Google Cloud and NVIDIA announced an expanded partnership today to provide new AI infrastructure and software to help customers build and deploy large language models and speed up data science workloads.
The announcement came at the Google Cloud Next conference where Google Cloud CEO Thomas Kurian and NVIDIA CEO Jensen Huang shared the stage to unveil new A3 instances powered by NVIDIA's latest H100 Tensor Core GPUs, purpose-built to accelerate generative AI models. They highlighted how the collaboration will bring advanced machine learning capabilities to customers worldwide.
The newly released A3 instances are designed to meet the burgeoning compute requirements. The NVIDIA H100 GPUs are optimized for Transformer Engine technology, giving an unprecedented performance boost to LLMs and other AI applications. Huang and Kurian emphasized the integral role played by NVIDIA's H100 and A100 GPUs in Google’s internal research divisions like DeepMind.
A key part of the expanded partnership is optimizing Google's PaxML framework for NVIDIA GPUs. PaxML enables the development of massive AI models and has been used by Google's own teams, including DeepMind. Originally engineered to work with Google's TPU slices, PaxML now allows developers to use NVIDIA's H100 and A100 Tensor Core GPUs. A GPU-optimized PaxML container is now available on NVIDIA's NGC container registry, offering immediate accessibility to a broad array of developers.
NVIDIA H100 GPUs will also be available in the coming weeks on Google's Vertex AI platform for faster training of large language models. And Google Cloud will be one of the first providers to offer access to NVIDIA's new DGX GH200 AI supercomputer for exploring generative AI capabilities.
The partnership builds on a long history of Google and NVIDIA working together, from L4 GPUs on Google Cloud to AI Enterprise software on the Google Marketplace. The latest integrations span hardware, software and services to advance artificial intelligence.
"We're at an inflection point where accelerated computing and generative AI have come together to enable new innovations," said Jensen Huang. "Our collaboration with Google Cloud will help developers accelerate their work through optimized infrastructure, software and services."
The partnership aims to make robust AI solutions easily accessible for customers. As Thomas Kurian noted, "Many of Google's products are built on NVIDIA GPUs, and customers are seeking NVIDIA acceleration for efficient development of large language models."
With powerful new tools for generative AI now available through Google Cloud and NVIDIA, businesses can tap into transformative AI to drive solutions and gain strategic advantages. The companies are committed to pushing the boundaries of what's possible with artificial intelligence.