this uncertainty, managing challenges and unlocking opportunities for sustainable AI deployment.

These include:

• Leveraging AI deployment for decarbonization.
• Transparent and efficient AI energy use.
• Innovation in technology and design.
• Effective ecosystem collaboration.

AI’s energy consumption

AI presents opportunities and challenges in the energy landscape. With around 72% of surveyed companies leveraging AI for at least one business function, its transformative potential is clear.

According to aggregated estimates from Accenture based on data from Goldman Sachs, the International Energy Agency and the Organization for Economic Co-operation and Development, accompanying this rise in adoption, AI-related electricity consumption can be expected to grow by as much as 50% annually from 2023 to 2030, posing a challenge to power systems.

The electricity demand of data centres, from hyper-scale facilities to micro edge deployments, is projected to grow from 1% of global energy demand in 2022 to over 3% by 2030.

However, such projections can vary. Uncertainty remains around how profound AI’s overall energy impact will be and which strategies could mitigate challenges that arise or enable new solution opportunities.

Despite AI’s rapid expansion, AI data centre electricity consumption will still likely account for only a small fraction of global electricity demand.

However, when combined with other major demand drivers (such as the electrification of transport and buildings), it can still contribute to an increased strain on power grids and energy providers.

To address this, strategies such as energy-efficient hardware, AI-optimized cooling, and smarter data centre design and operations are being explored to limit AI’s energy consumption. Moreover, advancements in chips and algorithms (e.g., small language models) may further mitigate AI’s energy consumption.

External link