Combined capital expenditures from Amazon, Alphabet, Meta, Microsoft and Oracle are estimated to rise by 1.0 percentage point of US GDP in 2026, pushing total AI infrastructure spending among the five companies to a record 2.4% of GDP. That compares with a 1.0% share as recently as 2023, according to figures compiled by The Kobeissi Letter — a pace of increase the firm called rarely seen in modern economic history.
Individually, Amazon is projecting roughly $200 billion in 2026 capital expenditures, Alphabet has guided to $175-185 billion, Microsoft is tracking toward $190 billion, and Meta has guided to $115-135 billion, according to company disclosures compiled across recent earnings cycles. Combined, the four companies' AI infrastructure spending this year rivals the entire GDP of a mid-sized European economy — a scale of investment concentrated in a handful of companies that has few historical parallels.
Where the money is actually going
The spending is flowing overwhelmingly into data center capacity and the power infrastructure needed to run it — which is exactly the demand that has been pulling bitcoin miners into the AI business over the past several months. Riot Platforms' newly signed $9.1 billion, 20-year lease with Anthropic is a direct beneficiary of this spending wave, and it followed a similar $19 billion deal between Anthropic and fellow miner TeraWulf roughly a month earlier. Both deals illustrate how AI labs are increasingly paying a premium for exactly the kind of power-dense infrastructure that crypto miners spent years building for an entirely different purpose.
The scale of spending has also raised questions among analysts about its effect on the companies' own cash positions, with some projections warning that combined free cash flow across the group could fall sharply in 2026 as capital expenditure continues to outpace the revenue AI products are currently generating.
A double-edged signal for markets
For now, markets have broadly rewarded the spending story rather than punished it, treating aggressive AI capex as a signal of confidence in future demand rather than a warning sign about capital discipline. Whether that reading holds through a full earnings cycle of shrinking free cash flow remains one of the more closely watched macro questions heading into the back half of 2026.
Related: Riot's $9.1B Anthropic Lease Could Ease Its Bitcoin Selling Pressure
Either way, the AI infrastructure buildout is proving to be one of the largest capital reallocations in recent corporate history — and one with second-order effects, like the miner-to-AI-landlord pivot, that are still working their way through adjacent markets.
