The Electric Power Research Institute (EPRI) recently released a whitepaper titled “Powering Intelligence: Analyzing Artificial Intelligence and Data Center Energy Consumption,” which sheds light on the exponential growth potential of AI power requirements. According to the report, U.S. data centers alone could see a more than double increase in power consumption by 2030, reaching 166% of current levels. This surge in demand is largely driven by generative AI, which requires significantly more power per query compared to traditional search methods. The report estimates that AI queries, excluding images and rich content, consume 10 times the electricity of traditional Google queries.
#The Impact of AI on Energy Consumption
The EPRI report examines various use cases, including Google search, ChatGPT, BLOOM, and AI-powered Google search. While ChatGPT was found to be the least energy-intensive among AI-based queries, the integration of Google’s AI capabilities into Google Search could make it over three times more energy-intensive than ChatGPT. These findings highlight the significant energy requirements associated with AI applications and the potential impact on overall data center power consumption.
#The Future of Data Center Energy Usage
EPRI developed four different forecasts for electricity usage in U.S. data centers between 2023 and 2030. Even under the lowest growth scenario, data center electricity usage is projected to increase by 29% by 2030. The highest growth scenario predicts a staggering 166% increase in data center electricity usage by 2030. The uneven distribution of this growth poses localized challenges, with certain states like Virginia projected to have data centers comprise a significant portion of total electricity demand.
#Meeting the Demands of AI-Powered Applications
As the demand for AI-powered applications continues to rise, enterprises are not only focusing on acquiring the necessary hardware, data, and models but also grappling with the power requirements of these energy-intensive systems. Companies like Amazon, Google, and Microsoft, which have embraced AI and operate hyperscale data centers, have realized the importance of securing long-term data center capacity to support their growth plans. This involves negotiating multi-year contracts with power providers, facilities operators, and contract manufacturers to ensure a steady supply of resources. For other enterprises, this may require a fundamental shift in how they approach data center procurement, moving away from a “three bids and a buy” model to forging longer-term partnerships with data center and equipment providers.
#The Need for Strategic Thinking and Long-Term Planning
To navigate the changing landscape of data center procurement, enterprise IT leaders must adopt a proactive and forward-looking approach. This shift towards capacity contracting involves close collaboration between IT, facilities, and finance teams, as well as upfront investments in infrastructure that may not yield immediate returns. However, it is crucial for enterprises serious about competing in an AI-driven future to secure the resources needed to stay ahead of the curve.
