Google, NVIDIA Extend AI Partnership With Massive Infrastructure Expansion
Techstrong.ai, Thursday, April 23rd, 2026
Google Cloud and NVIDIA expanded their partnership with new AI Hypercomputer infrastructure featuring A5X instances and Vera Rubin architecture.
Google Cloud and NVIDIA announced an extended partnership centered on AI Hypercomputer, a co-engineered stack designed for industrial and agentic AI applications. The announcement featured the debut of A5X bare-metal instances powered by NVIDIA's Vera Rubin architecture, delivering 10x reductions in inference costs and 10x increases in throughput per megawatt.
Google Cloud can now scale clusters to 80,000 GPUs at a single site using next-generation Virgo networking, with infrastructure specifically designed for physical AI, robotics, and industrial digital twins.
The partnership includes the first Confidential Computing offering for NVIDIA Blackwell GPUs and integrates NVIDIA's Nemotron models into Gemini Enterprise Agent Platform for automating complex workflows. Companies like OpenAI, Snap, Schrodinger, and CrowdStrike are already leveraging the platform for demanding inference tasks, cost reduction, and accelerated simulations.