AWS re:Invent 2025 - Behind the curtain: How Amazon’s AI innovations are powered by AWS (INV211)
Source: Dev.to
Overview
AWS re:Invent 2025 – Behind the curtain: How Amazon’s AI innovations are powered by AWS (INV211)
In this session Amazon leaders demonstrate how AWS powers innovation across Amazon.com, Zoox, and Prime Video. Highlights include:
- Amazon Stores: Over 21,000 AI agents deployed, delivering $2 billion in cost savings and a 4.5× increase in developer velocity through AI‑native development with Kiro and Spec Studio.
- Amazon Rufus (generative AI shopping assistant): Scaled to 80,000 Trainium and Inferentia chips during Prime Day, processing 3 million tokens per minute; customers were 60 % more likely to complete purchases.
- Zoox: Autonomous robotaxis run trillions of calculations using tens of thousands of GPUs on AWS, leveraging SageMaker HyperPod and S3 for simulation and training.
- Prime Video: Features such as Defensive Alerts and the Burn Bar (built on ECS, Kinesis, and Bedrock) deliver real‑time AI predictions for NFL and NASCAR broadcasts.
All innovations showcase how AWS services enable massive‑scale AI deployment across Amazon’s businesses.
Day One Culture: The Foundation of Customer‑Centric Innovation
“We’re very comfortable that we’re doing a better job for customers, and the way you do that is just keep working on it every day, raising the bar every day. And that’s the heart of Day One.”
“If you want a culture that behaves like a Day One culture, you need people who wake up hungry to improve the customer experience, hungry to invent, hungry to move quickly, hungry to learn, and hungry to figure out what else could I be doing for customers to make their lives better that we’re not trying to do right now.”
Behind the Curtain: How Amazon Powers AI Innovations on AWS
Paul Roberts, Director of Technology, AWS opens the session:
“We’re going to go behind the curtain and show you how Amazon is powering its AI innovations on AWS. We’ll dive deep into Amazon Stores, Zoox, and Prime Video—some of our largest businesses built on AWS.”
Key takeaways
- Amazon Stores: AI agents orchestrated via Kiro and Spec Studio, running on Amazon SageMaker, AWS Lambda, and Amazon Bedrock.
- Zoox: Real‑time perception and planning workloads run on AWS EC2 GPU instances, with training pipelines using SageMaker HyperPod and data stored in Amazon S3.
- Prime Video: Real‑time analytics pipelines built with Amazon Kinesis Data Streams, AWS Lambda, and Amazon ECS, feeding AI models hosted on Amazon Bedrock for live sports insights.
Prime Day at Massive Scale: AWS Services Powering Amazon’s “Super Bowl”
Prime Day is Amazon’s flagship shopping event—its “Super Bowl.” With over 200 million Prime members and 9 billion packages shipped in a single year, the event demands extreme scalability.
“Prime Day doesn’t just happen. It takes months of planning and a massive, coordinated effort across the entire AWS ecosystem.”
Core AWS services that enable Prime Day at scale
| Service | Role |
|---|---|
| Amazon EC2 Auto Scaling | Dynamically provisions compute capacity for traffic spikes. |
| Amazon SageMaker | Powers recommendation engines and demand‑forecasting models. |
| Amazon DynamoDB | Provides low‑latency, highly available data storage for inventory and order state. |
| Amazon S3 & Glacier | Stores product media, logs, and archival data. |
| Amazon Kinesis | Streams click‑stream and transaction data for real‑time analytics. |
| AWS Lambda | Executes event‑driven functions for order processing and fraud detection. |
| Amazon CloudFront | Delivers static assets globally with low latency. |
| AWS Wavelength | Reduces latency for mobile shoppers on edge locations. |
These services work together to ensure a seamless shopping experience for millions of customers, handling billions of requests, and delivering real‑time AI‑driven insights throughout the event.