• Google Cloud (GOOG) becomes the first cloud provider to offer NVIDIA (NVDA) RTX PRO 6000 Blackwell Server Edition GPUs via G4 virtual machines, integrated with Cloud Run for serverless high-end AI inference.
  • The move targets surging demand for inference workloads, simulations, and visual tasks, with performance gains of up to 4x compute and 6x bandwidth over previous G2 VMs.
  • Industry experts highlight potential for accelerating enterprise AI adoption in robotics, digital twins, and content creation, amid a competitive cloud GPU market growing over 50% annually.

Google Cloud has launched a preview of G4 virtual machines powered by NVIDIA’s RTX PRO 6000 Blackwell Server Edition GPUs, marking it as the first cloud provider to offer this hardware, with general availability expected soon. The VMs integrate with Cloud Run to deliver a serverless experience for high-end AI inference, simulations, and visual workloads, according to company announcements and industry sources.

Efforts to expand AI infrastructure have hit a stride, with Google Cloud leveraging its partnership with NVIDIA to address growing enterprise needs. "This integration allows developers to access pro-grade GPUs on a pay-per-use basis, reducing barriers for smaller firms," said a person familiar with the matter, who spoke on condition of anonymity. The move comes as Alphabet (GOOGL) reported Q3 2025 revenue of $88.3 billion, with Google Cloud contributing $11.4 billion, up 35% year-over-year, driven by AI demand amid global GPU shortages.

Without such innovations, companies might struggle with costly on-prem setups, especially as energy costs rise. The RTX PRO 6000 GPUs, optimized for inference via FP4/FP6 precision and RT Cores, offer significant throughput gains—experts predict up to 168% improvements via custom interconnects. This positions Google Cloud in the agentic and physical AI era, with applications in robotics and Omniverse simulations. NVIDIA Omniverse and Isaac Sim are now generally available as VMIs on Google Cloud Marketplace, enhancing digital twin capabilities for industries like manufacturing and automotive.

Parallel developments see AWS (AMZN) and Azure (MSFT) rolling out Blackwell equivalents, but Google Cloud’s serverless approach with Cloud Run could give it an edge. "It’s a strategic play to capture market share in inference workloads, where demand is exploding," noted an industry analyst. The cloud GPU market is projected to grow over 50% annually through 2028, reflecting broader trends in multimodal AI and physical simulations.

In related news, CoreWeave (CRWV) also offers RTX PRO 6000 instances, expanding enterprise data center transformations. Google Cloud plans to introduce smaller G4 VM variants and full Cloud Run serverless support in the short term, complementing its A-series for hybrid workloads. Attempts to reach Google Cloud for additional comments were not immediately successful, but sources indicate positive industry buzz on performance for agentic AI.

Correction: An earlier version misstated the GPU model; it is the RTX PRO 6000 Blackwell Server Edition, not a consumer variant.