.png)
How It Works
Most cloud providers historically billed compute time in full-minute or full-hour blocks. Even if a workload ran for 61 seconds, customers paid for two full minutes. Per-second billing eliminates that rounding tax. The clock starts when a function or instance begins execution and stops the moment it completes. AWS Lambda and AWS Fargate both use per-second billing, with a short minimum charge applied to very brief executions. GCP charges Compute Engine VMs by the second after a one-minute minimum. Azure applies per-second billing to several compute services as well.
Why It Matters for Cloud Cost
For teams running short-lived or variable workloads, per-second billing directly reduces waste. Serverless functions that process events in under a minute, batch jobs that spin up and down frequently, and container tasks with unpredictable runtimes all benefit from not paying for unused time. At scale, the savings from eliminating billing rounding can be meaningful, particularly for high-frequency Lambda invocations or Fargate tasks. Teams that ignore this model often overprovision to avoid partial-minute penalties, which compounds the waste further.
Usage AI’s Flex Savings Plan covers EC2, Fargate, and Lambda, saving 40 to 60% versus on-demand with no upfront cost and a full buyback guarantee on any underutilization.