Nvidia Corp. is doubling down on its partnership with Amazon Web Services Inc. to expand what’s possible in the realms of artificial intelligence, robotics and quantum computing development.

The two companies took to the stage during Amazon’s annual customer conference, AWS re:Invent this week, where they made a host of announcements regarding their ongoing collaboration.

The updates include the availability of Nvidia’s NIM microservices on various AWS AI services, which should enable faster inference with lower latency for AI developers, plus the launch of Nvidia’s DGX Cloud on AWS, and various other developments in AI.

For developers, the biggest news is the expanded availability of NIM microservices on AWS.

Nvidia’s NIM provides developers with easy access to a range of easy-to-use microservices that make it easy to deploy high-performance AI model inference workloads in any environment, such as the cloud, on-premises data centers and workstations. With today’s update, they can now be accessed from the AWS Marketplace and the new AWS Bedrock Marketplace, as well as Amazon SageMaker Jumpstart, making it even easier for developers to deploy them from whatever interface they’re working with, the companies said.

What’s more, users will be able to deploy them across multiple AWS services, including Amazon Elastic Compute Cloud, Amazon SageMaker and the Amazon Elastic Kubernetes Service.

The NIM microservices are available as prebuilt containers and they come with a choice of inference engines, including Nvidia Triton Inference Server, Nvidia TensorRT, Nvidia TensorRT-LLM and PyTorch. Moreover, they support hundreds of different AI models, including those available in the AWS Bedrock Marketplace, Nvidia’s own AI foundation models, plus customer’s custom models.

DGX Cloud now on AWS

In addition to the NIM microservices, developers are also getting access to a new infrastructure offering, namely the Nvidia DGX Cloud. It’s now available through AWS Marketplace Private Offers, and gives customers access to a fully managed, high-performance compute platform for training, customizing and deploying AI models.

DGX Cloud is a cloud-hosted AI supercomputing service that gives enterprises access to Nvidia’s graphics processing units and the software they need to train advanced models for generative AI and other types of applications.

One advantage of using the DGX Cloud is its flexible deployment terms, Nvidia said, and customers will also get direct access to the company’s experts, who will be on hand to provide the technical expertise needed to scale their AI deployments.

The DGX Cloud platform currently provides access to Nvidia’s most powerful GPUs, the Nvidia H100 and H200, and will soon be expanded to include the next-generation Blackwell GPUs, slated to launch in the new year.

AWS said the Blackwell chips will be available as part of the GB200 NVL supercomputing system, which will benefit from its new, liquid cooling system to deliver the highest performance with greater energy efficiency than other cloud platforms.

AI blueprints, robotics simulations and drug discovery

In other AI-related announcements, Nvidia said it’s making a number of new AI Blueprints available for instant deployment on AWS. The blueprints provide ready-to-deploy AI agents for tasks such as video search, container vulnerability analysis and text summarization, that can easily be integrated into existing developer workflows.

They enable a number of possibilities, Nvidia said. For instance, developers can use the AI Blueprints for video search to quickly create a visual AI agent that’s able to analyze video in real-time. It can then generate alerts for security teams, or identify health and safety violations in the work place, spot defective products on a manufacturing line and so on.

Nvidia is also making advances in terms of AI-powered robots. The company has long been a believer in the potential of AI to help automate robots so they can perform more useful tasks in the real world, and its latest update aims to accelerate how these use cases can be simulated.

Key to this is the Nvidia Omniverse platform. The company said it’s making a reference application available on Nvidia Omniverse, which is used to create realistic virtual environments and digital twins. The application is said to be powered by high-performance AWS EC2 G6e instances accelerated by its L40S GPUs. According to Nvidia, developers will be able to use it to simulate and test AI-powered robots in any kind of environment, with highly realistic physics.

Meanwhile, Nvidia and AWS are also trying to accelerate AI’s application in the development of new pharmaceuticals. They said Nvidia’s BioNeMo NIM microservices and AI Blueprints for advancing drug discovery are now available with AWS HealthOmics, which is a fully managed biological data compute and storage service designed to support clinical diagnostics.

The collaboration extends the capabilities of AWS HealthOmics, giving researchers the chance to experiment with more AI models, the companies said.

Advancing quantum computing

Finally, Nvidia said it’s working with AWS to help accelerate quantum computing development. The chipmaker’s Nvidia CUDA-Q platform, which is used to develop “hybrid quantum/classical computing applications” that span traditional and quantum computers, is being integrated with the Amazon Braket service.

Amazon Braket makes it easier for users to set up, monitor and execute hybrid quantum-classical algorithms on quantum processors. With the integration, CUDA-Q users will be able to tap into Amazon Braket’s quantum resources, Nvidia said, while Braket users will be able to take advantage of CUDA-Q’s GPU-accelerated workflows for development and simulation.

Holger Mueller of Constellation Research Inc. said no cloud vendor’s conference can be considered complete without an appearance by Nvidia, and the GPU maker definitely didn’t disappoint.

“The two companies are deepening an already deep relationship with more services and offerings coming to the AWS cloud, and that’s good news for enterprises who are customers of both,” the analyst said. “The key announcement for most customers is the imminent arrival of Blackwell, but they’ll be encouraged to see the arrival of additional Nvidia offerings on AWS, from healthcare to blueprints and even quantum computing.”

Image: Nvidia