Announced 27 Oct 2026 · Sources checked
What Oracle and Google announced
Oracle and Google announced at Oracle AI World on 27 October 2026 that Google's Tensor Processing Units will be available through Oracle Cloud Infrastructure. OCI will be the first hyperscale cloud platform outside Google to offer TPUs.
Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, said: 'Different AI workloads have different compute needs. Bringing Google TPUs to OCI gives customers access to multiple leading AI compute architectures in one cloud.'
Why it matters
Google TPUs are custom accelerators co-designed with open software to power the entire AI lifecycle, from frontier training and large-scale inference to the multi-step reasoning demands of agentic AI. OCI will provide the power, cooling and datacenter networking to deploy them at scale.
The deal is significant because Google has historically kept TPUs as an internal advantage powering its own services and Google Cloud. Offering them through a competitor's cloud infrastructure is a major step in Google's broader effort to sell TPUs as a commercial product, and it gives OCI a differentiated AI compute option in a competitive market.
The broader chip expansion
The TPU deal is part of a wider Oracle chip expansion. Oracle and NVIDIA said the cloud firm will expand to multi-gigawatt scale with NVIDIA's Vera Rubin AI servers, and Oracle plans to offer chips from Cerebras Systems and SambaNova Systems, both inference specialists.
Thiagarajan's statement — 'Different AI workloads have different compute needs' — signals Oracle is deliberately building a multi-vendor AI compute portfolio rather than tying OCI to a single accelerator supplier.
What to watch
Watch when TPUs actually go live on OCI and what the pricing looks like relative to Google Cloud's own TPU offerings.
Watch whether this opens the door for other hyperscalers to offer Google TPUs, and how NVIDIA responds to its largest cloud partners increasingly offering competing accelerators.
Watch Oracle's investor day on 28 October, where the company is expected to provide more detail on its AI infrastructure strategy, including how the TPU and multi-chip offerings fit into its fiscal 2027 capital spending plans of about $90 billion to $95 billion.
The AI hardware market context
The TPU arrangement lands as AI companies diversify their chip supply. OpenAI is reportedly planning to buy Google TPUs, Anthropic uses multiple chip families including TPUs for training, and Oracle's own announcement expands its AI compute menu with NVIDIA Vera Rubin, Cerebras and SambaNova alongside the new Google TPU option.
For Google, letting a competitor's cloud offer TPUs is a bet that selling the hardware broadly generates more revenue than keeping it exclusive to Google Cloud. For Oracle, the deal is a way to differentiate OCI in a crowded hyperscaler market by giving customers access to a second major accelerator ecosystem.
What the executives said
Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, said bringing Google TPUs to OCI gives customers choice across performance, throughput and economics per workload. Google's chief technologist for AI and infrastructure, Amin Vahdat, said the next generation of AI will require a diverse range of compute, and that TPUs give customers a power-efficient way to run the full AI lifecycle.
The quotes frame the deal as a customer-choice story rather than a partnership between equals, which is notable given that Oracle and Google are normally competitors in cloud and that Google is investing heavily in its own AI infrastructure.
What 'first for a hyperscaler' means
The 'first hyperscaler outside Google' framing needs context. Google has sold TPUs as a Google Cloud service for years, and this year began selling TPU hardware directly to some customers. Oracle is the first major cloud provider outside Google to put TPUs in its portfolio, which gives OCI an accelerator option that AWS, Azure and most other clouds do not offer as a standard service.
That differentiation matters as cloud providers compete for AI workloads. Enterprises that want to hedge their accelerator dependence, or that have engineers already familiar with TPU tooling, may pick OCI specifically for TPU access. NVIDIA's response to its largest cloud partners carrying competing silicon will be one of the more interesting business developments to follow.
The OpenAI connection
The Oracle-Google TPU arrangement also sits alongside OpenAI's reported plan to buy Google TPUs, which would make OpenAI both a Google Cloud customer and a buyer of Google silicon. AI Lookout covered that report on 27 October, and together the two stories signal a broader shift: the largest AI labs and cloud providers are no longer tying themselves to a single accelerator vendor.
Google's willingness to sell TPUs to both OCI and reportedly OpenAI suggests the company sees bigger value in selling silicon widely than in keeping it exclusive to Google Cloud. For AI infrastructure buyers, that means more choices and, potentially, more competitive pricing across the market.
Common questions
Is OCI the first outside Google to offer TPUs?
Yes. Oracle says OCI will be the first hyperscale cloud platform outside Google to offer Google TPUs.
When will Google TPUs be available on OCI?
The companies announced the plan on 27 October 2026; a specific availability date was not stated in the announcement.
Sources & further reading
How this story was made
Written by Kristian Kostov with AI assistance and checked against the linked sources. Company performance claims are attributed to the company. Analysis reflects AiLookout’s interpretation; we have not independently tested the products discussed. The cover is an AI-generated editorial illustration, not a photograph or evidence of an actual event.
Our editorial standards
