NVIDIA and Google Cloud Expand AI Infrastructure for Agentic and Physical AI

A Decade of Co-Engineering AI at Scale

For more than ten years, NVIDIA and Google Cloud have jointly built a full-stack AI platform that spans every technology layer—from performance-optimized libraries and frameworks to enterprise-grade cloud services. This deep collaboration has created a foundation that allows developers, startups, and enterprises to move agentic and physical AI from research into production. Applications now range from intelligent agents that manage complex workflows to robots and digital twins operating on factory floors.

At this week's Google Cloud Next event in Las Vegas, the partnership reached a new milestone with advancements that expand Google Cloud AI Hypercomputer for AI factories. These innovations are designed to power the next frontier of agentic and physical AI. Key announcements include new NVIDIA Vera Rubin-powered A5X bare-metal instances, a preview of Google Gemini on Google Distributed Cloud running on NVIDIA Blackwell and Blackwell Ultra GPUs, confidential VMs with NVIDIA Blackwell GPUs, and agentic AI on the Gemini Enterprise Agent Platform using NVIDIA Nemotron open models and the NVIDIA NeMo framework.

Next-Generation Infrastructure: From Blackwell to Vera Rubin

Google Cloud announced the A5X instance, powered by NVIDIA Vera Rubin NVL72 rack-scale systems. Through extreme co-design across chips, systems, and software, these systems deliver up to 10x lower inference cost per token and 10x higher token throughput per megawatt compared to the previous generation. The A5X uses NVIDIA ConnectX-9 SuperNICs combined with next-generation Google Virgo networking, scaling to up to 80,000 NVIDIA Rubin GPUs within a single site cluster and up to 960,000 NVIDIA Rubin GPUs in a multisite cluster. This enables customers to run their largest AI workloads on NVIDIA-optimized infrastructure.

“At Google Cloud, we believe the next decade of AI will be shaped by customers’ ability to run their most demanding workloads on a truly integrated, AI‑optimized infrastructure stack,” said Mark Lohmeyer, vice president and general manager of AI and computing infrastructure at Google Cloud. “By combining Google Cloud’s scalable infrastructure and managed AI services with NVIDIA’s industry‑leading platforms, systems and software, we’re giving customers flexibility to train, tune and serve everything from frontier and open models to agentic and physical AI workloads—while optimizing for performance, cost and sustainability.”

Comprehensive Blackwell Portfolio for Every Workload

Google Cloud's broad NVIDIA Blackwell portfolio includes a range of virtual machines:

• A4 VMs with NVIDIA HGX B200 systems
• A4X VMs with NVIDIA GB200 NVL72 rack-scale systems
• A4X Max with NVIDIA GB300 NVL72 systems
• Fractional G4 VMs with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs

Customers can right-size their acceleration capabilities, whether using multiple interconnected NVL72 racks that scale out to tens of thousands of NVIDIA Blackwell GPUs, a single rack that scales up to 72 Blackwell GPUs with fifth-generation NVIDIA NVLink and NVLink 5 Switch, or as little as one-eighth of a GPU. This comprehensive platform gives organizations the flexibility to match infrastructure to workload demands.

Enhanced Security with Confidential VMs

Security remains a top priority for AI workloads. Google Cloud introduced confidential VMs with NVIDIA Blackwell GPUs, providing hardware-based memory encryption and isolation for sensitive AI training and inference. This allows enterprises to protect proprietary models and data while taking advantage of NVIDIA's latest GPU technology.

Advancing Agentic AI with Gemini and Nemotron

On the software side, the partnership brings agentic AI to the Gemini Enterprise Agent Platform. Customers can now leverage NVIDIA Nemotron open models combined with the NVIDIA NeMo framework to build, customize, and deploy intelligent agents that reason, plan, and act. This integration enables agents to manage complex business processes, interact with users naturally, and operate securely within enterprise environments.

Together, these advancements—from infrastructure to software—provide a complete ecosystem for organizations to build and scale agentic and physical AI applications. Whether deploying a single agent on a fractional GPU or training frontier models on a massive cluster of Rubin GPUs, the NVIDIA-Google Cloud partnership continues to deliver the tools needed for the next decade of AI innovation.

For more details, explore the comprehensive Blackwell portfolio or learn about the next-generation infrastructure.