Reserved capacity
Buying compute in advance (1-3 years) at a discount — saves money if you can predict usage.
Think of it like
Buying a season pass to the gym instead of paying per visit.
Example
You know you'll need 100 GPUs for 2 years; commit upfront and save 40% vs on-demand.
How it actually works
Reserved instances require guessing usage correctly. Overcommit and you waste money. Undercommit and you miss the discount. They're also hard to change once purchased. Good for steady, predictable workloads; bad for growing startups or experimental projects.
For product teams
Reserved capacity is profitable for mature, predictable products.
For engineers
Use reserved instances for baseline load; burst with on-demand or spot.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome