AWS AI Factory: Enhancing AI Infrastructure Performance in Data Centers

Build dedicated AI infrastructure right in the data center.

As governments and large organizations seek to expand artificial intelligence (AI) projects, many have begun to move toward the concept of "AI factories" to meet specific needs regarding data sovereignty and regulatory compliance.

However, building a high-performance AI factory requires a comprehensive system encompassing management processes, databases, storage, and security, with a complexity that few customers are willing to shoulder themselves.

Organizations in heavily regulated sectors and the public sector face significant AI infrastructure challenges when deploying large-scale projects. Building their own AI capabilities requires massive investments in GPUs, data centers, and energy, as well as navigating complex procurement processes, selecting the right model for each use case, and licensing multiple AI vendors.

All of these factors cause the implementation process to drag on for many years, increase operational complexity, and distract organizations from their core business objectives.

To address this challenge, on December 11th, AWS launched AWS AI Factories, a new service that provides businesses and governments with a dedicated AI infrastructure, right in the customer's data center, operated exclusively for them.

AWS AI Factories integrates advanced AI acceleration tools, from cutting-edge NVIDIA AI computing technology to Trainium chips, combined with high-speed, low-latency networking capabilities, high-performance storage and databases, comprehensive security, energy-efficient infrastructure, and powerful AWS AI services such as Amazon Bedrock and SageMaker AI.

This enables customers to quickly develop and deploy AI applications at scale, optimizing performance and meeting specific business needs.

AWS AI Factories operate as separate AWS Regions, providing secure, low-latency access to computing, storage, databases, and AI services. This approach leverages available data center space and power while opening access to the full AWS AI infrastructure and services, from cutting-edge AI chips for training and inference to tools for building, training, and deploying AI models.

Furthermore, managed services provide access to leading foundation models without the need for separate contracts with individual providers, while helping customers fully meet requirements for security, data sovereignty, and regulatory compliance regarding where data is processed and stored.

With nearly two decades of leadership in cloud computing and superior experience in designing large-scale AI systems, AWS can deploy secure, reliable AI infrastructure far faster than most organizations can build it themselves, saving years of effort and significantly reducing operational complexity.

The 15-year partnership between AWS and NVIDIA

The relationship between AWS and NVIDIA began 15 years ago, when AWS launched the world's first GPU cloud. Since then, AWS has provided customers with the most diverse portfolio of GPU solutions.

Building on a long-standing partnership to deliver advanced AI infrastructure, AWS and NVIDIA help customers build and run large-scale language models faster, more efficiently, and more securely than anywhere else, right within their own data centers.

With the NVIDIA-AWS AI Factories integration, AWS customers gain seamless access to NVIDIA's accelerated computing platform, a comprehensive AI software suite, and thousands of GPU-accelerated applications, delivering superior performance, high efficiency, and flexible scalability for developing next-generation AI solutions.

The two sides continue to combine cutting-edge technologies: AWS Nitro System, petabit-scale Elastic Fabric Adapter (EFA) networking, and Amazon EC2 UltraClusters supporting the latest NVIDIA Grace Blackwell and next-generation NVIDIA Vera Rubin platforms.

In the future, AWS will add support for NVIDIA NVLink Fusion high-speed chip connectivity technology to next-generation Trainium4 and Graviton chips, as well as in Nitro Systems. This integration helps customers shorten time to market while achieving superior performance.

Ian Buck, Vice President and General Manager of Hyperscale and HPC Computing at NVIDIA, shared: “Large-scale AI requires a holistic approach from advanced GPUs and networking infrastructure to software and services that optimize every layer of the data center. Together with AWS, we bring all of these elements directly into our customers’ environments.”

"By combining NVIDIA's latest Grace Blackwell and Vera Rubin architectures with AWS's secure, high-performance infrastructure and comprehensive AI software suite, AWS AI Factories helps organizations deploy powerful AI capabilities in a fraction of the time, while focusing entirely on innovation instead of worrying about integration," said Ian Buck.

Supporting the public sector to accelerate AI adoption.

AWS AI Factories are designed to meet AWS's most stringent security standards, giving government customers the utmost confidence in running sensitive workloads at all classification levels: Unclassified, Sensitive, Classified, and Top Secret.

At the same time, AWS AI Factories provides governments worldwide with the readiness, reliability, security, and control needed to drive economic growth and maximize the benefits of AI technology.

AWS and NVIDIA are strategically partnering with HUMAIN, a global company based in Saudi Arabia specializing in developing comprehensive AI capabilities.

As part of this collaboration, AWS will build the first pioneering “AI Zone” in Saudi Arabia, with a scale of up to 150,000 AI chips, including GB300 GPUs, AWS's dedicated AI infrastructure, and AWS AI services, all deployed in a custom-designed data center by HUMAIN.

Tareq Amin, CEO of HUMAIN, shared: “The AI factory that AWS is building in our new AI Zone marks the beginning of a multi-gigawatt journey for HUMAIN and AWS. From the outset, this infrastructure has been designed to serve both the accelerating AI computing needs locally and globally.”

"Through our shared commitment to expanding global markets, we are creating an ecosystem that will shape the future of how AI ideas are built, deployed, and scaled worldwide," added Tareq Amin.

NGUYEN BAO

Source: https://nhandan.vn/nha-may-tri-tue-nhan-tao-aws-nang-cao-hieu-nang-ha-tang-ai-trong-trung-tam-du-lieu-post929637.html