Tech giant Google announced that Nvidia’s latest AI powerhouse, the Blackwell platform, will be available on Google Cloud in early 2025. This collaboration expands the capabilities of Google Cloud, offering customers access to cutting-edge AI processing.
Blackwell arrives in two flavors: the HGX B200 and the GB200 NVL72. The HGX B200 tackles the most demanding workloads in AI, data analytics, and high-performance computing. Whereas the GB200 NVL72 is specifically designed for the next frontier of large-scale model training and real-time inference, particularly suited for trillion-parameter models.
This integration comes alongside Google Cloud’s announcement of their most powerful TPU accelerator yet, the TPU v5p. With four times the processing power per pod compared to the previous generation, the v5p caters to training and running AI models. Additionally, Google Cloud announced expanded support for Nvidia’s existing H100 GPUs through their DGX Cloud platform.
The combined arrival of Nvidia’s Blackwell platform and Google Cloud’s advancements in TPU technology signifies a significant boost for cloud-based AI capabilities. This collaboration empowers businesses and researchers with powerful tools to tackle demanding tasks in areas like scientific discovery, drug development, and large language model development.
