Your raise used to go offshore. Then it went to a buyback. Now it’s going to a data center - Fortune
Summary
A Goldman Sachs global strategy paper argues the world has entered a new investment regime defined by rising real rates, geopolitical fragmentation and a synchronized capital-spending surge, replacing the buyback-driven model of the prior cycle. Capital is now flowing into data centers, semiconductor fabs, power grids and fiber networks for AI infrastructure.
Market Impact
The thesis reframes capital allocation across sectors, with the top five hyperscalers projected to spend about $755 billion on capex in 2026 and S&P 500 capex growth far outpacing buybacks. It implies a broadening of market leadership toward industrials, energy and real assets. This analysis is informational and avoids any directional trading claims.
Why It Matters
It articulates a structural shift in where corporate capital flows, away from buybacks toward physical AI and energy infrastructure, with broad cross-asset implications.
Key Points
- Goldman's Peter Oppenheimer argues the world has entered a structurally different era of rising real rates, fragmentation and surging capital spending.
- The top five hyperscalers, Amazon, Meta, Google, Microsoft and Oracle, are expected to spend roughly $755 billion on capex in 2026, up 84% from 2024.
- S&P 500 companies reported year-over-year capex growth of +38% in Q1 2026, versus just +1% for buybacks.
- Goldman says the capital is flowing to data centers, semiconductor fabs, power grids and fiber networks rather than primarily to workers.
Key Entities
Evidence
The top five hyperscalers— Amazon , Meta , Google , Microsoft , and Oracle —are expected to spend approximately $755 billion on capital expenditures in 2026, up 84% from 2024.Supports: Supports the hyperscaler capex figure.
S&P 500 companies reported year-over-year capex growth of +38% in Q1 2026, compared to just +1% for buybacks—an almost exact inversion of the dynamic that defined the previous decade and a half.Supports: Supports the capex-vs-buybacks inversion.
It is going to data centers, semiconductor fabs, power grids and fibre networks, as the physical infrastructure of artificial intelligence requires extraordinary amounts of capital and relatively few people to build a...Supports: Supports where capital is flowing in the summary.