Generated by Apex Intelligence. Verify key information against the sources. For information and education, not personalized investment advice. See full disclaimer

Worth notingMarket eventTue, Oct 272 min read

Oracle Cloud to offer Google TPUs, widening AI chip choices

Google gains another route to AI customers; buyers' potential savings depend on pricing and the work needed to adapt their software.

Oracle and Google announced on October 27, 2026, that Oracle Cloud Infrastructure (OCI) will offer Google's Tensor Processing Units, or TPUs—chips designed to train and run AI models. The companies say OCI will be the first hyperscale cloud outside Google to offer them: a provider operating large networks of data centers that rent computing capacity to customers.^[1]

Oracle is also expanding its AI chip relationship with Nvidia while adding offerings from Google and Cerebras.^[2] The supplied TPU announcement gives no specific launch date or prices.^[1]

Offering TPUs through OCI gives Google access to more customers buying AI computing from Oracle, while Oracle gains a wider hardware selection with which to compete for AI business.^[1] Customers can compare the cost of running their work on different chips, but potential savings depend on service prices, whether TPUs suit the task and the effort required to adapt software.^[1] Oracle's parallel expansion with Nvidia leaves chip suppliers competing for customer spending within its cloud; the announcements do not establish how that spending will be divided.^[2]

Impact

Back to the market

Cloud computing servicesOpportunity
Oracle can use a broader AI chip menu to attract customers. Rival cloud providers face a new competing route to Google's specialized chips, although Oracle's ability to win orders depends on prices, available capacity and service performance.^[1]
AI chip suppliersNeutral
Google gains a new channel for supplying TPUs to OCI customers, giving buyers more alternatives to compare. Oracle is also expanding with Nvidia and adding Cerebras, leaving the eventual allocation of orders unresolved.^[1]^[2]
Businesses building and running AIOpportunity
OCI customers will be able to assess TPUs for their AI workloads. More hardware choices may improve cost efficiency, but buyers need to include software adaptation costs in their comparisons; the announcement provides no quantified savings.^[1]