OpenAI has reached a massive $38 billion agreement with Amazon, allowing the ChatGPT creator to run its artificial intelligence systems on Amazon’s U.S.-based data centers.

Under the deal announced Monday, OpenAI will gain access to “hundreds of thousands” of Nvidia AI chips through Amazon Web Services (AWS) to power and expand its AI tools. Following the announcement, Amazon’s shares rose 4%.

The agreement comes just days after OpenAI restructured its longstanding relationship with Microsoft, which had previously been its exclusive cloud computing partner. Regulators in California and Delaware also approved OpenAI’s new corporate structure last week, enabling the San Francisco-based company — originally founded as a nonprofit — to raise capital more easily and operate for profit.

“The rapid advancement of AI technology has created unprecedented demand for computing power,” Amazon said in a statement. The company noted that OpenAI will “immediately begin using AWS computing capacity,” with all infrastructure expected to be in place by the end of 2026, and room to expand further into 2027 and beyond.

Developing and maintaining AI systems like ChatGPT requires enormous amounts of energy and computing resources. OpenAI has made over $1 trillion in financial commitments to secure such infrastructure, including partnerships with Oracle, SoftBank, and major chipmakers Nvidia, AMD, and Broadcom.

Some investors have questioned the sustainability of these deals, given that OpenAI remains unprofitable and relies on future revenue to cover its growing infrastructure costs. CEO Sam Altman, however, dismissed such concerns, saying on a recent podcast with Microsoft CEO Satya Nadella that “revenue is growing steeply” and that OpenAI is “making a forward bet” on continued expansion.

Amazon already serves as the primary cloud provider for Anthropic, one of OpenAI’s top competitors and the developer of the Claude chatbot.