According to a recent report by the Dell'Oro Group, global data centre capital expenditure (capex) is set to increase at a compound annual growth rate of 21%. A significant contribution to this surge is expected from hyperscale cloud service providers, who will account for half of the projected $1.2 trillion global data centre capex by 2029.
Baron Fung, Senior Research Director at Dell'Oro Group, highlights, "GPUs and custom AI accelerators now account for roughly one third of total data centre capex, making them the single largest driver of growth." These remarks emphasise the substantial role these technologies play in current and future data centre expansions. This trend is further supported by robust spending across the infrastructure, which includes racks, compute solutions, storage, networking, and physical facilities.
The hyperscalers, known for leveraging vertically integrated solutions and custom architectures, are at the forefront of optimising data centre performance while reducing computation costs. Both public and private sector investments are contributing to the expansion wave. To cater to this global demand, hyperscalers and collocation providers are anticipated to introduce over 50 gigawatts of new capacity within five years. Although the forecast predicts a transient deceleration in 2026, investments are poised to sustain growth in the long run.
Additional insights from the report include:
- Accelerated servers for AI tasks are anticipated to represent a substantial portion of infrastructure spending by 2029.
- The top US-based cloud service providers (Amazon, Google, Meta, and Microsoft) will dominate nearly half of the global data centre capex in 2025.
- The Rest-of-Cloud category, including neo-cloud and GPU-as-a-Service entities, is experiencing an impressive CAGR of 39%.
The Dell'Oro Group's Data Center IT Capex 5-Year Forecast Report provides an extensive review of market trends from 2014 onward. It offers insights into how the landscape for data centre and server capex is evolving, as well as projections for various customer segments and revenue breakdowns for different infrastructure sub-segments.
