Snowflake has signed a $6 billion, five-year agreement with Amazon Web Services, the companies announced on Wednesday, May 27, 2026. The deal centers on Snowflake securing access to AWS’s ARM-based CPU chip, Graviton, to support its growing AI workloads.
To put the scale in perspective, Snowflake has generated $7 billion in total sales through the AWS Marketplace since its founding in 2012 — meaning this single contract is nearly equivalent to its entire cumulative revenue from that platform. Snowflake’s customer spending on AWS has been accelerating, doubling in 2025 to $2 billion for that calendar year alone.
The driver behind the deal is AI adoption. Snowflake has offered its AI building tool, Cortex AI, for several years. The tool allows enterprise users to query databases using plain language, generate summary reports, and build AI-powered features on top of data already stored in Snowflake. As AI shifts from model training toward daily use and automated agents, CPU demand has surged — CPUs handle the bulk of tasks associated with AI agents, while GPUs are primarily used for training and reasoning.
Amazon CEO Andy Jassy stated last month that the company’s homegrown AI chips offer better price-performance than Nvidia’s, and that savings are passed on to customers. AWS recently signed a separate deal to supply millions of Graviton chips to Meta, following Meta’s $10 billion agreement with Google Cloud.
The Snowflake deal adds to a pattern of large cloud contracts built around non-Nvidia chips. Nvidia CEO Jensen Huang responded last week by pointing to his company’s new AI-specific CPU, called Vera, which he described as representing a $200 billion market opportunity, with $20 billion in sales already secured.
The deal suggests that AWS’s investment in its own chip infrastructure may be drawing enterprise customers who are looking for cost-effective AI compute options, even as Nvidia remains a dominant force in the broader market.
Source: TechCrunch