Amazon.com Inc.'s cloud unit, Amazon Web Services (AWS), has secured a monumental $38 billion deal to provide computing power to OpenAI, the company behind ChatGPT. This seven-year agreement grants OpenAI access to hundreds of thousands of Nvidia Corp. graphics processing units (GPUs) hosted by AWS. The announcement on Monday highlighted OpenAI's transformation from a research lab into a dominant artificial intelligence powerhouse, reshaping the tech industry. OpenAI has previously committed an unprecedented $1.4 trillion to infrastructure for building and operating advanced AI models, a scale raising concerns about a potential investment bubble in the sector.

For Amazon, the deal endorses its expertise in building and managing vast data center networks, an area where it has pushed to compete in the AI race. Matt Garman, AWS CEO, stressed the partnership's value: “AWS's infrastructure will serve as a crucial backbone for OpenAI's ambitious AI endeavors.” Post-announcement, Amazon shares rose 4.5% to $255.29 in New York, while Nvidia's climbed 3.3% to $209.20.

As the world's largest rented computing power provider, AWS was previously an outlier among U.S. majors partnering with OpenAI. Microsoft Corp. OpenAI's top investor and former exclusive cloud provider, announced a $250 billion commitment for its Azure unit. Oracle Corp. signed a $300 billion data center deal, and earlier this year, OpenAI named Alphabet Inc.'s Google Cloud Platform among its partners. The AI startup also has a $22.4 billion agreement with CoreWeave Inc., a rising “neo cloud” for AI developers.

Analysts like Bloomberg Intelligence's Anurag Rana and Andrew Girard suggest adding AWS could ease pressure on OpenAI, especially as it spreads contracts to smaller neo-clouds like CoreWeave. They noted AWS's global data center network could aid OpenAI's international growth. Under the deal, OpenAI begins using AWS capacity immediately, with full deployment by end of 2026 and an option to extend. Amazon will deploy hundreds of thousands of advanced Nvidia chips, including GB200 and GB300 AI accelerators, in custom clusters to boost ChatGPT responses and train next-gen models.

Sam Altman, OpenAI CEO, emphasized the critical need for massive, reliable compute to scale frontier AI. “Scaling frontier AI requires massive, reliable compute,” he said. “Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”