Amazon's $220B AI Spending Pushes Free Cash Flow Negative Despite Surging AWS Demand

Amazon plans $220B in 2026 capital spending as AWS revenue jumps 37%, but free cash flow falls to negative $7.6B amid AI infrastructure costs.

Amazon's $220B AI Spending: AWS Grows, Cash Flow Turns Negative

Key takeaways

  • Amazon raises 2026 capital spending to $220 billion
  • AWS revenue surges 37% as backlog reaches $496 billion
  • Free cash flow turns negative amid infrastructure expansion

Amazon is investing at an unprecedented pace to expand its artificial intelligence infrastructure, with projected 2026 capital spending reaching $220 billion even as its latest annualized free cash flow measure has fallen below zero.

The spending surge comes despite exceptional demand for Amazon Web Services (AWS), which reported 37% revenue growth in the second quarter and a $496 billion contract backlog. Amazon is now confronting power shortages, equipment constraints and construction challenges as it races to deliver computing capacity.

An October 19 Business Insider investigation describes how the company's ambitious expansion has increased financial commitments long before newly constructed facilities can begin generating revenue.

Amazon's $220 Billion AI Investment Drives Cash Flow Negative

Amazon raised its full-year cash capital expenditure forecast from approximately $200 billion to $220 billion in July, partly because of rising memory-chip costs and accelerating demand for AI computing.

According to its second-quarter earnings release, trailing-12-month operating cash flow increased 33% to $161.4 billion.

However, free cash flow declined from positive $18.2 billion a year earlier to negative $7.6 billion, primarily reflecting a $66.1 billion year-over-year increase in net property and equipment purchases.

The $220 billion budget includes broader company infrastructure, not exclusively AI projects. It also represents planned spending rather than an amount already deployed.

Amazon's spending illustrates the same challenge affecting the wider AI data center boom: substantial investment must occur before facilities can begin delivering contracted computing services.

AWS Revenue Jumps 37% While Power and Hardware Costs Rise

AWS generated $42.2 billion in second-quarter revenue, its strongest growth rate in 18 quarters, while operating income reached $16.6 billion.

Its backlog expanded to approximately $496 billion, representing contracted future business rather than revenue already collected.

Amazon CEO Andy Jassy has said the company expects insufficient capacity to satisfy demand through 2027, despite the enormous construction program.

According to Business Insider, AWS infrastructure executives are managing risks involving electricity connections, labor availability and server equipment deliveries, including expensive Nvidia hardware.

These challenges resemble the power-grid delays affecting other AI infrastructure developers, although Amazon has significantly greater financial resources than smaller operators.

Increasing hardware costs also create pressure across the semiconductor supply chain. Recent selling in chip stocks showed how concerns about AI investment returns can affect Nvidia, AMD and Micron even when demand remains substantial.

Wall Street Questions When AI Spending Will Pay Off

Amazon's cash flow deterioration comes during heightened scrutiny of the technology sector's infrastructure commitments.

CoreWeave recently experienced a 22% weekly stock decline as investors questioned its borrowing requirements, while Meta faces its own $145 billion spending test.

The companies have different financial profiles, but each faces pressure to demonstrate that infrastructure investment will produce durable returns.

AWS Chief Financial Officer John Felton told Axios that the company's AI spending remains sustainable, pointing to strong customer demand and a backlog approaching half a trillion dollars.

Amazon argues that infrastructure costs arrive years before the resulting revenue, making negative free cash flow a temporary consequence of rapid expansion rather than evidence of weak customer demand.