The ability to dynamically and automatically adjust cloud resources up or down based on actual demand in real-time. Like a balloon that expands and contracts based on how much air you put in.
During a viral marketing campaign, elastic cloud infrastructure automatically adds more servers as traffic spikes, then removes them when traffic returns to normal, optimizing costs.
Elasticity is a capability implemented through autoscaling services. All four providers can automatically add/remove compute instances based on metrics (CPU, requests, queue depth) and can scale across instance groups; they differ mainly in configuration style, supported metrics/integrations, and ecosystem defaults.