NVIDIA and Google Cloud Propel AI Innovation with Expanded Developer Resources

Introduction

At the annual Google I/O conference, NVIDIA and Google Cloud announced significant expansions to their collaborative developer program, now supporting over 100,000 participants. This joint initiative provides curated learning paths, hands-on labs, and events designed to help developers build AI applications using the full-stack NVIDIA AI platform on Google Cloud. The community has become a vital resource for data scientists, machine learning engineers, and developers eager to advance their skills with cutting-edge technologies.

Community Growth and Enhanced Offerings

Launched at Google I/O the previous year, the community has rapidly evolved into a central hub for AI builders. Over the past twelve months, members have created production-ready retrieval-augmented generation (RAG) applications on Google Kubernetes Engine (GKE) and implemented advanced observability for agent-based workloads. Additionally, they've explored new large language model (LLM) research and prototyped hybrid inference setups — combining on-premises and cloud resources — for real-world scenarios like sports analytics and enterprise data pipelines.

This year, the community introduces several fresh resources:

• A dedicated learning path for using the JAX library on NVIDIA GPUs.
• A new NVIDIA Dynamo codelab focused on inference optimizations.
• Monthly developer livestreams with expert insights and demos.

New Learning Opportunities

The expanded curriculum equips developers with practical skills to accelerate AI projects. The JAX learning path covers everything from single-GPU experiments to multi-rack deployments, ensuring a consistent experience across scales. Meanwhile, the NVIDIA Dynamo codelab teaches techniques to optimize large-scale inference, especially for mixture-of-experts (MoE) models, enabling more efficient AI serving on NVIDIA-accelerated Google Cloud infrastructure.

Monthly livestreams will feature NVIDIA and Google Cloud engineers demonstrating best practices, troubleshooting common challenges, and showcasing innovative use cases from the community.

Real-World Applications and Tools

Developers are already leveraging these resources for breakthrough applications. Using NVIDIA cuDF in Google Colab Enterprise or Dataproc, they accelerate data science and analytics workflows. For multi-agent systems, combinations of Google DeepMind's Gemma 4 models, NVIDIA Nemotron open models, and the Google Agent Development Kit on Google Cloud G4 VMs (powered by NVIDIA RTX PRO 6000 Blackwell GPUs) enable sophisticated agent deployments on Google Cloud Run or with spot instances.

Collaboration on Open Frameworks

NVIDIA and Google Cloud are deeply invested in open frameworks like JAX. Their joint efforts ensure developers can build, scale, and productize JAX workloads on NVIDIA AI infrastructure within Google Cloud — from single experiments to massive multi-rack clusters — while maintaining strong performance and consistency. This work extends to Google Cloud AI Hypercomputer, where the MaxText framework uses these JAX optimizations to train large models efficiently on NVIDIA GPUs.

Inference Optimization with NVIDIA Dynamo

NVIDIA Dynamo on GKE helps developers optimize large-scale inference, particularly for MoE models. This integration allows AI applications to serve more efficiently on NVIDIA-accelerated Google Cloud infrastructure, reducing latency and costs while improving throughput.

Getting Hands-On with New Resources

Starting next month, community members can access the new learning path on running and scaling JAX on NVIDIA GPUs, as well as the NVIDIA Dynamo on GKE inference codelab. These free resources are available to all members of the Google Cloud and NVIDIA developer community and are designed to accelerate the journey from experimentation to production.

For more information on joining the community and accessing these materials, visit the official NVIDIA and Google Cloud developer hub.