Re:Invent 2024: How AWS & Nvidia are Powering Ahead With AI
Global chipmaker Nvidia, alongside cloud computing giant Amazon Web Services (AWS), unveiled impressive infrastructure developments aimed at data centre operators deploying AI workloads at AWS Re:Invent 2024.
A significant announcement was Nvidia's Neural Interface Models (NIM) being deployed on AWS. The NIM system enables data centres to process AI models with reduced computational overhead. Such an integration is designed to increase demand for AI inference, the process of running AI models in production environments.
AWS has also introduced liquid-cooled data centres designed for Nvidia's Blackwell platform, which will power Amazon EC2 P6 instances and DGX Cloud. The cooling system combines air and liquid cooling to manage rack-scale AI computing systems including the Nvidia GB200 NVL72.
Expanding AWS infrastructure capabilities
DGX Cloud, available through AWS Marketplace Private Offers, provides organisations with direct access to Nvidia's technical specialists for AI implementation support. It aims to close the gap for companies lacking internal AI expertise.
"Gen AI is transforming cloud workloads and putting accelerated computing at the foundation of diverse content generation," commented Jensen Huang, CEO of Nvidia.
- DGX Cloud provides direct access to Nvidia experts, enabling businesses to scale their AI capabilities quickly
- AWS has introduced liquid-cooled data centres
- Nvidia Isaac Sim allows developers to simulate and test AI-driven robots in virtual environments
- Nvidia BioNeMo NIM microservices accelerate drug discovery processes for biotech companies
- A-Alpha Bio achieved a 12-fold increase in inference speed using the BioNeMo framework on AWS
- Nvidia's latest AI Blueprints for video analysis and cybersecurity are available for instant deployment on AWS
- CUDA-Q allows developers to build hybrid quantum-classical applications using GPU-accelerated workflows
Nvidia's Isaac Sim robotics simulation platform has expanded to AWS, utilizing Amazon EC2 G6e instances powered by Nvidia L40S GPUs. The platform generates synthetic data for training AI models, reducing the need for physical prototyping in robotics development.
Robotics companies Aescape, Cohesive Robotics and Swiss Mile are using Isaac Sim for performance validation of their systems through virtual testing environments.
The integration of Nvidia's CUDA-Q platform with Amazon Braket, AWS's quantum computing service, enables developers to create applications that combine classical computing with quantum processing capabilities.
"All quantum computing is hybrid. We do not subscribe to this theory that quantum computing will displace classical computers," says Stefan Natu, Head of Product for Amazon Braket at AWS. "We think quantum computers are going to operate in tandem with classical computers."
The platform provides tools for quantum simulation and development through GPU acceleration, now pre-installed on Amazon Braket's quantum processors.
Advancing AI data centre infrastructure
The integration of Nvidia's BioNeMo AI Blueprints into AWS HealthOmics is an advancement in the application of AI to healthcare and pharmaceutical research. Here, Nvidia’s AI expertise is combined with AWS’s robust cloud infrastructure to create a powerful drug discovery platform.
The 12-fold increase in inference speed achieved using BioNeMo on AWS infrastructure demonstrates how these technologies can dramatically accelerate research processes in biotechnology.
Likewise, database company DataStax has implemented Nvidia's NeMo Retriever and NIM microservices for Wikimedia Deutschland, creating a vectorised database of Wikidata for developers.
Data management firm Cohesity has also integrated NeMo Retriever microservices into its AI-powered search assistant, Cohesity Gaia, to enhance retrieval-augmented generation - a technique that improves AI model responses by incorporating external data.
Significantly, data analytics platform provider Cloudera is also using Nvidia AI on AWS to enhance its inference solution, as it hopes to support humanitarian organisation Mercy Corps' aid distribution technology.
"Driven by a common mission to deliver cost-effective state-of-the-art Gen AI to every customer, Nvidia and AWS are collaborating across the entire computing stack, spanning AI infrastructure, acceleration libraries, foundation models, to Gen AI services," says Jensen Huang.
"What we really need to advance this industry and move it forward is this ecosystem of AWS, hardware providers, software vendors and ultimately application builders who have potential applications that they want to explore and experiment with quantum computing," Stefan adds.
"That's what's going to move the needle in the industry."
Explore the latest edition of Data Centre Magazine and be part of the conversation at our global conference series, Tech & AI LIVE and Data Centre LIVE.
Discover all our upcoming events and secure your tickets today.
Data Centre Magazine is a BizClik brand
