Compute rental lets a user or organization pay for processors, memory, storage, and related infrastructure without purchasing and maintaining the physical machines. Capacity may be billed by time, workload, reservation, or resource use.
AI workloads often rent graphics processing units, or GPUs, because they can perform many numerical operations in parallel. Access remains subject to the provider's identity, payment, usage, and acceptable-use controls.
ELI5
Compute rental is like hiring a powerful computer for a while instead of buying it. You pay for the amount of processing time or capacity that your job uses.
For example, a small research team might rent a graphics processor for several hours to analyze a large dataset, then stop paying when the calculation is finished.
Why rent computing power instead of buying hardware?
Renting can provide quick access, flexible capacity, and specialized hardware without the upfront cost and maintenance of owning machines.
Can providers control rented computing resources?
Yes. Providers can require accounts and payment, enforce usage policies, limit capacity, monitor service health, and suspend access under their terms.
