Rotating Data
Explore the map
Home/Deal Room/AWS brings Project Rainier AI compute cluster online for Anthropic

Cloud capacity · Evidence brief

AWS brings Project Rainier AI compute cluster online for Anthropic

Amazon Web Services + Anthropic

Primary sourcePress release only2026-05-26
Deal structureDedicated frontier-model AI compute cluster / custom accelerator deployment
GeographyMulti-site U.S. AWS data-center footprint, including St. Joseph County, Indiana
Infrastructure layersCloud capacity + chips + data centers
Disclosed scaleNearly 500,000 AWS Trainium2 chips online now, with Anthropic expected to use more than 1M Trainium2 chips for Claude training and inference by year-end

Why it matters

The strategic read

Adds the clearest AWS/Anthropic hyperscaler AI cluster to the map. Rainier is not just generic cloud capex; it is a named, custom Trainium2 UltraCluster that Anthropic is already using for Claude, making AWS custom silicon, Indiana data-center sites, power/cooling design, and Anthropic frontier-model demand one connected infrastructure signal.

Source-supported terms

What was disclosed

  1. 01

    AWS says Project Rainier is fully operational and represents one of the world’s largest AI compute clusters.

  2. 02

    The cluster includes nearly 500,000 Trainium2 chips deployed less than one year after announcement.

  3. 03

    AWS says Anthropic is actively using Project Rainier to build and deploy Claude.

  4. 04

    Anthropic is expected to run Claude on more than 1M Trainium2 chips for training and inference by year-end.

  5. 05

    Project Rainier is spread across multiple U.S. data centers, with Amazon identifying St. Joseph County, Indiana as one Project Rainier site with outdoor-air cooling advantages.

Disclosure boundary

What is known—and what is not

Primary AWS product/infrastructure announcement; full Anthropic commercial terms, complete site list, power capacity, capex, utility interconnection agreements and site-level construction schedule are not public.