Edge computing
Also known as: edge cloud, edge processing
Edge computing means processing data near where it's produced, on the device itself, in a local server at a store or factory, or at a nearby network site, instead of sending everything to a distant cloud data centre.
That reduces delay, saves bandwidth, and keeps working if the internet connection drops. While edge computing is essential for things like self-driving machines and factory automation, managing many small, spread-out systems is harder than managing a few big ones.
Edge computing, explained at three levels
At home
Instead of sending everything to a faraway data centre, edge computing does the thinking close by, like a smart doorbell recognizing a package on the camera itself instead of in the cloud.
At work
Edge computing lets businesses process data on-site, which matters for fast decisions, limited internet, and privacy.
At home
Sending video or sensor data across the internet takes time and bandwidth. Processing it nearby gives faster responses and means only important results need uploading.
At work
Edge deployments range from devices with built-in AI chips to small servers in stores, factories, or cell towers. Common uses include quality inspection with cameras, retail analytics, predictive maintenance, and local caching.
At home
Keeping data on your device can protect privacy, since less of it leaves your home. In my opinion, that's one of the most practical benefits of edge AI on phones.
At work
Challenges include remote management and patching of many devices, physical security, consistent deployment (often using containers and lightweight Kubernetes), connectivity loss handling, and deciding which data and models live at the edge versus the cloud.
Real-world examples
At home
Lena's home security camera alerts her when a person, not a passing car, is at the door. The camera recognizes people itself rather than sending all its video to the cloud.
At work
Jerome manages technology at a food processing plant in Manitoba. Cameras with edge AI check products on the line in milliseconds, rejecting defects without waiting for a cloud response.
How it compares
| feature | Edge computing | Cloud computing |
|---|---|---|
| Where processing happens | Near the data source | Central data centres |
| Delay | Very low | Higher |
| Works offline | Often | No |
| Management | Many distributed systems | Centralized |
Edge is fast and local. Cloud is centralized and easier to manage.
For analysts
I'd put work at the edge only when delay, bandwidth, or offline operation truly require it. Everything else is usually cheaper and simpler in the cloud.
History & origin
Content delivery networks such as Akamai, founded in 1998, pioneered the idea of serving content from servers close to users. Cisco promoted 'fog computing' around 2012 for processing Internet of Things data near devices. As IoT and video analytics grew in the mid-2010s, the term edge computing caught on. Cloud providers extended their platforms outward: AWS announced Outposts and Wavelength in 2019, putting cloud services in customer sites and telecom networks, and Microsoft and Google followed with similar offerings. Running AI models on phones and local devices became a major edge use in the 2020s. In my view, edge computing is a reminder that the cloud is a tool, not a destination for every byte.