Big Red gets cozy with Google to offer TPUs to AI customers
Oracle is expanding Oracle Cloud Infrastructure’s AI offerings by adding Google’s Tensor Processing Units, giving customers an alternative to Nvidia hardware. The move broadens Oracle’s chip options as it seeks to serve different AI workloads and build on its growing cloud business.
Google TPUs are due to become available on OCI in 2027, with capacity increasing over several years; Oracle did not disclose how many it would provide. The company also announced new ChatGPT integrations and OpenAI-powered products for business tasks and cybersecurity, a Snowflake service on OCI, and the availability of Nvidia Vera Rubin NVL72 systems and Vera CPUs. Oracle reported revenue growth of 30 per cent year on year for the quarter ending 31 August.
In short
- Oracle plans to offer Google TPUs on OCI from 2027.
- The move gives customers another option alongside Nvidia chips.
- Oracle also announced OpenAI and Snowflake cloud services.
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Offering Google TPUs alongside Nvidia hardware could give Oracle customers more choice, help match chips to different AI workloads and reduce dependence on a single supplier. The partnership may also strengthen OCI’s appeal as demand for computing grows, while adding competition that could encourage better prices and innovation.
The case against
Adding another chip option may bring little immediate benefit: Google TPUs are not due on OCI until 2027, and Oracle has not disclosed how much capacity it will offer. Customers could face added complexity in choosing and adapting workloads across platforms, while the announcement’s broad mix of partnerships and products may make it harder to judge how much the deal will contribute to Oracle’s business.