What changed
AWS launched spending limits on September 16, 2026. Google Cloud launched Spend Caps in July. Both work the same way: you pick a monthly dollar ceiling for a project, starting as low as $20 on AWS. When your usage reaches that number, the cloud provider pauses your project instead of letting the bill grow.
AWS is rolling this out to new customers first. Google's feature is in preview for services including Gemini API, Cloud Run, and Maps. The cap uses estimated costs to trigger faster than a bill would arrive, though it is not instant and any overage that slips through still gets charged.
What you can do with it
If you have been afraid to try AWS or Google Cloud for a side project or a pilot because a mistake might cost you thousands, this changes the math. Set a $50 or $100 cap on a new project. Test an idea. Build a proof of concept. If something breaks or runs wild, you get an error instead of a bankruptcy notice.
For a small team testing AI tools hosted in the cloud, this is the safety rail you set before you hand any task to an agent. The agent might spin up compute or call an API hundreds of times. A hard cap means you control the maximum damage.
What to keep in your own hands
Set the cap before you deploy anything. Do not rely on a warning email. Warnings arrive after the spending starts. A hard cap stops it.
Check what the cap covers. Google's version applies to specific services in one project. AWS pauses the whole project. Neither cap covers every type of charge, like some subscription fees. Read the provider's list of what is included before you assume you are fully protected.
Know that lifting a cap is manual. If your service hits the ceiling at midnight, it stays paused until you raise the limit yourself. That is the right default, but it means you need a plan for who gets the alert and how fast they can act if the cap trips during business hours.