Home/Deal Room/Snowflake commits $6B to AWS AI infrastructure

Cloud capacity · Evidence brief

Snowflake commits $6B to AWS AI infrastructure

Snowflake + Amazon Web Services

Primary sourcePress release only2026-05-27
Deal structureStrategic cloud infrastructure commitment / AI compute capacity
GeographyAWS global cloud footprint / Snowflake AI Data Cloud regions
Infrastructure layersCloud capacity + chips + AI data platforms
Disclosed scale$6B multi-year infrastructure commitment to AWS over five years, covering Graviton compute and AI spend for Snowflake data and AI workloads

Why it matters

The strategic read

Adds a large enterprise AI-demand commitment to the AWS custom-silicon lane. Snowflake is not just reselling software here; it is reserving a multi-billion-dollar cloud infrastructure path tied to Graviton processors, GPU-accelerated EC2, model training, inference, and agentic AI workloads that run close to governed enterprise data.

Source-supported terms

What was disclosed

  1. 01

    Snowflake signed a multi-year strategic collaboration agreement with AWS to accelerate enterprise agentic AI adoption.

  2. 02

    Snowflake committed $6B in Graviton compute and AI spend on AWS over five years.

  3. 03

    The companies described it as Snowflake’s largest AWS infrastructure commitment to date.

  4. 04

    Snowflake said the expanded collaboration includes deeper generative AI and agentic AI integrations, AWS Marketplace go-to-market, workload migrations and customer success programs.

  5. 05

    Snowflake uses AWS Graviton processors and high-performance GPU-accelerated Amazon EC2 instances for AI model training and inference.

  6. 06

    Snowflake said launches are completed or underway in 10 new AWS regions, including Auckland, Cape Town, Bangkok and the AWS European Sovereign Cloud, to support data residency and bring AI deployment closer to customers.

Disclosure boundary

What is known—and what is not

Primary company announcement; full SCA contract, spend ramp, reserved-capacity mechanics, region-by-region allocation, AWS chip mix, pricing terms, termination rights, and customer workload allocation are not public.