The latest data and forecasts from Synergy Research Group indicate that the average capacity of hyper-efficient data centers that will be opening in the next six years will be more than double that of current hyper-efficient data centers. While the trend has always been for the critical IT load of these centers to increase over time, the impact of generative artificial intelligence technology and services has added an additional push to the need for significantly more powerful facilities. Meanwhile, as the average IT load of individual data centers increases, so too will the number of operational hyper-efficient data centers. This will include some degree of remodeling of existing data centers to increase their capacity. As a result, the total capacity of all operational hyper-efficient data centers will nearly triple in the next six years.
Synergy’s analysis is based on the study of the space and operations of data centers of 19 of the world’s leading cloud and internet service companies that meet Synergy’s criteria for being considered hyper-efficient operators. This includes the largest operators in areas such as SaaS, IaaS, PaaS, search, social media, e-commerce, and gaming. By mid-2023, these companies had 926 large data centers in operation worldwide. Synergy’s well-known future data center plan includes 427 new installations, which is a key input to Synergy’s forecasting model.
The mix of hyper-efficient data centers continues to change, region by region, and between owned versus leased, but overall the total number of data centers worldwide has doubled in the past five years. The impact of recent advances in generative AI is not so much to increase the number of data centers—which will continue to grow by over a hundred per year—but to substantially increase the amount of energy needed to operate those data centers. As the number of GPUs in hyper-efficient data centers skyrockets, primarily driven by AI, the energy density of associated racks and data center facilities must also increase substantially. This is leading hyper-efficient operators to rethink some of their data center architecture and deployment plans.