PwC projects global AI infrastructure investment to reach about $31.6 trillion by 2050, with data center capex rising from roughly $800 billion in 2026 to around $1.8 trillion by 2050.

The United States is expected to account for about $15.1 trillion of this spend (roughly 48%), with Asia Pacific following at $8.2 trillion, led by China and India; Europe and the Middle East would make up the remainder.

Oxford Economics modeled the outlook across 46 countries and five regions, highlighting regional variations in drivers and risks.

Key shaping factors include power supply availability, chip constraints, and sovereign regulatory trends, which could reallocate regional winners and losers.

A practical constraint is infrastructure readiness—transformer lead times, grid connections, cooling, and planning approvals—that can slow deployment even when funding is available.

Recent restrictions on New York data center construction illustrate potential regulatory headwinds, while regional opportunities and risks align with the investment growth outlook.

The core question is which regions, operators, and institutions are best positioned to capture the AI infrastructure opportunity, not whether the capital or demand exists.

PwC emphasizes identifying regions and providers most capable of seizing opportunities and avoiding missed ones, given evolving requirements and risks.

Environmental and financial risks accompany buildout, including higher cooling water demand, transmission needs, and potential asset underutilization if AI demand diverges from projections.

Industry voices stress that AI infrastructure requires active capital allocation and should be viewed as hybrid assets with complex risk, potentially facing policy-driven constraints in regions like North America.

PwC’s Clara Cutajar notes AI infrastructure spans tech, energy, real estate, supply chains, regulation, and financing, demanding proactive investor positioning.

Three constraints shape investment geography: power supply, data sovereignty, and semiconductor chains, with power availability as the single most significant determinant.