AWS CEO: How Amazon Makes a Comeback in the AI Era? Delivering Cheaper and More Reliable AI at Scale

Amazon Web Services (AWS) is reshaping the cloud computing market landscape by directly deploying AI infrastructure into customer data centers. This new product model, named "AI Factory," enables governments and large enterprises to deploy AI projects at scale while maintaining full control over data processing and storage locations, all while meeting compliance requirements.

On Tuesday, AWS unveiled this product at the Re:Invent 2025 conference in Las Vegas. The AI Factory deploys Nvidia GPUs, Trainium chips, and AWS infrastructure such as networking, storage, and databases into customers' own data centers, operating exclusively for a single customer, functioning like a private AWS region.

This product is developed based on AWS's Project Rainier, created for Anthropic, and has been applied in collaboration with Humain in Saudi Arabia. Last month, AWS and Humain announced an expansion of their partnership, planning to deploy approximately 150,000 AI chips, including Nvidia GB300 and Trainium chips.

This model reflects a strategic shift for cloud service providers in the AI era: by offering flexible deployment options and more cost-effective dedicated infrastructure, they are competing for high-value customers with strict data sovereignty and compliance requirements.

Dual Chip Strategy Meets Different Needs

AWS AI Factory offers customers two technology route options. Customers can choose the Nvidia-AWS AI Factory integrated solution, which provides Nvidia hardware, a full-stack Nvidia AI software suite, and the Nvidia computing platform. The AWS Nitro system, Elastic Fabric Adapter (EFA) trillion-scale network, and Amazon EC2 UltraClusters support Nvidia Grace Blackwell and the next-generation Nvidia Vera Rubin platform.

At the same time, AWS offers a solution based on its self-developed Trainium chips. The company announced the Trainium3 UltraServers at the Re:Invent conference and revealed planning details for the Trainium4 chip. Notably, AWS plans to make future Trainium4 chips compatible with Nvidia NVLink Fusion, enhancing interoperability between the two solutions.

Ian Buck, Vice President and General Manager of Nvidia's hyperscale and HPC business, stated:

Large-scale AI requires a full-stack approach—from advanced GPUs and networking to optimized software and services at every layer of the data center. By combining Nvidia's latest Grace Blackwell and Vera Rubin architectures with AWS's secure, high-performance infrastructure and AI software stack, AWS AI Factory enables organizations to build powerful AI capabilities in a very short time, allowing them to focus entirely on innovation rather than integration

Saudi Project Validates Business Model

The Humain project in Saudi Arabia provides large-scale commercial validation for the AWS AI factory model. Humain CEO Tareq Amin stated, "The AI factory built by AWS in our new AI zone marks the beginning of Humain and AWS's journey of thousands of megawatts. From the outset, this infrastructure was designed to meet the growing local and global AI computing demands."

Tareq Amin emphasized the reasons for choosing AWS: "We chose AWS because of their experience in building infrastructure at scale, enterprise-level reliability, extensive AI capabilities, and deep commitment to the region. Through our shared commitment to global market expansion, we are creating an ecosystem that will shape how AI concepts are built, deployed, and scaled for the entire world."

The project involves the deployment of approximately 150,000 AI chips, including Nvidia GB300 and Trainium chips, showcasing AWS's capabilities in delivering ultra-large-scale AI infrastructure.

Targeting Government and High Compliance Demand Markets

The AI factory products are primarily aimed at government agencies and large organizations with strict data sovereignty and compliance requirements. This dedicated infrastructure model allows customers to run AWS-hosted services, including foundational models, within their own data centers while maintaining control over data processing and storage locations.

This positioning aligns with AWS's recent market moves. According to media reports, AWS recently announced plans to invest $50 billion to expand AI and high-performance computing capabilities for the U.S. government.

Through the AI factory model, AWS combines the flexibility of cloud services with the compliance of on-premises deployment, providing customers with a third option. This "private AWS region" operation allows organizations to leverage AWS's managed services and technological capabilities while meeting regulatory requirements for data localization and sovereignty