fundamentals
AI Capex as % of Cloud Revenue (Amazon, Alphabet, Microsoft, Oracle)
Latest
119.9%
Methodology
## What this measures
How much the four cloud providers spend on capital equipment for every dollar of cloud revenue: capex.cloud_providers.ttm divided by cloud_revenue.hyperscalers.ttm, as a percentage. Both cover the same companies (Amazon (AWS), Alphabet (Google Cloud), Microsoft (Intelligent Cloud) and Oracle (cloud)) and the same four quarters. Quarterly from the four quarters to June 2022.
## Why it matters
It puts the AI infrastructure debate in one number: the build-out (capex) against the revenue it is meant to earn (cloud sales). Above 100%, the companies are spending more on equipment in a year than their cloud businesses take in.
## How to read it
Higher means spending is running further ahead of revenue. It was 53.4% in the four quarters to December 2023, the low of the series, and 119.9% in the four quarters to June 2026, the high. A rising ratio says capex is growing faster than cloud revenue; a falling one says revenue is catching up.
## What it does not say
The numerator is company-wide capex, which also pays for retail, search, devices and other businesses, while the denominator is only the cloud line, so the level overstates what is spent per dollar of cloud revenue. Data centers last for years, so one year's capex is not meant to be repaid by one year's revenue. Microsoft's August 2024 redefinition of Intelligent Cloud lowers the revenue sum from September 2024, which lifts the ratio by a few points at that step. Meta, a large AI spender with no cloud revenue, is excluded.
## How it's built
Palanor divides the two four-company sums for each calendar quarter in which all four companies have reported, and rounds to one decimal. Oracle's quarters count in the calendar quarter they end in.
## Source
Palanor calculation from the companies' 10-Q and 10-K filings (U.S. Securities and Exchange Commission, EDGAR; public information). Refreshed weekly.
Read this signal inside the lattice.
Palanor weighs every signal against the world your organization is watching. Methodology is public; integration is the platform.
See it in Palanor →