Google will make its Tensor Processing Units (TPUs) available through Oracle Cloud Infrastructure (OCI) in 2027, giving Google another distribution route as it says AI-computing demand exceeds any single cloud provider’s footprint.
The companies announced the plan Oct. 27. Oracle said OCI would be the first hyperscaler beyond Google’s own cloud to offer the accelerators. The companies have not announced the planned capacity, TPU models, prices or deployment locations.
OCI is to supply power, cooling and data center networking for the deployments. The financial terms have not been disclosed.
Alphabet’s Q2 2026 filing says Google recognized revenue during the quarter ended June 30, 2026, from a limited number of TPU-system supply agreements. The contracts cover specialized, high-scale workloads on on-premises infrastructure that customers require or provide. Alphabet said the significant majority of revenue from those agreements is expected in 2027. Google Cloud generated $24.768 billion in the quarter, up 82% from a year earlier, but Alphabet did not report TPU revenue separately.
Google announced a separate joint venture with Blackstone on May 19. Blackstone committed an initial $5 billion in equity. The venture is expected to bring 500 megawatts of capacity online in 2027, with Google supplying TPUs, software and services.
“Bringing Google TPUs to OCI gives customers access to multiple leading AI compute architectures in one cloud,” said Mahesh Thiagarajan, president of Oracle Cloud Infrastructure.
The OCI offering is planned to begin in calendar 2027, with access expected to expand over several years.
