NVIDIA and Google Cloud Partner to Bring Agentic AI to Enterprises

NVIDIA has announced a strategic partnership with Google Cloud to introduce agentic AI to enterprises. This collaboration enables businesses to utilize the Google Gemini family of AI models locally through NVIDIA’s Blackwell HGX and DGX platforms and NVIDIA Confidential Computing for enhanced data security.

Enabling Compliance and Security with NVIDIA Blackwell

The NVIDIA Blackwell platform, integrated with Google Distributed Cloud, empowers on-premises data centers to meet regulatory requirements and data sovereignty laws. Enterprises can ensure compliance while maintaining operational efficiency by restricting access to sensitive information such as patient records, financial transactions, and classified government data. Additionally, NVIDIA Confidential Computing safeguards sensitive code within the Gemini models, protecting it from unauthorized access and potential data breaches.

Sachin Gupta, Vice President and General Manager of Infrastructure and Solutions at Google Cloud, highlighted that the partnership enables enterprises to securely harness the full potential of agentic AI, combining NVIDIA Blackwell’s performance and confidential computing capabilities with operational simplicity.

Dual-Layer Security

Confidential computing with NVIDIA Blackwell provides enterprises with the technical assurance that user prompts to the Gemini models’ application programming interface (API) and the data used for fine-tuning remain secure. This dual-layer security protects against unauthorized access or tampering, allowing enterprises to innovate confidently with Gemini models while preserving data privacy.

The On-Premises Dilemma

Despite the advancements in multimodal reasoning — integrating text, images, code, and other data types to solve complex problems — enterprises with stringent security or data sovereignty requirements have faced challenges in adopting these technologies. This partnership addresses these challenges by making Google Cloud one of the first providers to offer confidential computing capabilities for securing agentic AI workloads in all environments, whether cloud-based or hybrid.

Powered by the NVIDIA HGX B200 platform featuring Blackwell GPUs and NVIDIA Confidential Computing, this solution enables customers to protect their AI models and data while achieving groundbreaking performance and energy efficiency. Enterprises can now maintain data security and model integrity without compromising on innovation.

AI Observability and Security for Agentic AI

Scaling agentic AI in production requires robust observability and security to ensure reliable performance and compliance. To address this, Google Cloud has introduced the GKE Inference Gateway, designed to optimize the deployment of AI inference workloads through advanced routing and scalability. By integrating with NVIDIA Triton Inference Server and NVIDIA NeMo Guardrails, the gateway provides intelligent load balancing that enhances performance and reduces serving costs. It also enables centralized model security and governance.

Looking ahead, Google Cloud plans to enhance observability for agentic AI workloads by integrating NVIDIA Dynamo, an open-source library designed to serve and scale reasoning AI models across AI factories. This forward-looking approach ensures enterprises can confidently scale their agentic AI applications while maintaining security and compliance.