Technology giants Nvidia and Google Cloud are teaming up to bring Agentic AI to companies that need stronger security and more control over their data. The collaboration allows businesses to run Google’s Gemini AI models on-premises using Nvidia’s new Blackwell HGX and DGX platforms, along with Nvidia Confidential Computing.
The goal is to support businesses that must meet strict data privacy and sovereignty rules. With Nvidia Blackwell on Google Distributed Cloud, sensitive information — such as medical records or financial data — stays protected under local regulations.
“By bringing our Gemini models on premises with Nvidia Blackwell’s breakthrough performance and confidential computing capabilities, we’re enabling enterprises to unlock the full potential of Agentic AI,” said Sachin Gupta, vice president and general manager of infrastructure and solutions at Google Cloud. “This collaboration helps ensure customers can innovate securely without compromising on performance or operational ease.”
Adding another layer of protection
Confidential computing adds a layer of protection by keeping user prompts and fine-tuned data private, while also preventing unauthorized access to the models.
Agentic AI goes beyond typical AI functions. Instead of only answering questions or retrieving data, it can analyze, make decisions, and take action. For example, in IT support, agentic AI can diagnose and fix issues. In finance, it can block suspicious transactions before they happen.
Google Cloud is also rolling out GKE Inference Gateway to help manage AI workloads more efficiently. It works with Nvidia’s Triton Inference Server and NeMo Guardrails to improve performance, reduce serving costs, and support security controls.
“Securing both data and models is key to building trust in AI,” said Ian Buck, vice president of hyperscale and HPC at Nvidia.
The solution supports both cloud and hybrid setups, giving enterprises more options for using AI while protecting their data.