The AI Capital Expenditure Cycle Has Not Peaked — And That Changes The Investment Calculus
The AI Capital Expenditure Cycle Has Not Peaked — And That Changes The Investment Calculus
Publish Date: 2026-07-03 16:14:00
Source Domain: www.forbes.com
-
AI Infrastructure Investment: Investment in AI infrastructure by major hyperscalers exceeded $360 billion in 2025, indicating long-term commitments to data centers, computing infrastructure, and model development.
-
Physical Resource Constraints: Regardless of AI model architecture, power, memory, chips, and data centers are critical inputs, suggesting focus on these physical resources may be more valuable than picking software company winners.
-
Energy Demand as a Constraint: Energy requirements, particularly electricity, are the most significant near-term constraints on AI development, impacting data centers’ expansion and driving growth in the power industry and related infrastructure.
-
Infrastructure as Investment Opportunity: For investors, considering the infrastructure layer like power, cooling, connectivity, and data center Real Estate Investment Trusts (REITs) might offer AI-related exposure without paying top valuations usually associated with high-cost software and semiconductor stocks.
