Amazon is redesigning its Indiana data center campus into an AGI SuperCluster to train its next frontier AI model, with over 6,000 Trainium servers deployed ahead of re:Invent.
Amazon is redesigning part of a massive AI data center campus in rural Indiana into a single, powerful cluster of computers built to train its next frontier models, Business Insider reported, citing internal planning documents.
The documents describe an effort to consolidate multiple data centers and deploy thousands of Trainium-powered AI servers. People familiar with the matter said the project is part of a broader initiative called "AGI Pivot" that supports the company's AGI organization and its future in-house AI models. These people asked not to be identified discussing private plans.
The redesign suggests Amazon's frontier-model ambitions remain intact despite recent job cuts in its AGI organization and the wind-down of the previous Nova model series. The documents describe faster deployment schedules intended to provide enough computing capacity for the AGI organization to train its next big model before the end of this year, showing the urgency amid industrywide capacity constraints, the people said.
One update describes an "emergent request" to deploy more than 6,000 Trainium-powered AI servers, speeding up launch schedules by several weeks. The accelerated timeline was intended to prepare Amazon's next AI model for this year's re:Invent conference, typically held in early December, according to this document.
The documents also describe a broader effort to consolidate computing across the Indiana campus into what Amazon calls an AGI SuperCluster, creating a "larger, more efficient" AI system to maximize performance.
To do that, Amazon is redesigning networking, storage, and fiber-optic infrastructure so multiple data centers function as one. The project converts some existing buildings into what Amazon calls "annexes," connecting them to neighboring data centers so they can share core networking equipment instead of operating independently. Some facilities are planning to replace older Trainium 2 systems with newer Trainium 3 servers.
The redesign reflects Amazon's broader effort to maximize returns on its record AI infrastructure investments. By linking existing data centers into a larger, more efficient AI cluster, the company is making better use of infrastructure it has already paid for and operates. The approach aligns with CEO Andy Jassy's comment last week that AI infrastructure should generate attractive long-term returns because data centers remain productive for decades while servers and networking equipment can be refreshed over time.
The Indiana campus includes facilities that are part of Project Rainier, the Trainium-powered AI supercomputer Amazon built primarily for Anthropic. The new AGI initiative is run out of the same huge data center complex. It will not affect the existing Project Rainier servers, the people said.
"We're always designing, upgrading, and improving our data center infrastructure to serve the diverse needs of our customers and teams," an Amazon spokesperson told Business Insider. "That work improves speed, cost, and sustainability so we can meet growing demand for compute to train and serve models on AWS."
Amazon is significantly expanding its AI investment. The company recently raised its projected 2026 capital expenditures to $220 billion from $200 billion as demand for AI computing continues to outstrip available capacity and prices for some components rise.
The latest infrastructure effort fits a broader pattern inside Amazon. Business Insider previously reported on the company's push to upgrade its AI infrastructure through projects such as Houdini and Titus, while reorganizing its AGI division around a new frontier-model effort.
The initiative highlights the growing importance of Trainium inside Amazon. Last week, the company said its custom chip business, including Trainium AI chips and Graviton processors, is on pace to generate more than $25 billion in annual revenue, up from last quarter's $20 billion projection.
The planning documents also offer a glimpse of what Amazon's next frontier-model effort will require. They call for expanding data storage to support multimodal AI training, which requires processing large numbers of high-resolution images and repeatedly saving the model's progress.
One person familiar with the effort said the push has "not slowed down," despite last month's layoffs in the AGI organization.
For broader context on how these infrastructure moves fit into the market, see our stock market analysis.
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.