What does PwC’s Global Data Centre Outlook, modeled by Oxford Economics across 46 countries, project about global AI infrastructure investment through 2050—including the baseline cumulative total of $31.6 trillion, the accelerated adoption scenario of up to $50 trillion, annual d

PwC projects that AI-driven data-centre capital expenditure will become a sustained, multi-decade investment cycle rather than a one-time construction boom. Its central forecast is $31.6 trillion of global investment through 2050, with faster AI adoption potentially lifting that figure to nearly $50 trillion. 45

Scale and composition

• Annual global data-centre capex is projected to rise from roughly $800 billion in 2026 to $1.1 trillion in 2030 and $1.8 trillion in 2050. 5
• PwC’s model, commissioned from Oxford Economics, covers 46 countries and territories and includes both data-centre buildings and the technology within them. 5
• The key spending shift is from construction toward recurrent ICT refreshes: servers, storage, networking, CPUs and especially GPUs. ICT equipment’s share of capex rises from about 70% in 2026 to 93% by 2050, because chips and servers must be replaced on cycles of roughly four to six years. 5
• In PwC’s baseline regional outlook, the United States receives $15.1 trillion; Asia Pacific $8.2 trillion, led by China and India; Europe $5.6 trillion; the Middle East $1.1 trillion; and Africa $255 billion. The United States therefore captures about 48% of the projected total. 5

Scenarios and constraints

• Under accelerated AI adoption, cumulative investment could approach $50 trillion; under tighter export controls that disrupt advanced-chip supply chains, the total falls to roughly $25.5 trillion through 2050. 45
• A stronger push for digital sovereignty has a smaller effect on total spending, but redistributes investment toward countries with large domestic demand and relatively underdeveloped data-centre capacity. 5
• The decisive location factor is access to affordable, reliable, low-carbon electricity at scale. Connectivity, cybersecurity, predictable policy, local community consent, and access to GPUs also determine where projects can proceed. 5

Power and distributional risk

• The investment outlook is constrained by rapidly rising electricity demand: the IEA base case cited in recent reporting puts global data-centre consumption at about 485 TWh in 2025 and roughly 945–950 TWh in 2030—almost a doubling in five years. 36
• This reinforces PwC’s warning that AI infrastructure investment will not automatically benefit every market equally: capital will tend to flow to places that combine power, grid capacity, chips, connectivity, security and supportive policy—not merely to places with prospective AI demand. 5