- Projected power requirement for data centres running AI-optimised servers is expected to reach 500 terawatt-hours annually by 2027, a figure that is 2.6 times higher than the levels recorded in 2023.
- Intersection of AI, GenAI, and energy consumption presents significant challenges that require immediate and strategic attention.
- As the demand for data centres escalates, it is imperative for organisations to navigate the complexities of power availability, cost management, and sustainability to ensure the continued growth and responsible deployment of AI technologies.
The rapid advancement of artificial intelligence (AI) and generative AI (GenAI) is significantly impacting electricity consumption, particularly within data centres.
According to research firm Gartner, Inc, forecasts indicate a staggering 160 per cent growth in electricity usage by these facilities over the next two years. The surge in power demand raises critical concerns regarding the sustainability of energy resources and the operational viability of AI infrastructure.
Bob Johnson, VP Analyst at Gartner, highlights that the burgeoning need for hyperscale data centres to support GenAI applications is outpacing the capacity of utility providers to expand their services.
By 2027, it is anticipated that 40 per cent of existing AI data centres will face operational constraints due to power availability. Such constraints threaten to disrupt energy supply, potentially leading to shortages that could stifle the growth of new data centres essential for advancing GenAI technologies.
The projected power requirement for data centres running AI-optimised servers is expected to reach 500 terawatt-hours (TWh) annually by 2027, a figure that is 2.6 times higher than the levels recorded in 2023.
Increased electricity prices
The exponential increase is primarily driven by the substantial data demands associated with training large language models (LLMs). However, the infrastructure needed to support this surge in power demand is lagging, as new power generation and transmission capabilities may take years to implement.
Gartner’s analysis suggests that the imminent power shortages will lead to increased electricity prices, which will consequently elevate the operational costs of running LLMs.
Organisations reliant on these technologies must proactively assess the potential impact of power shortages on their products and services.
Johnson advises that significant power consumers should secure long-term contracts with energy producers to ensure reliable access to electricity, independent of broader grid demands.
Moreover, the pursuit of zero-carbon sustainability goals is likely to be compromised as energy suppliers scramble to meet the rising demand. In many cases, this urgency has resulted in the extended operation of fossil fuel plants previously slated for retirement, thereby exacerbating carbon dioxide emissions.
The reliance on renewable energy sources such as wind and solar is insufficient to meet the continuous power demands of data centres, necessitating the use of hydroelectric, fossil fuel, or nuclear power for reliable energy supply.
Alternative power solutions
In light of these challenges, Gartner recommends that organisations reevaluate their sustainability objectives concerning CO2 emissions, considering the evolving requirements of data centres and their energy sources.
The future landscape of AI and GenAI will be shaped not only by technological advancements but also by the critical balance between energy consumption, sustainability, and operational efficiency.
Organisations must adopt innovative approaches that prioritise energy efficiency and explore alternative power solutions, such as improved battery storage technologies or small nuclear reactors, to align their operations with long-term sustainability goals.