IT Sustainability Think Tank: The energy challenge of AI datacentres in 2024 and beyond
The soaring enterprise demand for AI services is causing all sorts of problems for datacentre operators and their sustainability goals, which will need addressing in 2025
The rise of generative AI (GenAI) is reshaping industries, but it’s also creating an unprecedented demand for energy.
Datacentres supporting AI workloads are expanding at a breakneck pace to meet the needs of increasingly complex large language models (LLMs).
The issue isn’t just about availability. AI datacentres are forecast to increase electricity consumption by 160% within the next three years. Such a surge threatens to overwhelm utility providers, disrupt energy availability, and undermine sustainability goals as fossil fuel plants remain in operation longer to keep up with demand.
Powering innovation or fueling a crisis?
The insatiable energy appetite of hyperscale datacentres is outstripping the ability of power grids to cope. AI models require immense computational power for training and operations, making 24/7 energy availability essential.
Yet renewable energy sources like wind and solar are not yet capable of providing the reliable baseline power needed without significant advancements in energy storage.
This mismatch between demand and supply has far-reaching consequences. Rising energy costs will drive up operational expenses for AI products and services, impacting organisations across industries.
Meanwhile, the concentration of datacentres in regions like Ireland and Singapore is already forcing local governments to limit their expansion due to power constraints.
Sustainability at stake
The strain on energy grids is having a knock-on effect on sustainability goals. In the short-term, many datacentres will need to rely on fossil fuels, increasing their carbon footprints and delaying progress toward net-zero targets. While longer-term solutions like advanced battery storage or modular nuclear reactors hold promise, they are not yet viable at scale.
For organisations committed to sustainability, this presents a dilemma. Balancing the deployment of energy-intensive GenAI applications with environmental responsibility requires innovative approaches, such as adopting smaller language models, leveraging edge computing, and collaborating with datacentre providers to optimise energy use.
Rethinking the future of AI
The current trajectory of AI-driven innovation poses hard questions for business and IT leaders. How to sustain growth in the face of energy constraints? What strategies will mitigate the impact of rising costs and environmental pressures? Organisations must prioritise efficiency in AI workloads, re-evaluate sustainability goals, and actively support the development of greener energy alternatives like clean hydrogen and small nuclear reactors.
As the demands of GenAI reshape the global energy landscape, success will require more than just technological prowess. It will demand foresight, collaboration, and a willingness to innovate sustainably.
Read more from the IT Sustainability Think Tank
- The past 12 months have seen companies row back on their sustainability commitments, in response to economic pressures, but could doing so harm their bottom line in the long run?
- As the global transition towards developing low-carbon economies continues apace, Gartner shares its take on the actions enterprises must take now to navigate an increasingly volatile energy landscape.
- The hype around AI is increasingly being matched with discussions about how the technology's adoption will affect the environment, so what can IT leaders do to ensure they keep the companies they work for on the forefront of innovation, without compromising the environment - or their firm's own corporate sustainability agenda?
- The pursuit of sustainability is not without its challenges but, with the right strategies and a dedicated approach, it is an attainable goal for enterprises in 2025
