Global investment in artificial-intelligence infrastructure could reach $31.6 trillion through 2050, according to PwC, as data centers, chips, networking and power systems absorb record amounts of capital.
PwC projects annual data-center capital expenditure rising from about $800 billion in 2026 to $1.8 trillion by 2050. Reuters reported that the scale of the current AI buildout now exceeds the inflation-adjusted investment waves associated with earlier technology booms.
The spending case depends on future revenue
The central question is whether new AI products and productivity gains will generate enough revenue to justify the outlays. Bain estimates cited by Reuters suggest large U.S. hyperscalers and other AI companies may need more than $4.2 trillion of additional revenue over the next five years to fund current expansion plans.
Economists remain divided on how quickly that payoff will arrive. Broad productivity gains have so far been limited, and history suggests general-purpose technologies can take many years before their benefits show up fully in economic data.
Power and chip replacement are major cost drivers
PwC expects recurring chip upgrades to account for a large share of long-term spending rather than construction alone. Power availability is also becoming a key factor in where new data centers can be built.
The physical infrastructure may retain value even if AI revenue growth arrives more slowly than investors expect, but the gap between present spending and proven returns remains one of the biggest financial questions in the sector.
Sources: PwC Global Data Centre Outlook; Reuters, Oct. 3, 2026.
