Microsoft maps major data center buildout

Microsoft (MSFT) is planning a substantial increase in its global data center power capacity, outlining a roadmap that extends to 2032. The company intends to grow total data center capacity from about 12 gigawatts today to more than 38 gigawatts by the end of that period. This represents a multi-year, phased buildout of computing infrastructure to meet anticipated demand.

The expansion focuses on the aggregate power capacity available across Microsoft’s data center fleet. It is framed as a broad program rather than a single project, with growth distributed across multiple facilities and regions over time. The figures reflect internal planning targets for the size of the company’s overall data center footprint by 2032.

A key element of the plan is a deliberate shift toward infrastructure optimized for artificial intelligence workloads. Roughly one-third of the enlarged data center capacity is expected to be dedicated to AI-specific silicon. This portion of the buildout is aimed at supporting compute-intensive AI applications that require specialized chips rather than general-purpose processors.

By allocating a significant share of future capacity to AI-focused hardware, Microsoft’s roadmap highlights the central role of AI in its long-term infrastructure mix. The emphasis on AI-specific silicon is embedded within the broader power capacity target, signaling an integrated approach to cloud and AI infrastructure rather than separate buildouts.

The planned capacity increase is described as covering both company-owned and leased data center facilities. This means the more than 38 gigawatts of targeted capacity would come from assets directly under Microsoft’s operational control, whether on campuses it owns or sites it leases. The approach combines long-term ownership with flexible leasing to scale capacity.

Notably, the figures do not include computing capacity rented from specialist neocloud providers. By excluding these third-party arrangements, the roadmap focuses on infrastructure that is part of Microsoft’s own data center fleet. This distinction clarifies that the growth targets relate to the company’s owned and leased footprint rather than its total accessible compute across all possible partners.

The planned tripling of data center power capacity underscores a strategic preparation for rising cloud and AI usage. Expanding from about 12 gigawatts to more than 38 gigawatts positions Microsoft to handle larger and more complex workloads over the coming years. The integration of AI-specific silicon into roughly one-third of that capacity further aligns the infrastructure roadmap with expected advances in AI services.

By charting this long-term buildout through 2032, Microsoft is setting quantitative targets for the scale and composition of its data center fleet. The plan’s emphasis on owned and leased assets, combined with a defined AI-focused share, provides a structured view of how the company aims to grow its core computing backbone while distinguishing it from capacity sourced through external neocloud providers.