A surge in generative AI could leave 40% of data centres short of power by 2027, warns Gartner
The race to deploy artificial intelligence, particularly generative AI (GenAI), is driving electricity consumption in data centres to unprecedented levels. Gartner, a research firm, in its report predicts that over the next two years, energy usage by data centres could soar by as much as 160%. By 2027, this surge may leave 40% of existing AI data centres operationally constrained due to insufficient power availability.
“The explosive growth of new hyperscale data centres to implement GenAI is creating an insatiable demand for power that will exceed the ability of utility providers to expand their capacity fast enough,” says Bob Johnson, a vice-president at Gartner. This mismatch threatens not only energy availability but also the expansion of new data centres for GenAI and other uses from 2026 onwards.
Gartner estimates that the power required for data centres to run additional AI-optimised servers will reach 500 terawatt-hours (TWh) per year in 2027—2.6 times the level in 2023. The proliferation of large language models (LLMs), which underpin many GenAI applications, necessitates vast amounts of data and computational power, driving this surge in energy demand.
However, expanding power transmission, distribution, and generation capacity is a slow process. “Short-term power shortages are likely to continue for years,” notes Mr Johnson. This could result in the number of new data centres and the growth of GenAI being governed more by the availability of power than by technological innovation or market demand.
An Inevitable Rise in Electricity Costs
Impending power shortages are set to push electricity prices upward, inflating the costs of operating LLMs. Significant power users are already collaborating with major producers to secure long-term, guaranteed sources of power, independent of the general grid. In the meantime, data centre operators may leverage economic means to secure the power they need, passing increased costs onto AI and GenAI product and service providers.
Gartner advises organisations to anticipate higher power costs in their strategic planning. Negotiating long-term contracts for data centre services at reasonable rates could mitigate some financial impact. Companies should also factor in significant cost increases when developing new products and services, exploring alternative approaches that require less power.
Sustainability Goals Under Strain
The scramble to meet soaring energy demands may undermine zero-carbon sustainability objectives. In some cases, suppliers are keeping fossil fuel plants operational beyond their scheduled retirement to boost production. “The reality is that increased data centre use will lead to increased CO₂ emissions to generate the needed power in the short term,” acknowledges Mr Johnson. This development makes it more challenging for data centre operators and their clients to meet aggressive CO₂ emissions targets.
Data centres require uninterrupted power—a demand that renewable sources like wind or solar cannot consistently fulfil without alternative solutions during periods of low generation. Currently, reliable 24/7 power is predominantly provided by hydroelectric, fossil fuel, or nuclear plants. In the long term, advancements in battery storage technology or the adoption of small nuclear reactors may help align energy needs with sustainability goals.
Gartner recommends that organisations re-evaluate their sustainability objectives in light of future data centre requirements and power sources over the coming years. When developing GenAI applications, firms should focus on minimising computing power usage and consider alternatives such as edge computing and smaller language models.