Oracle has expanded its AI infrastructure by increasing its supply agreement with NVIDIA and adding chips from Google and other competitors to its cloud platform. The move gave Oracle more options for running AI workloads as demand for computing power continues to grow.
Announced on Tuesday (October 27, 2026), the expanded agreement will allow Oracle to deploy NVIDIA’s latest Vera Rubin AI server systems at a scale measured in multiple gigawatts. The companies did not disclose the deal's financial value, but the expansion highlights Oracle’s growing investment in infrastructure for advanced AI applications.
At the same time, Oracle is widening its supplier base. The company plans to offer Google’s Tensor Processing Units (TPUs) through Oracle Cloud Infrastructure (OCI) for the first time, alongside processors from Cerebras Systems and SambaNova Systems.
NVIDIA remains a major supplier of chips used to train large AI models. Its upcoming Vera Rubin platform is designed to handle demanding computing tasks, including model training and inference.
Oracle and NVIDIA said their expanded agreement would reach “multi-gigawatt scale,” although neither company provided a specific deployment schedule or contract value. NVIDIA CEO Jensen Huang has previously said Vera Rubin systems could generate up to USD 40 billion in revenue per gigawatt for NVIDIA.
Oracle is also reducing its reliance on NVIDIA for some networking components. The cloud provider plans to use its own Acceleron networking chips in certain systems rather than relying entirely on NVIDIA’s networking hardware.
Oracle’s decision to introduce Google TPUs gives customers another option for AI computing. These specialized processors accelerate machine-learning workloads and can be used for both training models and running AI applications.
Google and Oracle said the TPUs would become available through OCI in 2027, with deployment expected to expand over several years. This would make Oracle the first major cloud provider outside Google to offer the company’s TPUs directly through its own infrastructure.
Oracle also plans to offer chips from Cerebras and SambaNova, companies that focus on AI inference. Inference is the stage when a trained AI model generates responses to user requests. Unlike AI training, where NVIDIA has a strong position, the inference market includes a wider range of competing hardware providers.
Also Read: Oracle AI Updates: What’s New in 2026
Oracle has forecast capital spending of around USD 90 billion to USD 95 billion for its 2027 fiscal year. The company has said customers are increasingly funding the chips required for their workloads, though it has not disclosed how it will finance the latest purchases.
By combining NVIDIA systems with Google, Cerebras, and SambaNova hardware, Oracle is seeking to give customers more flexibility in balancing performance, computing capacity, and cost.
The strategy also reflects a broader shift in the AI industry: cloud providers are expanding infrastructure while exploring alternatives to a single dominant chip supplier. For Oracle, the challenge will be meeting growing demand without letting the cost of building AI capacity outpace customer revenue.