10.7 Cloud fundamentals and cost
You can deploy to a cloud and predict the monthly bill.
Cloud fundamentals — compute, object storage, managed databases, GPU pricing — are what let you deploy and predict the monthly bill before it arrives, with spot instances and serverless each having a place once their failure modes are known. It sits here as the deployment substrate. The horror stories are real and preventable: resources left running are the classic ML cloud bill, and alerts plus caps belong in place before the first GPU spins up.
Work through these
Compute, object storage, managed databases
The three cloud services almost every machine learning system uses: machines to compute on, storage for objects, and a managed database. Knowing these three covers most deployments.
GPU instance types and spot pricing
Accelerator instance types differ enormously in memory and speed, and interruptible capacity is far cheaper in exchange for being reclaimed. Designing for interruption is what makes that discount usable.
Serverless vs. containers vs. VMs
Functions, containers and virtual machines trade operational work against control and cost. The right answer depends on how steady the traffic is.
Cost monitoring and budget alerts
Watching what is being spent and being warned before the bill arrives. For accelerator workloads this is not optional, because the amounts involved are large.
Sign in to keep your progress.
Free resources
We haven't checked most of these for screen reader use yet.
Links last checked 29 Aug 2026.
Stuck here?
Ask a mentor. A real person answers, and they can see exactly which topic you're on. Usually within a couple of working days.
Checking your session…
Topics shown in module order.