NVIDIA & Google: Blackwell & Gemini Partnership Update
- NVIDIA and Google are expanding their long-term partnership to boost AI innovation and support developers.
- NVIDIA AI software, such as Nvidia Nemo and NVIDIA tensorrt-LLM, is integrated across Google Cloud, including Vertex AI and Google Kubernetes Engine (GKE), to accelerate performance and simplify...
- The A4 VMs, accelerated by NVIDIA HGX B200, are now generally available.
NVIDIA and Google Cloud are deepening their AI collaboration, a move that promises significant advancements in artificial intelligence.This partnership integrates NVIDIA Blackwell platforms, supercharging the performance of Google’s Gemini models and Gemma family. This powerful combination is designed to optimize AI inference, offering developers enhanced capabilities across various deployment architectures. The integration of NVIDIA AI software across Google Cloud, including Vertex AI and GKE, streamlines AI deployments, providing users with a potent and efficient AI experience. News Directory 3 offers crucial updates in this evolving tech landscape and highlights how Google Cloud was the first to offer NVIDIA HGX B200. With Blackwell’s robust performance and confidential computing, user data is protected, enabling innovation while adhering to privacy standards. Discover what’s next as this powerhouse partnership continues to evolve and deliver transformative AI solutions.
NVIDIA and Google Cloud Expand AI Collaboration with Blackwell
Updated May 28, 2025
NVIDIA and Google are expanding their long-term partnership to boost AI innovation and support developers. The collaboration focuses on optimizing the computing stack and community software, including JAX, OpenXLA, MaxText, and llm-d. These efforts directly enhance the performance of Google’s Gemini models and Gemma family.
NVIDIA AI software, such as Nvidia Nemo and NVIDIA tensorrt-LLM, is integrated across Google Cloud, including Vertex AI and Google Kubernetes Engine (GKE), to accelerate performance and simplify AI deployments. Google Cloud was the first to offer NVIDIA HGX B200 and NVIDIA GB200 NVL72 via its A4 and A4X virtual machines (VMs).
The A4 VMs, accelerated by NVIDIA HGX B200, are now generally available. Google Cloud’s A4X VMs deliver notable computing power and support scaling, enabled by Google’s Jupiter network and NVIDIA ConnectX-7 NICs. Google’s cooling infrastructure ensures efficient performance for large AI workloads.
The Gemini models’ reasoning capabilities are already powering cloud-based AI applications. Now, with NVIDIA Blackwell platforms on Google Distributed Cloud, organizations can deploy Gemini models securely within their own data centers.This is notably beneficial for sectors with strict data and regulatory requirements.
NVIDIA Blackwell’s performance and confidential computing capabilities ensure user data protection. This allows customers to innovate with Gemini while maintaining control over their information and meeting privacy standards. The collaboration optimizes AI inference performance for Google Gemini and Gemma.
The Gemma family has been optimized for inference using the NVIDIA TensorRT-LLM library and are expected to be offered as NVIDIA microservices. These optimizations maximize performance and accessibility for developers across various deployment architectures.
NVIDIA and Google Cloud are also supporting the developer community by optimizing open-source frameworks like JAX for scaling on Blackwell GPUs. A new joint developer community will accelerate cross-skilling and innovation.
What’s next
The collaboration aims to make it easier for developers to build, scale, and deploy next-generation AI applications, leveraging engineering excellence and open-source leadership.