NVIDIA and Google Cloud Expand Partnership on Agentic and Physical AI, Announce New GPU Instances and Enterprise Solutions
Key Takeaways
- ▸NVIDIA and Google Cloud expanding partnership with new A5X instances capable of scaling to nearly 1M Rubin GPUs for agentic and physical AI workloads
- ▸Enterprise customers achieving significant efficiency gains: Snapchat realized 76% cost savings on A/B testing with GPU-accelerated Apache Spark
- ▸OpenAI running production ChatGPT inference on NVIDIA GB300 and GB200 systems on Google Cloud
Summary
NVIDIA and Google Cloud announced a significant expansion of their partnership focused on agentic and physical AI at Google Cloud Next. The collaboration introduces several major initiatives, including NVIDIA Vera Rubin-powered A5X instances capable of scaling to nearly 1 million Rubin GPUs, and Gemini integration on Google Distributed Cloud. The partnership also encompasses a growing ecosystem of enterprise customers and startups leveraging NVIDIA's AI infrastructure on Google Cloud.
The announcements highlight real-world adoption across multiple sectors. CrowdStrike is utilizing NVIDIA NeMo open libraries with Gemini Enterprise Agent Platform for synthetic data generation and domain-specific cybersecurity fine-tuning. Snapchat achieved 76% daily cost savings on production A/B testing by migrating data pipelines to GPU-accelerated Apache Spark with NVIDIA cuDF. OpenAI is running production inference for ChatGPT on NVIDIA GB300 and GB200 systems, while Schrodinger is dramatically accelerating drug discovery simulations from weeks to hours using NVIDIA accelerated computing.
The partnership also focuses on supporting the startup ecosystem through NVIDIA Inception and Google for Startups programs, with emerging companies like CodeRabbit AI, FactoryAI, and others building autonomous software agents and managed inference solutions powered by Nemotron-based models on Google Cloud.
- Drug discovery acceleration: Schrodinger reducing simulation times from weeks to hours using NVIDIA accelerated computing
- Startup ecosystem growth through NVIDIA Inception and Google for Startups, with focus on autonomous agents and managed inference solutions
Editorial Opinion
This partnership represents a strategic convergence of two major AI infrastructure players at a critical moment when enterprises are moving from experimental AI to production agentic systems. The diversity of use cases demonstrated—from cybersecurity to drug discovery to content creation—underscores how GPU acceleration has become foundational across industries. The combination of NVIDIA's hardware prowess and Google Cloud's scale, coupled with real enterprise traction, positions this alliance as a formidable force in shaping the infrastructure layer of the AI economy.
