Sending every piece of data to a distant data centre adds delay and uses a lot of bandwidth. Edge computing processes data near where it is produced, on local devices or nearby servers, so results come faster and only what matters is sent onward.
This suits tasks that need quick responses or generate huge data, such as cameras, factory sensors and vehicles, and it keeps sensitive information local. The trade-off is managing many small, distributed systems instead of one central one, so edge and cloud are usually combined, with quick decisions made locally and heavier analysis done centrally.