Amazon Revolutionizes Its AI Data Center in Indiana for the AGI Era

An Ambitious Transformation for AI

Amazon is undertaking a significant transformation of its data center campus in Indiana, aiming to create a network of powerful computers for its future artificial intelligence models. According to internal documents, this initiative plans to consolidate several data centers and deploy thousands of AI servers equipped with Trainium technology.

Sources close to the company revealed that this project is part of a broader initiative called "AGI Pivot," which supports Amazon's AGI organization and its future AI models. These sources preferred to remain anonymous due to the confidentiality of the information.

The documents indicate that Amazon maintains its ambitions for cutting-edge AI models, despite recent layoffs in its AGI organization and the discontinuation of the Nova model series. They also highlight the urgency of providing sufficient computing capacity to train the next major model of the AGI organization before the end of the year, amid industry-wide capacity constraints.

Optimizing Investments

The redesign of the campus also reflects Amazon's effort to maximize the return on its colossal investments in AI infrastructure. By connecting existing data centers to form a larger and more efficient AI cluster, Amazon is optimizing the use of the infrastructure it has already funded.

This strategy aligns with statements from CEO Andy Jassy, who recently asserted that AI infrastructure should generate attractive long-term returns. Data centers remain productive for decades, while servers and networking equipment can be renewed over time.

Collaboration with Anthropic

The Indiana campus also houses facilities for Project Rainier, an AI supercomputer powered by Trainium, primarily built for Anthropic. The new AGI initiative is managed from the same complex, but it will not affect the existing servers of Project Rainier.

A spokesperson for Amazon stated that the company is constantly designing, upgrading, and enhancing its data center infrastructure to meet the diverse needs of its customers and teams. This work aims to improve speed, cost, and sustainability to meet the growing demand for computing on AWS.

Towards an AGI SuperCluster

Planning documents show that Amazon is looking to increase computing capacity for its next cutting-edge AI model. A recent update mentions an "emerging demand" to deploy more than 6,000 AI servers powered by Trainium, accelerating launch timelines by several weeks. This rapid deployment aims to prepare Amazon's next AI model for this year's re:Invent conference, typically held in early December.

Amazon also plans to consolidate computing on the Indiana campus into what it calls an AGI SuperCluster, creating a "larger and more efficient" AI system to maximize performance. To achieve this, the company is redesigning the network, storage, and fiber optic infrastructure so that multiple data centers operate as a single entity. The project includes converting some existing buildings into "annexes," linking them to neighboring data centers to share core networking equipment.

A Strengthened Commitment to AI

Amazon is significantly increasing its investment in AI. The company recently raised its capital expenditure forecast for 2026 to $220 billion, up from $200 billion, as demand for AI computing continues to outstrip available capacity and prices for certain components rise.

This latest infrastructure effort fits into a broader pattern within Amazon. The company has previously reported its intention to modernize its AI infrastructure through projects such as Houdini and Titus, while reorganizing its AGI division around a new cutting-edge model effort.

The initiative underscores the growing importance of Trainium within Amazon. The company recently stated that its custom chip business, including Trainium AI chips and Graviton processors, is on track to generate more than $25 billion in annual revenue, up from a projection of $20 billion the previous quarter.

Planning documents also provide insight into the needs for Amazon's next cutting-edge model effort. They anticipate expanding data storage to support multimodal AI training, requiring the processing of a large number of high-resolution images and the repeated backup of model progress. A source close to the project claimed that the initiative "has not slowed down," despite recent layoffs in the AGI organization.