In a move that redraws the map of the artificial intelligence arms race, OpenAI has announced a staggering $38 billion, seven-year partnership with Amazon Web Services (AWS). The deal secures OpenAI access to a massive trove of computing power—what one analyst is calling “the new oil”—to fuel its voracious appetite for training next-generation AI models.

This isn’t just another cloud contract; it’s a tectonic shift.

For years, OpenAI and Microsoft have been the tech world’s definitive power couple. Microsoft’s multi-billion dollar investments and its Azure cloud platform were the exclusive engine behind the development of ChatGPT and GPT-4. But that exclusivity is now officially over.

The $38 billion agreement with Amazon fundamentally alters OpenAI’s strategy, transforming it from a single-partner protégé into a multi-cloud behemoth that can pick and choose the best hardware from the world’s biggest players.

At the heart of this deal is one thing: compute.

The global race to build more powerful AI is bottlenecked by a severe shortage of specialized computer chips, specifically the high-performance Graphics Processing Units (GPUs) made by NVIDIA. These chips are the only things on Earth capable of performing the quadrillions of calculations needed to train and run models like GPT-4.

The AWS partnership gives OpenAI a secured, priority-access pipeline to “hundreds of thousands” of these coveted NVIDIA GPUs. According to the press release, this includes NVIDIA’s latest and most powerful models, the GB200 and GB300 AI accelerators, which will be deployed on Amazon’s specialized EC2 UltraServers.

This isn’t a “maybe someday” plan. OpenAI will “immediately” begin using this new AWS capacity to train its models. The full deployment is aggressively fast-tracked, with all compute targeted to be online before the end of 2026 and options to expand even further into 2027.

In short, OpenAI just bought itself a dedicated supercomputer, spread across Amazon’s global data centers, for the better part of a decade.

The most fascinating part of this story is what it says about OpenAI’s relationship with its primary backer, Microsoft. The user-provided content noted that Microsoft is “loosening its grip,” and that’s putting it mildly.

This deal was made possible by a recent, complex restructuring at OpenAI. Just last week, the company finalized its shift from a “capped-profit” hybrid into a more conventional for-profit “public benefit corporation.” As part of that change, its foundational partnership with Microsoft was rewritten.

Microsoft lost two critical advantages:

1. It is no longer OpenAI’s exclusive cloud provider.
2. It has lost the “first right of refusal” to host OpenAI’s AI workloads.

This means that for the first time, OpenAI was free to shop around. And it went shopping with one of the biggest briefcases in history, walking straight to Microsoft’s number one cloud rival.

However, this is not a divorce. It’s an open relationship. OpenAI’s new deal with Microsoft is still gargantuan, reportedly involving a $250 billion commitment to use Azure services. What’s changed is that OpenAI is no longer putting all its eggs in one basket. The AI powerhouse is now diversifying, spreading its massive computational needs across Microsoft Azure, Amazon’s AWS, and even other players like Oracle, with whom it has a reported $300 billion deal.

Why? Because no single company—not even Microsoft—has enough compute to satisfy OpenAI’s ambitions.

For Amazon, this is a monumental victory.

In the past year, investors have been worried that AWS was falling behind in the generative AI gold rush. Microsoft’s Azure, thanks to its OpenAI exclusivity, was seen as the undisputed leader. Google had its own powerful in-house AI (Gemini) and custom chips. Amazon, while still the world’s largest cloud provider, lacked a clear “killer app” partner.

This $38 billion deal single-handedly silences those critics. It validates AWS as a top-tier platform for serious, large-scale AI development. The market’s reaction was immediate: Amazon’s stock surged on the news, adding an estimated $140 billion to its market capitalization.

Matt Garman, the CEO of AWS, called the deal a “powerful reminder” of why organizations trust AWS for “serious scale, security, and performance.” It’s a clear signal to the rest of the industry that Amazon is competing directly with Microsoft to power the future of AI.

The final goal: fueling the AGI engine

This move, combined with the recent Microsoft deal, sheds light on OpenAI’s ultimate endgame: Artificial General Intelligence (AGI).

The user’s prompt noted that OpenAI’s new deal with Microsoft gives the company rights to its technology “until it reaches advanced general intelligence (AGI).” This isn’t a casual mention; it’s the entire point.

OpenAI’s leadership, including CEO Sam Altman, has been clear that “scaling frontier AI requires massive, reliable compute.” The path to AGI—an AI with human-level cognitive abilities—is paved with data and electricity, running through millions of GPUs.

The company’s infrastructure commitments are now valued at over $1 trillion. This $38 billion deal with Amazon is just one more (very large) down payment on the astronomical cost of building a god. By securing compute from every major provider, OpenAI is ensuring that when it makes its next big leap—whether it’s GPT-5 or a true AGI—it won’t be slowed down by simply running out of power.