
What happened
Oracle said it will deploy Google TPU AI chips in Oracle Cloud Infrastructure, providing the power, cooling and in-datacenter networking to run them at scale, with services starting in 2027.
Why it matters
The move gives Oracle Cloud customers an AI processor option other than NVIDIA GPUs, and widens where Google TPU can run, Google having open-sourced TorchTPU for native PyTorch support on the chip last week.
What to watch
Google TPU services on Oracle Cloud are scheduled to begin in 2027.
WHO IT HITSEnterprise cloud teams that run AI training and inference workloads on Oracle Cloud Infrastructure may get a second AI chip option alongside NVIDIA GPUs from 2027, and developers using PyTorch may find Google TPU a more accessible target.
Summaries like this, every morning.
Oracle framed the Google TPU deployment as part of its effort to offer AI-ready cloud infrastructure, but the company is chasing AWS, Microsoft Azure and Google Cloud, which lead the cloud market and are each building out their own AI processors alongside NVIDIA GPUs. Bringing Google TPU into Oracle Cloud gives Oracle customers a processor choice beyond NVIDIA.
For Google, the arrangement extends Google TPU beyond its own Google Cloud, where the company has been strengthening the chip to challenge NVIDIA's hold on the AI processor market. Google Cloud last week open-sourced TorchTPU, a version of the PyTorch machine-learning library with native support for Google TPU, which is expected to let users run their own AI workloads on the chip using various frameworks and libraries.
The announcement came during the keynote of Oracle AI World 2026, Oracle's annual event that opened October 27 Japan time in Las Vegas.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Paul Gillin said at EVOLVE26 that Cloudera is positioning itself as an "anywhere cloud for AI," extending clou…
At the Bloomberg Invest conference in Dallas, Crusoe CEO Chase Lochmiller called the market angst over neoclou…

On Latent Space, Anthropic's Katelyn Lesse and Angela Jiang detailed the Claude Platform's architecture, inclu…

Google Deepmind introduced Genie 4, a real-time world model that generates sound directly, including backgroun…

OpenAI released GPT-6.1 Terra, pitched at image and screen understanding, PC operation, and natural writing, a…

Apple's Oct 27 refresh brings the M6 to a $1,499 iMac and $1,999 14-inch MacBook Pro, both shipping Nov 6 with…
