Industry analyst Gartner is predicting that 40% of existing AI data centres will be operationally constrained by power availability by 2027 as AI and generative AI (GenAI) drive rapid increases in electricity consumption, with potential impacts on prices and sustainability goals.
► Analyst predicts that demand for energy to run GenAI will exceed capacity
► Prices and sustainability goals may be impacted
Data centres are forecast to grow by as much as 160% over the next two years, and the firm says that energy suppliers will struggle to keep up. It expects this to limit the growth of new data centre for GenAI and other uses from 2026.
Gartner estimates the power required for data centre to run incremental AI-optimised servers will reach 500 terawatt-hours (TWh) per year in 2027 – 2.6 times the level required in 2023. As a result, it says, power prices for AI data centres are likely to rise with the increased costs passed on to service providers and end customers.
The analyst also notes that big power users are already working with major energy producers to secure long-term guaranteed sources of power independent of other grid demands. It also recommends that a rise in power costs is factored into future plans.
Zero-carbon goals are also a concern and may be negatively affected by the short-term need for more power. Gartner suggested that, when developing GenAI applications, organisations should focus on using a minimum amount of computing power and look at the viability of other options such as edge computing and smaller language models.
