Amazon Opens $11 Billion AI Data Center in Rural Indiana.
Built specifically to train and operate Anthropic’s AI models, Amazon has established a gigantic data center spanning 1,200 acres in Indiana.
While the majority of Amazon’s AI rivals are still discussing data centers that are years away, the $11 billion building is already fully operating.
Amazon Web Services (AWS) CEO Matt Garman characterized the swift change as:
“Almost overnight, from cornfields to data centers.”
From a farm to one of the biggest AI data centers in the world
Indiana’s New Carlisle This was farmland only a year ago. The 1,200-acre location next to Lake Michigan is now home to one of the biggest operational AI data centers in the world. Amazon is using its own processors to train state-of-the-art AI models under a project called Project Rainier.
More than $1 trillion has been committed to AI data center initiatives by Amazon and its competitors. Given how ambitious these ideas are, experts wonder if there will be enough funding, effort, and community support to see them through to completion.
“Stargate,” OpenAI’s massive network of AI data centers, is currently in development. Amazon’s answer is Project Rainier – a $11 billion facility that isn’t just an idea but fully built and functioning.
Anthropic, the firm that created the Claude chatbot and is one of Amazon’s largest cloud customers and AI partners, has built the facility especially to train and run its AI models.
Garman declared, “This isn’t some future plan.” “Their models are training and operating here today, so this is real.”
The Worldwide Competition for AI Supercomputing
In order to fulfill the enormous increase in demand for AI, IT firms are rushing to construct supercomputing hubs.
While Alphabet, the parent company of Google, has started construction on a new location in Arkansas, directly across the river from Elon Musk’s xAI Colossus Data facility, Meta is building a 2-gigawatt “Hyperion” facility in Louisiana.
In less than a month, OpenAI revealed 33 gigawatts of additional processing power, which CEO Sam Altman claims amounts to $1.4 trillion in future promises.
Amazon’s Advantage: Quickness and Experience
Because of its decades of experience in large-scale logistics, Amazon has been able to produce results faster than most of its competitors.
The company’s strong partnerships with state and municipal governments have enabled it to create AI infrastructure at remarkable pace, from its HQ2 project to enormous fulfillment centers and AWS facilities.
Anthropic Chief Product Officer Mike Krieger commended Amazon’s performance:
“These initiatives seem fantastic on paper. However, true success occurs when customers can truly utilize them, and Amazon is unparalleled in facilitating that.
Project Rainier was made public just one day prior to Amazon’s third-quarter financial results. Investors are keeping a careful eye on how soon the company’s capital expenditures will result in income and profits.
Amazon Opens $11 Billion AI Data Center in Rural Indiana
Amazon announced 14,000 layoffs earlier this week as part of a larger reorganization initiative that aims to streamline management and reallocate resources to the development of AI and its Trainium chip.
“From Fields to Data Centers” – Project Rainier’s Story
In the spring of 2023, Project Rainier’s foundation was laid.
Amazon began hunting sites in rural Indiana around six months after ChatGPT opened, working with Indiana Michigan Power, a division of American Electric Power. The business made the biggest investment in the state’s history when it signed a $11 billion contract a year later.
As of this month, seven buildings have been put online since work started in September 2024, and two more campuses are still being built. Eventually, the entire complex will include 30 buildings and use 2.2 gigawatts of electricity, which is sufficient to power more than 1.6 million homes.
Josh Sallabedra, who has built Amazon data centers for 14 years, is currently in charge of the Indiana location. To supervise the project, he relocated from the West Coast.
He stated, “We had to completely change our building model because the customer demand is so intense.” Amazon moves more quickly than I’ve ever seen.
Local Communities Voice Their Concern
The location still has the air of a construction site, with vehicles sending dust flying along unfinished roads, steel beams rising in the distance, and workers in safety jackets moving between trailers.
A $3.5 billion electric vehicle battery facility is being built nearby by General Motors and Samsung. In a town with only 1,900 residents, almost 4,000 laborers arrive each day.
However, not everyone is pleased with the adjustments.
New Carlisle Town Council President Marcy Kauffman stated:
It’s challenging to continue losing farms. Much of it was devoted to this endeavor.
Dan Caruso, a local, expressed that worry:
“Once they get a foot in, they’ll keep expanding,” my buddies cautioned me. And that is precisely what is taking place.
By the end of the decade, energy consumption is predicted by Indiana Michigan Power to more than quadruple, which might have an impact on residential electricity rates. According to reports, over the past five years, bills in the vicinity of new data centers have increased by up to 267%.
Innovative Rainier Powered by Amazon Chips
Amazon’s Trainium 2 chips are installed in all seven Rainier buildings; noticeably, no Nvidia chips are present.
This represents the world’s largest non-Nvidia compute cluster.
“In Indiana, we’re currently running about 500,000 chips,” Garman stated. “We’ve doubled our order because everything is going so well.”
By the end of the year, the company hopes to have one million chips.
In the upcoming months, the next-generation Trainium 3, created in partnership with Anthropic, will be made available.
The Growing Amazon-Anthropic Collaboration
Anthropic is one of the first significant AI labs to train models on Amazon’s own silicon because its whole infrastructure is powered by AWS.
Amazon’s larger AI strategy includes a $8 billion investment in Anthropic.
Although Trainium’s raw performance isn’t yet on par with Nvidia’s GPUs, AWS asserts that its technology offers greater efficiency and lower energy consumption.
According to Krieger and Garman, Anthropic directly influenced Amazon’s technical teams to improve energy efficiency, decrease latency, and speed up training.
Garman clarified that the objective of Trainium 3 is to provide support for the upcoming generation of frontier models. It will be implemented not only in Indiana but also in all of our data centers worldwide and provides better performance and reduced power consumption.
Anthropic’s Quick Development
Anthropic is rapidly expanding and making significant investments to stay competitive.
The business generates close to $7 billion in revenue annually. In just two years, the number of enterprises using its Claude chatbot has increased by 300 times to over 300,000.
Over the past year, the number of large enterprise clients—each of whom generates over $100,000 annually—has increased sevenfold.
Within two months of its launch, Anthropic’s newest product, Claude Code, an AI coding assistance, produced $500 million in annualized income.
However, Anthropic is not entirely dependent on Amazon.
It revealed last week that it had entered into a significant agreement worth tens of billions of dollars with Alphabet, the parent company of Google, that will grant it access to up to one million Google Tensor Processing Units (TPUs).
Krieger clarified,
“Without a multi-chip strategy, we could not have scaled this quickly due to the enormous demand for our models.”
While acknowledging Anthropic’s collaborations, Garman made it apparent that Amazon had no intention of purchasing the business.
He declared, “We adore the partnership just the way it is.”
Amazon Opens $11 Billion AI Data Center in Rural Indiana — Competitors Rush to Catch Up.
