OpenAI Inks ‘Historic’ Deal With Oracle, Papered at $30 Billion Annually for Services from Data Center: Report

OpenAI, which is building artificial general intelligence to benefit humanity and powers a suite of next-gen AI technologies, has inked one of the largest commercial cloud deals in history with Oracle to support the organization’s high-performance computing workloads.

New reports peg the agreement at a $30 billion per annum pact—a staggering sum that underscores the scale and intensity of the compute race animating generative AI development.

OpenAI CEO Sam Altman didn’t confirm the financials, but did acknowledge the partnership in a post on X (formerly Twitter), as well as an update to the company’s blog on Tuesday:

“We are working to build the infrastructure for not just the current generation of extremely large-scale models, but for future generations of even more powerful models,” Altman wrote, emphasizing that OpenAI requires scale and complexity that Oracle is equipped to handle.

The Billion-Dollar Backbone of AI

OpenAI’s demand for compute has skyrocketed in recent years, as the group continues to push the frontiers of AI research and implementation. For GPT-4, GPT-4o, enterprise tools, and ChatGPT’s premium offerings, the models need:

• Huge clusters of GPUs
• Fast networking
• Reliable uptime
  —to serve hundreds of millions of users globally.

This collaboration with Oracle will greatly enhance OpenAI’s access to GPU-packed data centers, enabling:

• Faster training of large language models (LLMs)
• Low latency for real-time serving
• Better scalability for enterprise-grade production

Under the agreement:

• Oracle will deliver a dedicated AI supercluster powered by NVIDIA GPUs
• It will utilize Oracle Cloud Infrastructure (OCI)
• The setup will feature:
  • Tens of thousands of high-speed chips
  • Liquid-cooling systems
  • Ultra-low latency interconnect technology

These elements are tailored specifically for the demanding workloads of generative AI.

“This is a lot more than a cloud contract. This is infrastructure that’s being laid down for the future of A.I.,” said Dan Morgan, senior portfolio manager with Synovus Trust.
“The $30 billion number, if it’s right, is a reflection of how far OpenAI is willing to push to secure reliable compute in an increasingly competitive AI landscape.”

Oracle’s Victory in the A.I. Cloud Wars

The deal is a major win for Oracle, a company often eclipsed by cloud giants like Amazon Web Services (AWS), Microsoft Azure, and Google Cloud.

Traditionally known for its enterprise database products, Oracle has expanded its offering over recent years to include:

• Image recognition
• Large-scale AI services
• Automation of enterprise tasks

Its AI-fueled capabilities have gained momentum during a time when the world finds itself in a full-blown arms race for AI dominance.

Oracle has:

• Invested billions in AI-capable data centers
• Partnered with NVIDIA to deliver elastic, high-performance GPU clusters

Through this deal, Oracle gains:

• A marquee customer in OpenAI
• A major endorsement of its infrastructure capabilities

“Oracle is now squarely in the discussion for a leading AI cloud provider,” said Lisa Tran, a cloud computing analyst with Forrester.
“This could change the competitive landscape for them practically overnight, especially if other A.I. companies adopt OpenAI’s lead.”

Oracle Chairman and CTO Larry Ellison, a long-time AI advocate, had previously claimed that Oracle Cloud could handle the most demanding AI models. This partnership might be the strongest validation yet of those assertions.

Scaling AI While Compute Costs Soar

The deal also highlights the massive cost and effort behind modern AI development.

Building, training, and operating frontier models like GPT-4o or multimodal systems requires:

• Vast infrastructure investment
• Significant energy consumption
• Ongoing capital expenditure

OpenAI’s scale is particularly notable:

• It is reportedly training multiple successor models simultaneously
• Developing tools in speech, vision, robotics, and enterprise AI
• Maintaining tens of millions of daily ChatGPT users

Its computational requirements are expected to increase exponentially.

By signing a long-term deal with Oracle, OpenAI:

• Secures access to essential compute resources
• Gains preferential GPU allocation—a key move in a market where chips are scarce and demand is high

“This is OpenAI making a big bet on continued scale,” said Jayesh Mehta, a machine learning engineer and industry observer.
“When your business is predicated on running models 24/7 at global scale, you need more than just a cloud vendor—you need a strategic infrastructure partner.”

Strategic Shifts and Industry Ripples

The OpenAI–Oracle partnership also raises questions about OpenAI’s overall infrastructure strategy, particularly its longstanding relationship with Microsoft, one of its largest investors.

Microsoft has:

• Invested over $10 billion in OpenAI
• Provided compute power through Azure
• Supported much of OpenAI’s research and API services

Industry observers say the Oracle deal is not a departure from Microsoft but a strategic diversification of compute providers.

“Given model training lasts for weeks or months and involves billions of parameters, OpenAI can’t afford any downtime or GPU shortage,” said Tran.
“Diversifying the load among different providers is a reasonable, albeit expensive, next evolution.”

This deal also signals to other AI firms that compute access is the new battleground. As startups and enterprises work to develop foundational models, providers like Oracle, AWS, and Azure will compete fiercely to supply the infrastructure.

Looking Ahead

While neither OpenAI nor Oracle confirmed the $30 billion annual figure reported by The Wall Street Journal, both acknowledged that the deal is long-term and strategic.

• For OpenAI, it is an investment in stability, performance, and growth
• For Oracle, it represents a breakthrough moment and a chance to lead the next generation of commercial AI infrastructure

“This is the campfire around which humanity organizes itself,” Altman wrote in a blog post.
“We’re honored to be partnering with Oracle, one of the world’s leading technology providers, as we work to get to what’s next.”

As artificial intelligence continues to evolve at breakneck speed, one thing is clear: the race for compute dominance has only just begun. And in that race, partnerships like this may well determine who leads—and who lags behind.