Electronics Weekly
Datacentre capex grew 51% to $455 billion in 2024, says Dell’Oro, with accelerated servers optimised for AI training workloads deployed by the hyperscalers accounting for most of the growth.
“The top 10 hyperscalers accounted for more than half of global datacentre capex in 2024, driven largely by heightened investments in AI infrastructure,” says Dell’Oro vp Baron Fung, “while NVIDIA’s Hopper architecture—followed by Blackwell 
systems later in the year—dominated spending, custom accelerators from Google, Amazon, and Microsoft also fuelled growth. Additionally, Tier 2 cloud providers like xAI and CoreWeave significantly ramped up their capex, approaching hyperscaler levels due to increased GPU deployments. Hyperscalers and colocation providers also expanded investments in infrastructure, including dedicated AI networks and high-power facilities to support these compute-intensive workloads,” explained Fung.
Global datacentre capex is projected to rise by more than 30% in 2025, with sustained demand on AI infrastructure and a broader recovery in general-purpose infrastructure for servers and networking.
Dell led all OEMs in server revenue for 2024, with HPE and Supermicro trailing behind. Accelerated servers accounted for an estimated 36% of OEM server revenue as AI adoption expanded in the non-hyperscale market. Meanwhile, white-box server vendors captured over 56% of total server revenue, driven by strong hyperscale demand for AI servers.
Ongoing economic uncertainties may curb enterprise IT spending, while supply constraints and the rise of cloud and consumption-based services could result in near-term fluctuations in AI investments within enterprise data centers
