Analysis
Oracle's cloud infrastructure revenue jumped 121% year over year, SiliconANGLE reported after the company's latest earnings update.
- Cloud infrastructure revenue -- $7.4B this quarter, +121% year over year.
- Total contracted backlog (RPO) -- $664B, Traders Agency detailed.
- New AI contracts this quarter -- more than $30B of the backlog growth.
Oracle is the smallest of the major cloud providers by absolute revenue -- AWS, Microsoft Azure, and Google Cloud all dwarf it -- but it's growing infrastructure revenue faster than any of them right now, largely on the back of its multiyear OpenAI compute commitment and a broader wave of AI labs needing capacity outside the traditional hyperscalers. The backlog figure has become Oracle's signature metric precisely because it's forward-looking: it represents contracted future revenue, not cash already collected.
“- Cloud infrastructure revenue -- $7.4B this quarter, +121% year over year.”
The timing lands one day after Oracle separately signed a $7 billion, 10-year software contract with the Pentagon covering on-premises data centers for the military, intelligence community, and Coast Guard -- a reminder that Oracle's AI-era growth isn't confined to hyperscale cloud, but is showing up across every part of its government and enterprise book.
What the backlog number doesn't show: Oracle has been funding this buildout substantially with debt, and Pulse has previously covered the company's debt-fueled AI capex tied to a round of roughly 21,000 layoffs elsewhere in the business. Backlog is also not revenue -- RPO converts to cash over years, not quarters, and a chunk of it depends on customers (including OpenAI) continuing to draw down capacity at the pace contracted. If AI demand growth slows before that backlog burns off, Oracle's balance sheet -- not just its income statement -- is what absorbs the mismatch.
For founders and infrastructure buyers, the number worth tracking isn't the headline backlog -- it's whether Oracle's actual delivered capacity keeps pace with contracted demand. Every hyperscaler is currently backlog-rich and capacity-poor; Oracle just put a specific number on how large that gap has become.