On May 27, Snowflake announced a multi-year strategic agreement with Amazon Web Services (AWS). Over the next five years, Snowflake will spend $6 billion on services and technologies from AWS, including Amazon’s in‑house Arm‑based Graviton general‑purpose processors and GPUs designed for AI training and inference. This represents the largest single procurement commitment in Snowflake’s history, marking a substantial increase from the $2.5 billion pledged in 2023. Since its inception, Snowflake has generated over $7 billion in sales via the AWS Marketplace; in 2025 alone, those sales exceeded $2 billion — double the previous year’s figure. On the same day, Snowflake released results for the first quarter of fiscal year 2026 (ended April 30): adjusted earnings per share came in at 39 cents, while revenue reached $1.39 billion, a 33% rise year‑over‑year. Both figures surpassed analyst expectations. The company also forecast product revenue for the second quarter between $1.415 billion and $1.420 billion, again exceeding market forecasts. Driven by these announcements, Snowflake’s stock surged more than 33% in after‑hours trading. Additionally, Snowflake revealed plans to acquire Natoma, an enterprise‑grade MCP (Model Context Protocol) platform, to strengthen the connectivity between AI agents and corporate data; financial details of the deal remain undisclosed.
This pact is yet another major AI infrastructure commitment attracted by AWS in recent times. Previously, Anthropic pledged to invest over $100 billion in AWS over ten years, while Meta signed a contract to purchase millions of Graviton chips. Unlike those deals, Snowflake’s arrangement involves no equity investment whatsoever. Looking at broader industry trends, as AI applications evolve from simple chatbots to task‑oriented agents, demand for CPUs such as Graviton is rebounding due to their proficiency in cross‑agent coordination and large‑scale data orchestration. Consequently, the Arm architecture is rapidly supplanting the x86 architecture that has dominated data centers for decades. Google and Microsoft have likewise rolled out their own Arm‑based chips, intensifying competition against NVIDIA in the realm of cloud‑provided computing power.