The U.S. is experiencing a significant business investment boom, but it's almost entirely driven by a handful of Big Tech companies.
This isn't a broad-based economic expansion but rather an extremely focused capital expenditure (capex) cycle led by what Apollo's Torsten Sløk calls a 'handful of mega-cap spenders'. Recent announcements have put a hard number on this concentration. In late April 2026, tech giants like Microsoft, Alphabet, Meta, and Amazon collectively signaled plans to invest over $700 billion in AI infrastructure for the year. This figure, once a distant forecast, is now a concrete budget item, confirmed by company earnings calls and media reports.
Macroeconomic data appears to support this narrative of strength. The Q1 2026 GDP report from the Bureau of Economic Analysis (BEA) showed business investment re-accelerating. However, the details are telling: spending on equipment (like servers and GPUs) and intellectual property (AI models and software) surged, while investment in nonresidential structures fell. This composition perfectly matches a tech-driven boom, not a widespread industrial recovery.
So, why is this investment cycle so heavily concentrated? The reasons are multifaceted. First, from an economic standpoint, the scale of AI infrastructure is immense. Building and operating massive data centers requires billions in upfront costs and can lead to negative free cash flow for years. Only companies with fortress-like balance sheets and massive cash flows, like the Big 4, can sustain such an investment pace. Smaller players simply cannot compete at this level.
Second, supply chain dynamics play a crucial role. With soaring demand for components like high-bandwidth memory (HBM), costs are rising. Microsoft, for instance, attributed a significant portion of its capex increase to higher component prices. In this environment, mega-caps can leverage their purchasing power to secure large, long-term supply agreements, leaving smaller firms to face higher prices and potential shortages.
Third, policy and financing conditions reinforce this trend. Government initiatives like the CHIPS Act provide subsidies that primarily benefit large-scale manufacturers and their biggest customers. Furthermore, in a crowded bond market, large, well-known tech companies can issue debt more easily and at better rates than smaller enterprises, further widening the gap.
Finally, there are physical bottlenecks. The demand for electricity to power these data centers is straining local grids. Securing power, land, and the necessary permits is a complex and capital-intensive process. Again, only the largest players have the resources and leverage to negotiate multi-gigawatt power purchase agreements and build bespoke infrastructure, cementing their dominance in the AI land-grab.
- Capex: Capital expenditure, or funds used by a company to acquire, upgrade, and maintain physical assets like property, buildings, and equipment.
- Hyperscaler: A large-scale cloud computing provider that offers massive, scalable infrastructure, such as Amazon Web Services (AWS), Microsoft Azure, and Google Cloud.
- TTM (Trailing Twelve Months): A financial metric that represents the data from the past 12 consecutive months, used to show the most recent performance.