Powering artificial intelligence
A study of AI’s footprint—today and tomorrow
Deloitte Global analysis estimates that global electricity use by data centres will nearly triple in the coming decade, necessitating a corresponding increase in renewable energy to avoid negative impacts on climate. This growth is driven by AI technologies permeating our economy and society, which is driving substantial growth in data and computational needs. To manage AI’s environment footprint, stakeholders must work together to ensure its development aligns with global sustainability goals.
This study focuses on potential ways to achieve “Green AI” in the coming decades and answers the critical questions emerging around the energy consumption and climate impacts of AI.
- What are the current trends, key drivers and regulatory frameworks that will help shape the evolution of AI and the energy consumption of data centres?
- How might the energy consumption of data centres evolve and what would the resulting carbon footprint be?
- What strategies can businesses and governments implement to help mitigate the environmental impact of AI-related energy consumption on climate?
Key findings
AI growth could double its electricity consumption in the next two years
In recent years, the AI market has grown into a US$200 billion industry - and it’s still growing, which could double electricity consumption over the next two years. Although AI and related technologies hold the potential to accelerate the transition to climate-neutral economies, the growth in electricity demand from data centres risks offsetting these gains and delaying the energy transition if not appropriately managed.
It's important to actively manage emissions to mitigate AI’s footprint
Energy consumption of data centres is estimated to hit 1,000 TWh in 2030 and about 2,000 TWh in 2050, which is about 3% of global electricity consumption. While in comparison to other electricity uses like e-mobility, electric heating and green hydrogen that far outpace growth in data centre power consumption, it’s important to actively manage the emissions for the estimated data centre growth to meet global emissions targets.
Achieving “Green AI” is essential to long-term success
“Green AI” relates to the efforts made to reduce AI’s environmental footprint by adopting leading practices across the entire value chain—from clean energy supply purchases to hardware improvements. Achieving “Green AI” will require not just renewable energy expansion, but also reporting transparency, an ecosystems approach and a commitment to efficient AI development and operation, so the AI boom can happen without potentially jeopardising its environmental footprint. This Deloitte Global study focuses on potential ways to achieve “Green AI” in the future.
There are several ways to design a sustainable energy landscape
Innovations in AI and data centre efficiency will be pivotal in shaping a sustainable energy landscape. Energy demands on data centres can be carefully managed in several ways, from improving the performance of IT and hardware equipment, to implementing a comprehensive energy management system to optimise the centre’s use. In addition, using clean energy will support climate neutrality - along with phasing out or retrofitting thermal assets.
Opens in new window
Businesses and governments play a critical role
Businesses and governments are crucial in reducing AI’s climate impact by adopting sustainable practices, aligning market incentives with operations, investing in new technologies and supporting favourable regulatory frameworks.
This study helps stakeholders ensure AI development aligns with global sustainability goals, advocating for collaboration to harness AI’s transformative potential while protecting the environment.
Prof. Dr. Bernhard Lorentz
Geoff Tuff
