Amazon, Alphabet, Meta, and Microsoft spent $170 billion on data centers and chips in a single quarter, up 72% from a year earlier. Wall Street now estimates the four companies' combined capital spending at roughly $1.5 trillion through the end of 2027, and the buildout is tightening the same chip and power markets that Bitcoin miners depend on.
Amazon, Alphabet, Meta, and Microsoft together spent $170 billion on data centers in a single quarter, a 72% jump from the same period last year. The four companies are now on track to spend roughly $1.5 trillion on data center and chip infrastructure through the end of 2027, according to Wall Street estimates.
Amazon and Meta drive the increase
Amazon led the spending in Q2 2026 with $53 billion in capex, a 69% jump year-over-year. The company also raised its full-year guidance to approximately $220 billion, citing soaring memory chip costs as a key driver.
Meta reported a 55% year-on-year increase in its infrastructure costs, pushing it to revise its 2026 capex target upward to $130 billion. Component price inflation alone added tens of billions of dollars to hyperscaler budgets during the quarter. In Q1 2025, industry capex was already up 53% year-over-year to $134 billion. All four companies have also pledged investments in new power generation capacity to keep their servers running.
The squeeze reaches crypto miners
Nvidia and AMD are allocating large portions of their output to hyperscaler contracts, and that same supply chain feeds crypto mining hardware and blockchain infrastructure. Publicly traded Bitcoin miners, including Marathon Digital and Riot Platforms, have already been navigating a tighter hardware market.
Core Scientific and Hut 8 have struck deals to repurpose mining facilities for AI workloads. As Big Tech competes for power generation capacity and signs long-term energy contracts, it reshapes the energy landscape for every operator sharing the grid.
A bull case and a bear case
When four companies spend at a $680 billion annual run rate on infrastructure, that capital comes either from operating cash flow or from debt markets, where it competes with every other borrower for capital. The bull case holds that these companies are building the infrastructure layer for the next decade of AI applications.
The bear case: spending $1.5 trillion over a few years only works if revenue materializes at sufficient margins. If it doesn't, the result is some of the most expensive stranded assets in corporate history. The relationship between AI data centers and Bitcoin mining facilities is increasingly symbiotic, with shared power infrastructure and even shared physical locations.
Source: Crypto Briefing
Trading involves risk.
Risk Disclaimer
