AWS’ $38 Billion OpenAI Deal ‘Exposes How Dependent AI’ Is On Centralized Infrastructure, AI Startup CEO Says

“The industry’s biggest players are now effectively trading power to control intelligence the way others trade energy—concentrating enormous financial and operational power in a few providers,” said Eric Yang, CEO of AI startup Gradient.

The world's largest cloud company signed a massive $38 billion deal with AI superstar OpenAI this week that sent shockwaves through the tech industry due to the eye-popping deal size. OpenAI has immediately started running large amounts of workloads on AWS’ cloud infrastructure with huge capacity expansions planned over the next several years.

Gradient CEO Eric Yang said the scale of the new AWS and OpenAI cloud deal shows how quickly AI has become a capital market in its own right.

“That model can deliver short-term efficiency but also exposes how dependent AI has become on centralized infrastructure,” Yang said in an email to CRN.

With the $38 billion deal, AWS will provide OpenAI with the computing power it needs for its popular ChatGPT AI and help scale AI agents faster.

OpenAI will access AWS’ compute resources that are comprised of hundreds of thousands of Nvidia GPUs, with the ability to expand to tens of millions of CPUs to rapidly scale agentic AI workloads.

Tech Industry Must Ensure AI ‘Intelligence Itself Doesn’t Stay Trapped’

Yang said having the majority of AI intelligence and workloads concentrated on centralized infrastructure, like the public cloud, is concerning.

“The next challenge is ensuring that intelligence itself doesn’t stay trapped there,” the CEO said.

“The real opportunity lies in building systems that can operate independently of those central clouds—where people and organizations can run and evolve AI on their own terms,” he said.

Founded in 2024, Yang’s AI startup Gradient is developing ways to build and run AI without needing to plug into the cloud infrastructure market led by AWS, Microsoft and Google Cloud.

“That’s the foundation of a more sovereign approach to intelligence, one that grows from the edge rather than the center,” Yang said.

This week Gradient unveiled it has open-sourced Parallax, the company’s new operating system for AI that lets people and teams build, host, and access AI applications on their own devices and networks.

“Parallax is the easiest way to host your own agents and applications privately, efficiently, and without relying on the cloud,” said Yang.

AWS And OpenAI’s $38 Billion AI Deal

The infrastructure deployment that AWS is building for OpenAI features a sophisticated architectural design optimized for maximum AI processing efficiency and performance.

AWS said clustering Nvidia GPUs—both its GB200s and GB300s—via Amazon EC2 UltraServers on the same network enables low-latency performance across interconnected systems, allowing OpenAI to efficiently run workloads with optimal performance. These clusters are designed to support various workloads, from serving inference for ChatGPT to training next-generation models.

“With this new $38B agreement, OpenAI will immediately start using our world-class infrastructure—including Amazon EC2 UltraServers packed with hundreds of thousands of state-of-the-art Nvidia GPUs and the ability to scale to tens of millions of CPUs,” said AWS CEO Matt Garman in a LinkedIn post this week.

OpenAI CEO and co-founder Sam Altman said his company needs massive amounts of compute capability to scale AI to the masses.

“Scaling frontier AI requires massive, reliable compute," said Altman in a statement. “Our partnership with AWS strengthens the broad compute ecosystem that will power this next era and bring advanced AI to everyone.”

AWS recently reported total sales of $33 billion during its third quarter of 2025, up 20 percent year over year compared to Q3 2024. The company now has a $132 billion annual run rate.