The Soaring Cost of AI Infrastructure
The artificial intelligence boom is driving unprecedented capital expenditure among the largest US tech companies. According to a Jefferies report by Global Head of Equity Strategy Christopher Wood, the four major US hyperscalers—Microsoft, Amazon, Alphabet, and Meta—collectively poured an astounding $165 billion into AI infrastructure during the second quarter of 2026. This figure represents a significant increase from $72 billion recorded in the first quarter of 2025, highlighting the accelerating pace of investment in AI capabilities.
Free Cash Flow Dries Up
Despite the spectacular earnings growth attributed to AI, a less comfortable financial indicator is emerging: free cash flow. The aggregate free cash flow for these four tech giants plummeted to just $7 billion in Q2 2026, a stark drop from $60 billion in Q4 2025. Jefferies further projects this figure to turn negative, reaching an estimated -$12 billion in Q3 2026.
This creates a striking financial paradox: while AI is increasingly vital to corporate earnings, the enormous capital required to build the necessary infrastructure is rapidly consuming available cash.
Who Benefits Now?
The immediate beneficiaries of this AI capital expenditure race are often the suppliers. Chipmakers, semiconductor equipment companies, and server manufacturers, often referred to as "picks and shovels" players, can recognize revenue and profits as they sell infrastructure components. In contrast, the hyperscalers themselves must absorb these substantial capital costs and depreciate the assets over time, deferring their returns.
The Rising Depreciation Burden
The cash outflow is only one aspect of the financial strain. As these tech behemoths construct more data centers and procure advanced computing equipment, their depreciation expenses are also climbing. The combined depreciation expenses for Microsoft, Amazon, Alphabet, and Meta reached $44.5 billion in Q2 2026, marking a 24% increase year-on-year.
Investor Concerns and Future Outlook
For investors, the critical question is shifting from whether Big Tech can afford to build this AI infrastructure to whether the eventual revenue generated will justify the colossal capital investment. An important source of optimism stems from the robust growth in cloud revenues. Microsoft, Google, and Amazon saw their combined cloud revenues rise 38% year-on-year to $126 billion in Q2 2026. This expanding cloud business is central to the argument that current AI spending could ultimately yield substantial returns.
However, the sheer scale of investment means the payoff will need to be equally substantial. For now, the AI boom continues to be earnings-accretive, with Jefferies estimating that the S&P 500's AI basket could deliver annualized EPS growth of 48% during 2026-27, significantly outpacing the broader S&P 500's 23% and the index excluding AI's 12%.