Meta and Amazon plan $725B capex for 2026 as Big Tech spending surges
In brief
- Meta and Amazon plan $725B combined capex for 2026 amid AI infrastructure competition
- U.S. hyperscaler capex projected to reach $805B in 2026, nearly doubling from $410B in 2025
- Capital investment focuses on AI data centers, semiconductor chips, and networking infrastructure
- Analysts debate sustainability of accelerating capex spending trajectory
Big Tech's capex surge
Meta and Amazon planned combined capital expenditures of $725 billion for 2026, according to recent projections. This follows a period of significant investment growth across the industry. Major hyperscalers including Meta, Amazon, Microsoft, and Alphabet spent $410 billion in 2025, marking a substantial increase in infrastructure deployment.
Morgan Stanley has projected that total U.S. hyperscaler capital expenditures could reach $805 billion in 2026, driven by competitive pressures and the expanding demands of AI systems. The surge is primarily directed towards AI data centers, chips, and networking equipment, reflecting the growing demand for computational resources.
Where the skeptics stand
Not all observers are convinced the scale of capex is sustainable or justified. Some analysts question whether improvements in AI model efficiency will materialize fast enough to justify the infrastructure investment, or whether oversupply of data center capacity could compress margins across the industry. The risk of stranded assets—data centers that prove underutilized—remains a concern for investors tracking long-term returns on these massive deployments.
The competitive dynamics are also shifting. Companies that can deploy infrastructure efficiently may gain advantages in model quality and inference costs, but the industry still faces uncertainty about whether the current capex trajectory aligns with actual demand for AI services over the next 3-5 years.
