Revenue Growth

• Google Cloud grew 82% to $24.8B-by far the fastest of the three and approaching a $100B annualized run rate.

• Amazon Web Services (AWS) grew 36.7% to $42.2B, its fastest growth in 18 quarters and a $169B run rate.

• Microsoft Azure grew 43%. Full-year Azure revenue surpassed $100B for the first time.

​All three said demand continues to exceed available capacity.

Cloud Commits

• Google Cloud backlog reached $514B, up $52B or 11.3% QoQ.

• AWS backlog reached $496B, up $132B or 36.3%. AWS drove almost two-thirds of the combined increase and narrowed its gap with Google.

• Microsoft's total commercial RPO reached $678B, up $51B or 8.1%. It does not disclose Azure-only commitments. We estimate ~$460B including OpenAI-up ~$25B (see my last MSFT post for methodology).

Add it up and modeled cloud commitments across the big three now exceed $1.47T, vs $1.26T last quarter. That is 16.6% growth in three months

What’s inside them?

Microsoft said all $51B of sequential growth came from customers outside frontier model companies.

AWS described demand across core cloud and AI.

Google said the majority comes from typical GCP contracts across a broad customer mix. Neither provides a split of Frontier labs (though last quarter AWS mentioned a $100B Anthropic contract that wasn't included then).​

AI Revenue and Scale

• AWS said its AI and chips businesses each crossed $25B annualized run rates, growing triple digits.

• Microsoft Foundry reached 100,000 customers and 2X+ revenue. Customers operating at a 1T token annualized rate increased 4X.

• Google Cloud reported Marketplace transactions grew more than 7X YoY. ~500 customers processing more than 1T tokens over the last year.

CapEx

• AWS raised its 2026 plan to $220B.

• Google raised guidance to $195–205B.

• Microsoft moved its calendar 2026 capex expectation to $175B from $190B. The reduction reflects lease accounting; underlying investment expectations remain unchanged.

​Combined plans now total ~$590–600B. Each still expects demand to outstrip available AI capacity.

What this means for alliance leaders

• Not all $1.47T is Marketplace-addressable. But the committed-spend pool that eligible software can help consume is rapidly expanding.

• Growth is broadening beyond frontier-lab megadeals and into enterprise production workloads.

• AI pulls databases, storage, CPUs, security and observability around it. ISVs attached to AI or adjacent workloads give cloud sellers a clearer consumption and co-sell story

The opportunity is to enter account planning before customers allocate their next wave of cloud commitments.