AWS Local Zones

Extend AWS services to 30+ metros worldwide so you can run latency-sensitive applications closer to end users, address data residency requirements for regulated workloads, and accelerate migration and modernization of legacy applications to the cloud.

Deploy AI/ML workloads with petabit-scale GPU clusters for large-scale training and inference in select metros (Atlanta, Dallas, Phoenix), and run distributed inference across Local Zones in 30+ metros—all without building data centers, negotiating colocation contracts, or managing dedicated infrastructure

Use familiar AWS infrastructure, services, APIs, and tools with full access to regional services through the AWS backbone. Built on the same AWS Nitro System, Local Zones inherit security controls from AWS Regions. Architect for high availability using familiar AWS resiliency strategies.

Scale your applications as needed with on-demand scaling, without overprovisioning, and only pay for what you use. Choose from multiple instance types, pricing models (On-Demand, Savings Plans, Spot), and deployment options to match your workload requirements.

Deliver high-quality experiences for interactive applications that require low latency to your users. Run applications close to end users in AWS Local Zones to enable real-time gaming, live streaming, augmented and virtual reality (AR/VR), virtual workstations, independent software vendor (ISV) application acceleration, and more.

Address data residency requirements when data needs to remain in a particular country, state, or municipality for regulatory, contractual, or information security reasons. This is often the case with betting & gaming, financial services, healthcare, and other highly regulated industries. With AWS Local Zones, you can control where your workloads run and where your data resides, with low-friction movement between cloud and edge locations to easily adapt to regulatory changes.

Run AI/ML workloads in more locations, closer to where data is generated and consumed. Enable low-latency predictions and real-time decision-making for use cases like fraud detection, patient diagnostics, and personalized recommendations.  

In select metros (Atlanta, Dallas, Phoenix), run large-scale Large Language Model (LLM) training and inference using the latest GPUs and AWS Trainium chips deployed with EC2 UltraClusters, interconnected using Elastic Fabric Adapter (EFA) networking for high-throughput AI/ML workloads. Use Amazon Bedrock and Amazon SageMaker to build, fine-tune, and deploy foundation models with your proprietary data. 

Migrate legacy on-premises applications to the cloud without disrupting latency-sensitive system dependencies. Move segments of your applications to a nearby AWS Local Zone, while still meeting the low-latency requirements in the hybrid deployment until you are ready to fully migrate.