Amazon’s $11 billion AI data center in Indiana opens as rivals race to break ground

NEW CARLISLE, Indiana — Transforming what was once farmland, Amazon has inaugurated a sprawling 1,200-acre facility near Lake Michigan, now recognized as one of the world's largest operational AI data centers. Named Project Rainier, this site is dedicated to training advanced artificial intelligence models utilizing Amazon's proprietary chips. With over $1 trillion pledged by tech giants for AI infrastructure, the ambitious nature of these projects has led many to question the feasibility of such investments. Amazon's $11 billion Rainier facility stands as a significant response to this wave, already operational and designed specifically for models developed by Anthropic, a key AI partner and cloud customer of Amazon. "This isn’t a future plan; it’s happening now," stated Matt Garman, CEO of Amazon Web Services, during a recent interview. This data center is part of a broader trend among major tech companies racing to establish supercomputing facilities to meet soaring demand. Competitors like Meta and Google are also planning substantial projects across the United States. The announcement of Rainier coincided with Amazon's third-quarter earnings report, which investors are keenly monitoring for insights into capital expenditures and revenue conversion timelines. The company also recently announced a reduction of 14,000 jobs as part of a strategic refocus on AI and other priority areas. The origins of the Rainier complex trace back to early 2023 when Amazon began scouting sites in rural Indiana. After signing the largest capital investment agreement in the state's history, construction kicked off in September 2022, culminating in the current operational status of seven buildings, with more planned. This site is set to consume over 2.2 gigawatts of electricity, enough to power around 1.6 million homes. Josh Sallabedra, the project's site lead, noted the rapid pace of construction, attributing it to strong customer demand for AI capabilities. Amazon's distinct approach contrasts with some competitors who are erecting temporary structures; it emphasizes a well-planned infrastructure designed for reliability and security. As construction progresses, the local community grapples with the rapid transformation of their landscape. While some residents express concerns over the loss of farmland, others acknowledge the economic opportunities that such large-scale projects can bring. Rainier is equipped with Amazon's custom chips, Trainium 2, marking a significant milestone as the largest deployment of non-Nvidia computing technology globally. Garman revealed that the facility is currently running around 500,000 chips, with expectations to increase that number to one million by year's end. Moreover, collaborations between Amazon and Anthropic have led to the development of Trainium 3, set to enhance performance and efficiency in AI model training. Anthropic's rapid growth and increasing revenue underscore the competitive landscape in the AI sector, where partnerships, like the recent agreement with Google, are essential to meet escalating demands. Amidst this technological evolution, Amazon remains committed to expanding its capabilities in response to customer needs, indicating that the construction and development of AI infrastructure is far from over. As Garman emphasized, the surge in demand necessitates continuous growth in capacity and innovation.