October 27, 2026 — Oracle is expanding its artificial intelligence chip supply arrangements with Nvidia while broadening its cloud infrastructure offerings to include processors from Google and Cerebras Systems, as the company seeks to meet growing demand for AI computing.
Oracle and Nvidia announced that their partnership would expand to “multi-gigawatt scale” using Nvidia’s latest-generation Vera Rubin AI servers. The companies did not disclose the financial terms of the agreement.
Nvidia CEO Jensen Huang has previously said Vera Rubin systems could generate up to $40 billion in revenue per gigawatt for Nvidia.
Oracle Diversifies Its AI Chip Supply
Alongside its expanded Nvidia partnership, Oracle plans to introduce Google’s tensor processing unit (TPU) chips to its cloud services for the first time.
The company will also offer chips from SambaNova Systems, which is backed by Intel, and Cerebras Systems. Both companies develop processors designed for AI inference — the stage when AI systems respond to user requests.
While Nvidia dominates the market for training AI models, inference is a more competitive segment with a broader range of suppliers.
Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, said different AI workloads require different computing capabilities. Offering customers more hardware options would allow them to select infrastructure based on performance, throughput and cost.
Oracle Plans to Use Its Own Networking Chips
Oracle is also increasing its reliance on internally developed technology. Even in systems built around Nvidia processors, the company plans to replace some of Nvidia’s networking chips with its own Acceleron networking technology.
The move reflects Oracle’s broader effort to diversify its infrastructure while maintaining flexibility over how its cloud services are built and delivered.
Cerebras Sees Opportunity to Expand Its Customer Base
Cerebras CEO Andrew Feldman described the Oracle agreement as a significant opportunity to expand the company’s reach among major enterprise customers.
He told Reuters that the deal would allow Cerebras technology to be deployed at scale inside other companies’ data centers.
The partnership could help Oracle provide more options for customers running AI inference workloads, while giving Cerebras access to a broader market through Oracle’s cloud infrastructure.
Financing and Spending Plans Remain in Focus
Oracle has not disclosed how it intends to finance its expanded chip purchases. The company is scheduled to hold an investor day this week, putting its infrastructure investment strategy under further scrutiny.
During an earnings call last month, Oracle said customers were increasingly paying for the chips used to support their workloads. The company also maintained its capital spending target of approximately $90 billion to $95 billion for fiscal 2027.
Oracle’s expanded supplier network comes as cloud providers compete to secure computing capacity and offer customers a wider range of AI hardware. By combining a larger Nvidia deployment with Google, SambaNova and Cerebras offerings, Oracle is seeking to accommodate different workloads while expanding its position in the AI cloud market.
Original article
Comments