Google partners with Oracle to offer TPUs outside its own cloud infrastructure

Google Expands TPU Availability

Google announced that Oracle will offer its Tensor Processing Units (TPUs) to customers. This marks the first instance where Google's custom AI chips, developed since 2013, will be accessible outside of Google's own cloud infrastructure. The decision reflects a shift in Google's strategy to broaden the reach of its AI hardware.

Strategic Shift in Cloud Playbook

Historically, Google focused on selling its chips, models, and cloud services as a unified stack within its own ecosystem. This new partnership with Oracle indicates a change in approach, with Google now pushing its equipment beyond its own walls. This strategy aims to attract more customers and grow its artificial intelligence chip business, a market currently dominated by Nvidia.

Market Impact and Growth Projections

Analysts anticipate significant growth for Google's TPU business. TD Cowen projects $9.4 billion in TPU sales for the second half of this year and $62.5 billion for 2027. Wells Fargo estimates approximately $9 billion in third-quarter sales and $99 billion next year. Google Cloud revenue is expected to increase by about 90% year-over-year, with TPUs contributing to this momentum.

Cloud Market Competition

In the second quarter, Google Cloud generated $25 billion in revenue, trailing Amazon Web Services ($42 billion) and Microsoft Azure ($29 billion). Despite being third in overall cloud revenue, Google Cloud is experiencing the fastest growth among the major providers. The expansion of TPU availability is part of Google's effort to further accelerate its growth and compete in the broader AI hardware market.