
Global investment in data centres is forecast to reach US$31.6 trillion through 2050 as the rapid adoption of artificial intelligence drives a capital-spending cycle that PwC says will surpass the railway, electricity-grid and internet build-outs of earlier eras.
PwC’s inaugural Global Data Center Outlook, released on Wednesday (September 2, 2026), projects that annual investment will rise from about US$800 billion in 2026 to US$1.1 trillion in 2030 and US$1.8 trillion by 2050.
Under a scenario in which AI is adopted more quickly than PwC’s baseline assumption, cumulative spending could exceed US$50 trillion over the next 25 years. The report compared that figure with the approximately US$30 trillion size of the US economy.
Consumers, businesses and governments are integrating AI into a growing range of daily activities and operations. The resulting demand for computing capacity is pushing technology groups such as Microsoft and Amazon, together with other data-centre providers, to accelerate the construction of new facilities around the world.
Most of the projected spending will go towards the equipment operating inside data centres rather than land or construction, according to PwC. AI processors and related hardware, including chips supplied by market leader Nvidia, will account for a substantial share.
This distinguishes the AI build-out from earlier infrastructure cycles that required heavy spending at the outset, such as railway construction, electricity grids, legacy memory-chip plants and fibre-optic networks.
AI infrastructure will instead require repeated capital expenditure as processors reach the end of their useful lives and need to be replaced with newer generations of chips.
“The AI cycle will reset and require upgrades every four to six years, with no end in sight,” PwC’s researchers wrote.
The refresh cycle means spending will not end when the initial buildings, power connections and computing systems are completed. Operators will need to continue investing in equipment to maintain processing performance and support increasingly demanding AI applications.
The report identified energy availability as one of the main factors determining how much of the forecast investment can proceed.
Developers need access to large volumes of electricity that are affordable, stable and available without interruption. At the same time, governments and operators face pressure to secure cleaner, low-carbon sources of power for energy-intensive data centres.
PwC has separately identified grid connections, backup generation, transformers, permitting, water availability, emissions rules, zoning and community acceptance as constraints that can determine which proposed facilities are ultimately financed and built.
Community resistance is already delaying projects. Data Center Watch found that at least 75 US data-centre projects valued at approximately US$130 billion were blocked or delayed during the first quarter of 2026 because of local opposition.
Protests cited in the outlook reflected concerns about environmental effects, competition for local resources and broader social consequences, including fears that AI could displace workers.
These barriers could slow expansion even where demand and investment capital are available, particularly in areas where electricity networks, water resources or local approval processes are already under pressure.
PwC’s baseline forecast assumes that the global trading system remains open and that processors and other semiconductor products can continue moving across borders.
A major interruption to the semiconductor supply chain could reduce projected worldwide data-centre investment by almost 20%, the report estimated.
The risk is amplified by the central role of hardware in the investment cycle. Restrictions on chip exports, production disruption or other barriers to cross-border supply could affect both the construction of new facilities and the regular replacement of existing equipment.
The United States is expected to receive about US$15.1 trillion of the projected investment through 2050, equivalent to almost half the global baseline total.
Among the other regional estimates cited by PwC, Asia-Pacific is forecast to attract US$8.2 trillion, followed by Europe with US$5.6 trillion, the Middle East with US$1.1 trillion and Africa with US$255 billion.
China and India are expected to be major sources of new demand because of their large populations, rapidly expanding digital economies and considerable remaining scope to apply AI in business operations and everyday life.
PwC said the central question was no longer whether investment capital was available or whether demand had emerged. It was which regions, data-centre providers and organisations were sufficiently prepared to capture the investment, and which would miss the opportunity.
Sources: PwC, Data Center Watch