At Big Compute 2022, NVIDIA partnered with Rescale to release their HPC-as-a-service platform, bringing full-stack AI software to the cloud. The platform will provide services and new capabilities for simulation and engineering software to be used by cloud service providers. 

NVIDIA’s chief Jensen Huang said that with this collaboration, developers will be able to use its AI tools on the cloud with their supported workflows and AI engines. Rescale will now offer the NVIDIA Modulus framework to build  physics machine learning neural network models to support a wide range of engineering use cases and applications. 

Additionally, Rescale is integrating NVIDIA Base Command Platform, which is an AI developer workflow management system, allowing orchestration of workloads on clouds along with NVIDIA DGX systems. 

Why partner with Rescale? 

NVIDIA said that there has been a rise in companies embracing working on the cloud platforms to build and architect supercomputers like Electrolux, Samsung, Nissan, and Virgin Orbit, and they all have been utilising the Rescale’s cloud platform. This is where NVIDIA decided to step into the picture and collaborate with Rescale to address the rising demand for AI in the cloud. 

Rescale is a high performance cloud computing platform that supports accelerated R&D computing and secure cloud operations. Their AI-enabled performance optimization is what attracts companies to fields of aerospace, energy, automotive, semiconductor, among others.

Other cloud partnerships 

Besides Rescale, NVIDIA also provides HPC-workloads and NVIDIA Quadro Virtual Workstation on Google Cloud to accelerate graphic rendering and intensive workloads anywhere. NVIDIA has partnered with AWS for more than a decade to provide GPU support and AI services on the cloud platform.

NVIDIA has also been working with Microsoft Azure to provide GPU-acceleration on cloud, and are expected to make announcements about their recent collaborations at the joint event—Supercomputing Conference 2022.

Catches with up trends 

NVIDIA’s shift to cloud systems is clearly visible since GTC 2022, when they announced that they are moving to offer SaaS—Nvidia Omniverse Cloud—a suite of cloud services for developers, enterprises, and artists. NVIDIA has been developing various models and taking big leaps to make a more realistic Metaverse.

Another trend that NVIDIA recently picked up on was text-to-image synthesis using diffusion models, and built their model called ‘eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers’. This comes after years of dedication to GAN for image generation.