Reserved instances are cloud pricing commitments that reduce the cost of predictable virtual machine or compute usage in exchange for a fixed term. They are usually a billing benefit, not a server that is set aside exclusively for one customer. Cloud providers apply the discounted rate when running usage matches the reservation's eligible attributes.
Reserved Instances
What are reserved instances?
Reserved instances are cloud pricing commitments that reduce the cost of predictable compute usage for a fixed term. They are usually a billing discount, not a virtual machine or physical server reserved exclusively for you.
When an eligible workload matches the reservation's conditions, the provider applies the discounted rate. Reservations can cover virtual machines, database compute, and other always-on services. Eligibility may depend on provider, region, instance family, operating system, tenancy, and reservation type. A separate capacity reservation may be needed to guarantee launch capacity.
How reserved instances work
A reservation matches stable usage with a longer-term pricing commitment.
- Measure compute that runs consistently, such as a production API or database.
- Select eligible attributes, including provider, region, instance family, operating system, and tenancy.
- Choose a provider term and payment option.
- Run workloads that match the eligibility rules.
- The provider applies the discounted rate to matching usage.
Reserved instances vs on-demand, savings plans, and spot instances
Rules differ among providers, but lower cost usually means less flexibility.
| Option | Commitment and flexibility | Best fit |
|---|---|---|
| On-demand | No long-term commitment, highest flexibility | New, changing, or short-lived workloads |
| Reserved instances | Term commitment tied to eligible usage attributes | Stable baseline servers and databases |
| Savings plans | Spend commitment, often broader coverage | Compute that may change instance types or services |
| Spot instances | Low-cost surplus capacity that can be interrupted | Fault-tolerant batch jobs and flexible workers |
Benefits of reserved instances
Reservations work best when usage is steady and can be forecast confidently.
- They reduce the effective cost of stable, always-on compute.
- They make infrastructure spending more predictable.
- They suit production APIs, databases, and workers with a reliable baseline.
- Some providers offer capacity-related options for workloads that need placement assurance.
Practical limits of reserved instances
A discount helps only while the committed usage remains relevant.
- Unused reservations can still create cost after an application is retired or downsized.
- Changes to architecture, region, operating system, or instance family can reduce eligibility.
- A reservation does not automatically guarantee new instances can launch during high demand.
- Terms can make rapid migrations or major platform changes harder to manage.
How to decide whether to reserve compute
Use observed usage, not launch-plan estimates.
- Review several billing periods to find the compute that runs continuously.
- Exclude temporary peaks from autoscaling, campaigns, or load tests.
- Check planned migrations, redesigns, and service retirements.
- Compare the reservation's break-even cost with on-demand pricing.
- Buy gradually for the stable floor and review utilization regularly.
Reserved instances for applications built with Emergent
Teams building products through Emergent's AI web app builder can validate production demand before committing to cloud spend. After launch, use application design and development practices to separate dependable baseline compute from variable traffic. Reserve only the baseline and keep seasonal or uncertain demand on flexible pricing. For more context, see the AI app builder glossary entry.
Start Building
on Emergent today
on Emergent today
Start Building
