All the numbers: Amazon Prime Day 2026 powered by AWS

Prime Day is, every year, the biggest live load test in the industry — and once again it was powered by AWS. Here are the metrics that matter, because they're the practical upper bounds of what each service sustains under real traffic, not benchmark slides.

The big numbers

• EC2 + Graviton — AWS Graviton powered up to 49% of the EC2 compute used by Amazon.com during Prime Day 2026.
• EBS — peaked at over 24.8 trillion I/O operations per day, moving over an exabyte of data daily.
• Lambda — handled over 2.3 trillion invocations per day.
• ECS + Fargate — launched an average of 158.3 million tasks per day, a 47.7% increase over the previous year's Prime Day average.
• CloudFront — delivered over 2.1 trillion HTTP requests during the global week, up 5% versus Prime Day 2025.
• DynamoDB — over the June 23–26 event, processed over 59 trillion requests, peaking at 192 million requests per second, with single-digit millisecond responses throughout.
• Aurora — hundreds of billions of transactions; stored 5,491 TB, transferred 1,194 TB of data.
• ElastiCache — peaked at over 2.3 quadrillion daily requests and 2.1 trillion requests in a minute.
• Kinesis Data Streams — processed a peak of 988 million records per second.
• SNS — delivered 5 trillion messages in a single day.
• SQS — received a peak of 213 million messages per second.

Why this matters beyond the press release

These are the numbers to remember when someone tells you a service "can't scale." 192M DynamoDB requests per second with single-digit millisecond responses is the strongest durability argument for serverless databases that exists. 2.3 trillion Lambda invocations a day is the answer to every "Lambda doesn't scale" objection. Graviton at 49% of the largest retailer's compute shows where the industry's default silicon is heading. And the 47.7% year-over-year growth in ECS/Fargate tasks tracks the container-mainstreaming of even the most extreme workloads.

Bottom line

Prime Day isn't a benchmark — it's the real thing, and every one of these services absorbed it without a headline incident. If you're scaling Spring Boot on ECS, DynamoDB, or serverless, the practical ceiling these services operate under is far above your traffic — and the architecture patterns Amazon ships (Graviton, event-driven, Fargate) are the mainstream's default future.