Seattle, Washington, United States
Amazon Web Services (AWS)
The largest cloud provider by market share, and the infrastructure most AI teams still default to first - GPU instances, managed model access through Bedrock, and the storage and networking underneath both.
Where it sits in the stack
- User - Amazon Web Services (AWS) does not work at this layer
- Application - Amazon Web Services (AWS) does not work at this layer
- ModelThe AI itself - Amazon Web Services (AWS) works at this layer
- InferenceRunning the model, live - Amazon Web Services (AWS) works at this layer
- ComputeThe processing power - Amazon Web Services (AWS) works at this layer
- GPUThe physical chips - Amazon Web Services (AWS) works at this layer
- Data CentreThe building - Amazon Web Services (AWS) works at this layer
- Power & Network - Amazon Web Services (AWS) does not work at this layer
Has a site in the UAE
AWS operates the Middle East (UAE) Region (me-central-1) in Dubai, with three Availability Zones, live since 2022.
- Middle East (UAE) - me-central-1, Dubai
What it offers
Amazon EC2
On-demand virtual compute, including GPU-accelerated instance families used for AI training and inference.
Amazon Bedrock
Managed access to Amazon's own and third-party foundation models through a single API.
Amazon S3
Object storage that sits under most AI training and retrieval pipelines built on AWS.
What it can do
- GPU and custom-silicon (Trainium/Inferentia) instances
- Managed foundation-model access via Bedrock
- Global fibre backbone and edge network
What its 5 layers mean
- Model
The trained system that actually produces the answer - GPT, Gemini, Claude, Llama and others like them - plus the platforms that make models like these available for developers to build on.
- Inference
The moment a trained model is actually used to produce an answer, in about the time it takes to read this sentence. Training a model happens once; inference happens every single time someone uses it.
- Compute
The raw computing capacity - rented by the hour, the month or the year - that trains and runs AI models. This is what "the cloud" mostly means in practice.
- GPU
The specialised chips - mostly made by NVIDIA - that do the actual arithmetic behind AI, built to run millions of small calculations in parallel far faster than a general-purpose computer chip can.
- Data Centre
The physical building - racks, servers, cooling - where everything above this actually lives. An AI model has no existence outside a room full of running machines.
Something out of date?
Tell us and we will check it against the official source.