Google Cloud has expanded its collaboration with NVIDIA by introducing a new line of virtual machines tailored for advanced AI applications in enterprise environments. Powered by NVIDIA's RTX Pro 6000 Blackwell server GPUs, the G4 virtual machines represent a significant technological advancement, according to Google. In addition, Google has made NVIDIA Omniverse available as a virtual machine image on Google Cloud Marketplace, along with the reference robotics simulation framework, NVIDIA Isaac Sim. In a blog post, Google Cloud's Roy Kim and Dai Vu stated, "G4 VMs deliver a profound leap in performance, offering nine times the throughput of G2 instances. This brings about a step-change improvement in results across various workloads, including multimodal AI inference, realistic design and visualization, and robotics simulation using applications developed on NVIDIA Omniverse." Initially, G4 VMs will be available with configurations of one, two, four, and eight RTX Pro Blackwell GPUs—providing up to 768 GB of GDDR7 memory. Fractional GPU options are also expected to become available. According to NVIDIA, the design integrates fifth-generation Tensor Cores for AI performance boosts and fourth-generation RT Cores for cinematic-quality graphics and realistic simulations. Google Cloud noted that customers will benefit across multiple domains, particularly in generative AI, where G4 accelerates fine-tuning and inference for large language models, enabling real-time applications such as multimodal and text-to-image models. There are also potential efficiency gains. Since G4 allows a single GPU to be divided into four independent instances, each with its own high-bandwidth memory and compute cores, multiple smaller and varied workloads can be run concurrently on a single VM. Additionally, the solution offers flexibility and scalability, supporting AI models ranging from less than 30 billion to over 100 billion parameters. Kim and Vu further explained how the availability of NVIDIA Omniverse on Google Cloud Marketplace would benefit users. "G4 VMs provide the essential infrastructure for demanding real-time rendering and physics-accurate simulations required by enterprise digital twins. Together with Omniverse, they offer a scalable cloud environment for developing, deploying, and interacting with applications focused on industrial digital twins or robotics simulations." For NVIDIA, the expanded partnership builds an end-to-end platform on Blackwell, delivering a unified architecture that provides a seamless experience for accelerating workloads within a single cloud ecosystem.