Hyperscale Data Centers Set to Double Capacity, Fueled by Generative AI Advancements

The average capacity of hyperscale data centers is predicted to more than double over the coming six years compared to the existing operational hyperscale facilities, Synergy Research Group said.
AI to drive Hyperscale Data Center Capacity
Historically, hyperscale data centers have expanded their critical IT loads, but the integration of generative AI technology and services has accelerated the demand for significantly more potent facilities.

This surge in capacity is due to a two-fold effect: not only will the average IT load of individual data centers rise, but the overall number of operational hyperscale data centers will also steadily grow. Additionally, retrofitting existing data centers to boost their capacity is part of the evolving strategy. Consequently, the total capacity of all operational hyperscale data centers is anticipated to increase nearly threefold within the next six years.

The research is grounded in an extensive analysis of the data center footprint and operations of 19 leading cloud and internet service firms globally, meeting Synergy’s criteria for being classified as hyperscale operators. This group encompasses major players in Software as a Service (SaaS), Infrastructure as a Service (IaaS), Platform as a Service (PaaS), search, social networking, e-commerce, and gaming. As of mid-2023, these companies collectively operated 926 major data centers worldwide. Synergy’s known pipeline of future data centers includes an additional 427 facilities, forming a fundamental basis for their forecast model.

The composition of hyperscale data centers is dynamically changing, with variances observed in different regions and ownership structures. The total count of global data centers has doubled in the last five years. However, the recent surge in generative AI advancements is chiefly amplifying the power requirements of these data centers rather than drastically increasing their numbers.

With the exponential growth in GPUs within hyperscale data centers driven primarily by AI, the power density of associated racks and data center facilities must also witness a significant boost. This reality is prompting hyperscale operators to reassess their data center architecture and deployment strategies, the report said.