infrastructure//cloud computing

Cloud computing is the supply of infrastructure and managed services over a network, with elastic provisioning and metered or contracted consumption, and it moves much operational machinery behind a provider interface. It centralizes shared capacity, where edge computing places selected computation near the data sources or the consequences.

Cloud computing is the supply of infrastructure and managed services over a network, with elastic provisioning and metered or contracted consumption, and it moves much operational machinery behind a provider interface. It centralizes shared capacity, where edge computing places selected computation near the data sources or the consequences.

That interface accelerates experimentation and absorbs demand spikes, but creates dependencies on pricing, service semantics, region availability and provider control planes. Cloud is neither inherently cheaper nor less sovereign. The answer depends on utilization, required services, exit cost and which layers the organization is willing to operate itself; proprietary managed services can make a later migration far more expensive than the first adoption suggested (vendor lock-in).

Underneath the interface the machinery is shared. A provider sells slices of the same servers to many customers (multi-tenancy), separated by virtual machines and containers, or rents whole machines when performance or compliance demands it (bare metal); all of it sits in data centers whose power and cooling ultimately set what the provider can offer.

The market has two shapes. Hyperscalers sell hundreds of integrated services on one account; neoclouds sell mostly GPU capacity for AI, often in larger blocks and on long contracts. Which is cheaper depends on the workload, the commitment and the exit, never on the category.

Elasticity has a bill. Capacity that appears with an API call also stays with an API call left running, so the cost of the cloud is managed continuously rather than decided once at purchase (FinOps), and growth is scaled out over more machines rather than up into bigger ones (horizontal and vertical scaling).

The cloud is somebody else's industrial plant with an excellent API.