VPC endpoints pricing: what to model (interface vs gateway endpoints)

Reviewed by CloudCostKit Editorial Team. Last updated: 2026-01-27. Editorial policy and methodology.

Start with a calculator if you need a first-pass estimate, then use this guide to validate the assumptions and catch the billing traps.


Use this page when you need to decide what belongs inside the endpoint bill model before you argue about optimization.

This guide is about bill boundaries: endpoint-hours, GB processed, endpoint type, and the transfer or architecture costs that should be tracked beside endpoints rather than confused with them.

What to model (interface vs gateway endpoints)

  • Interface endpoints: endpoint-hours (endpoints x AZs x hours) and sometimes a per-GB processing fee.
  • Gateway endpoints: a different cost model depending on service; treat them separately.
  • Transfer boundaries: cross-AZ/cross-region paths can create separate transfer line items.

The same "private access" goal can be achieved with different endpoint types. Always model the endpoint line item and the transfer path together.

Inside the endpoint bill vs outside the endpoint bill

  • Inside the endpoint bill: interface endpoint-hours, per-GB processing where applicable, and endpoint-type differences that change the direct line item.
  • Usually outside the endpoint bill: NAT charges that remain, cross-AZ transfer created by the path, and downstream service bills that endpoints merely expose or reroute.
  • Why that boundary matters: teams often blame endpoints for the whole network bill when the real multiplier sits in path locality, transfer boundaries, or services still using NAT.

Interface vs gateway endpoints: decision points

  • Interface endpoints (PrivateLink): great when you need private access to many AWS services, but they add endpoint-hours per AZ and can multiply quickly across environments.
  • Gateway endpoints: a different mechanism with a different cost model. Treat them separately and verify which services in your architecture can use them.
  • Path matters: if clients in one AZ consistently talk to endpoints in another AZ, you can introduce cross-AZ transfer even if the endpoint line item looks correct.

Fast estimate (baseline + scenario)

Use VPC Interface Endpoint Cost Calculator for a first-pass estimate of endpoint-hours + per-GB processing.

  • Baseline: current NAT GB processed and current endpoint inventory (if any).
  • Scenario: endpoints added + NAT traffic reduced + any expected cross-AZ path changes.

Worked estimate template (copy/paste)

  • Endpoint-hours = endpoints * AZs per endpoint * hours/month
  • Endpoint GB processed = GB/month expected to flow through endpoints
  • Endpoint cost = hourly baseline + per-GB processing (if applicable)

Quick sanity check: if you add many endpoints across 3 AZs, the hourly baseline can dominate even when GB processed is modest. Model the hours first, then refine GB.

NAT break-even checklist

  • How much NAT GB processed is actually AWS-service traffic that can move to endpoints?
  • How many endpoints will you deploy (and in how many AZs) across prod/staging/dev?
  • Will endpoint routing change traffic locality (cross-AZ) or keep it AZ-local?
  • Do you need endpoints everywhere, or only in a subset of VPCs and accounts?

Common pitfalls

  • Forgetting that interface endpoints scale with the number of AZs you use.
  • Moving traffic across AZs or regions and creating new transfer charges.
  • Assuming endpoints replace NAT costs without validating the traffic mix.
  • Counting endpoints once while they exist in multiple VPCs and environments.

How to validate the pricing model

  • Reconcile endpoint-hours in billing against endpoints * AZs * hours.
  • Spot-check the top destinations and verify they are the services you intended to route through endpoints.
  • After rollout, compare NAT GB processed and endpoint GB processed as a before/after sanity check.

When this is not the right page

Related guides

Sources


Related guides


Related calculators


FAQ

Do VPC endpoints always save money?
Not always. Endpoints add their own hourly fees (and sometimes per-GB processing), but they can reduce NAT GB processed and improve security by keeping traffic private. Model your traffic mix to find break-even.
What's the biggest multiplier for endpoint cost?
Availability Zones. Interface endpoints are deployed per AZ. If you attach endpoints in 3 AZs, your endpoint-hours are roughly 3x compared to 1 AZ.
Are gateway endpoints billed the same way as interface endpoints?
No. Gateway endpoints differ by service and can have a different cost model. The key is to treat endpoints as their own line items and model your actual traffic path.

Last updated: 2026-01-27. Reviewed against CloudCostKit methodology and current provider documentation. See the Editorial Policy .