Amazon, Google, Microsoft, Meta, and Oracle expected to bleed cash due to AI investments

The five companies that essentially run the internet’s backend are spending so much on AI that some of them are now losing money on a free cash flow basis. Alphabet just reported its first negative free cash flow quarter since going public in 2004. Amazon isn’t far behind.

Collectively, Amazon, Alphabet, Microsoft, Meta, and Oracle have guided for more than $725 billion in capital expenditure for 2026. That’s a 77% increase over 2025, and the vast majority is going toward data centers, custom chips, and other AI infrastructure.

The cash flow crunch is real

Alphabet posted negative free cash flow of roughly $5.9 billion in Q2 2026, driven by $44.9 billion in quarterly capex alone. To put that in perspective, Alphabet hadn’t been FCF-negative in a single quarter since it IPO’d over two decades ago.

Amazon’s trailing twelve-month free cash flow sits at negative $7.6 billion. Meta saw its cash generation collapse by 91% year-over-year. Oracle’s situation might be the most eyebrow-raising of the bunch: its fiscal 2026 capex hit 174% of operating cash flow, meaning it spent nearly twice as much on infrastructure as it generated from operations.

Oracle’s solution? Plans to raise between $45 billion and $50 billion through debt and equity offerings.

The scale is hard to comprehend

Analyst projections suggest AI-related capital spending could approach or exceed $765 billion in 2026 and potentially reach $1.2 trillion by 2027. Looking further out, estimates for aggregate hyperscaler spending on compute, data centers, and power from 2025 to 2030 land at over $5.3 trillion, with 2026-2031 projections stretching to $7.6 trillion.

Throughout the late July 2026 earnings season, most of these firms reported strong profit growth and healthy cloud revenue expansion. But Wall Street fixated on the cash flow deterioration. Rising memory chip costs added another layer of expense pressure on top of the already staggering construction and equipment bills.

Why this is happening now

The AI capex arms race traces back to late 2023, when the generative AI boom forced every major cloud provider to dramatically accelerate infrastructure buildouts. Customer demand for AI compute, particularly for training and running large language models, exceeded available capacity almost overnight.

In some cases, annual capex now rivals or exceeds total operating cash flow, a ratio that historically would have triggered alarm bells across any industry. Some companies have turned to debt markets to bridge the gap, a notable shift for firms that traditionally funded expansion entirely from operations.

What this means for investors

The Alphabet situation is particularly instructive. A company that generated positive free cash flow every single quarter for 22 years just posted a nearly $6 billion deficit.

The key metric to watch going forward is the ratio of capex to operating cash flow. When that number consistently exceeds 100%, as it already has for Oracle and is approaching for others, it means the business model depends on external financing to sustain its investment thesis.