By integrating the NVIDIA Blackwell platform with Google Distributed Cloud, on-premises data centres can remain aligned with regulatory conditions and data sovereignty laws. Additionally, NVIDIA Confidential Computing secures code in the Gemini models, protecting it from unauthorised access and data leaks. As a result, businesses can lock down access to sensitive data, including patient records, financial transactions, and confidential government documents.
NVIDIA Blackwell enables confidential computing, assuring enterprises that user prompts sent to the Gemini models' application programming interface and the data used for the fine-tuning are secure from viewing or modification. Moreover, model owners can safeguard them against unauthorised access or tampering, providing dual-layer protection.
Agentic AI aims to transform enterprise technology by offering advanced problem-solving capabilities. Unlike traditional AI models that rely on learned knowledge, the new AI systems are capable of reasoning, adapting, and making decisions in dynamic backgrounds. These capabilities are particularly valuable in the IT field, where a knowledge-based AI model can diagnose issues, solve complex problems, and detect potentially fraudulent transactions.
As many models use features such as integrating text, image, code, and other data types, problems like security and data sovereignty requirements are not covered.
In this announcement, Google Cloud, utilising Blackwell GPUs and NVIDIA Confidential Computing, is one of the first cloud service providers to offer confidential computing capabilities that protect agentic AI workloads across various environments, including both cloud and hybrid. Moreover, the NVIDIA HGX platform with Blackwell GPUs and NVIDIA Confidential Computing secures AI models and data, delivering performance and energy efficiency without compromising security or model integrity.
Following this announcement, Google Cloud stated a new GKE Inference Gateway designed to optimise the AI inference workloads through improved routing and scalability. By integrating with NVIDIA Triton Inference Server and NVIDIA NeMo Guardrails, it provides intelligent load balancing, ensuring performance and low serving costs while facilitating centralised model security and governance.
Looking forward, Google Cloud aims to improve observability for agentic AI workloads by incorporating NVIDIA Dynamo, an open-source library created for serving and scaling reasoning AI models across AI factories.
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now