How Sustained-Use Discounts work automatically on Google Compute Engine, which machine families qualify, and how they interact with other discounts.
What Are Sustained-Use Discounts?
Sustained-Use Discounts (SUDs) are automatic price reductions applied to Compute Engine VMs that run for a significant portion of the billing month. You do not purchase anything, sign anything, or configure anything. If your VM runs long enough, the discount appears on your bill.
SUDs apply retroactively at the end of each billing month based on cumulative usage. The longer a resource runs during the month, the bigger the discount.
How SUDs Work
SUDs use a tiered structure. As a VM runs more hours in a month, each additional tier gets a deeper discount on the per-hour rate.
For machine families with a maximum 30% monthly SUD (N1, M1, M2, as of May 2026):
| Usage Tier | Discount per Tier | Effective Overall Discount |
|---|---|---|
| 0% to 25% of month | 0% (full on-demand) | 0% |
| 25% to 50% of month | 20% off | 10% |
| 50% to 75% of month | 40% off | 20% |
| 75% to 100% of month | 60% off | 30% |
For machine families with a maximum 20% monthly SUD (N2, N2D, C2, as of May 2026):
| Usage Tier | Discount per Tier | Effective Overall Discount |
|---|---|---|
| 0% to 25% of month | 0% | 0% |
| 25% to 50% of month | ~13% off | ~7% |
| 50% to 75% of month | ~27% off | ~13% |
| 75% to 100% of month | ~40% off | 20% |
A VM running 100% of the month (roughly 730 hours) on N1 gets the full 30% discount on the entire month’s usage, not just the hours above 75%.
Eligible Machine Families
| Machine Family | Max Monthly SUD | Notes |
|---|---|---|
| N1 (general-purpose) | 30% | Also covers N1 custom types, sole-tenant, and GPUs attached to N1 |
| M1, M2 (memory-optimized) | 30% | Also covers M1/M2 sole-tenant |
| N2 (general-purpose) | 20% | Also covers N2 custom types and sole-tenant |
| N2D (general-purpose, AMD) | 20% | Also covers N2D custom types and sole-tenant |
| C2 (compute-optimized) | 20% | Also covers C2 sole-tenant |
| f1-micro, g1-small | 30% | Shared-core machine types |
NOT eligible for SUDs:
| Resource | Reason |
|---|---|
| E2 machine series | Lower on-demand pricing instead |
| C3, C3D, C4, C4A, C4D | Newer families with competitive on-demand pricing |
| N4, N4D, N4A | Newer families with competitive on-demand pricing |
| H3, H4D | Compute-optimized, newer generation |
| Spot VMs and preemptible VMs | Already deeply discounted |
| A2, A3 (accelerator-optimized) | GPU-focused, not eligible |
| Local SSDs | Storage, not compute |
| Premium OS licenses | Charged separately |
Key Point: Google’s newer machine families (E2, C3, C4, N4) do not get SUDs. They already have lower on-demand pricing that is intended to be competitive without additional discounts. For these families, use CUDs or Spot VMs for savings.
Usage Aggregation
You do not need to run a single VM continuously to earn SUDs. Google aggregates all usage for the same machine family in the same region within your billing account.
Example: You run 10 n2-standard-4 VMs in us-central1 for 73 hours each (10% of the month). Together, that is 100% equivalent usage of one n2-standard-4, which qualifies for the full 20% SUD.
Aggregation rules:
- Usage is aggregated across all VMs of the same machine family in the same region
- Applies across all projects under the same billing account
- SUDs reset at the beginning of each billing month
- Moving a project to a different billing account resets accumulated usage to zero
SUDs and CUDs Together
CUDs and SUDs are mutually exclusive per resource-hour, but they work together on your overall bill:
- CUDs apply first to any usage that matches a commitment
- SUDs then apply to the remaining on-demand usage
This means if you have a CUD covering 60% of your N2 usage, SUDs will apply to the other 40% if those VMs run long enough. You do not lose SUDs on uncommitted usage.
TL;DR
- SUDs are automatic. No purchase, no configuration. Just run VMs and the discount appears on your bill.
- N1 and M1/M2 get up to 30% off. N2, N2D, and C2 get up to 20% off.
- E2, C3, C4, N4, and newer families do not qualify for SUDs. Use CUDs or Spot VMs instead.
- Usage is aggregated across VMs of the same family in the same region under the same billing account.
- CUDs apply first, SUDs apply to remaining on-demand usage. They do not stack on the same resource-hour.
- Treat SUDs as automatic savings on VMs that already need to run. Do not keep idle VMs online just to earn a larger discount.
Resources
Sustained-Use Discounts Official documentation with detailed tier calculations and eligible resources.
Committed-Use Discounts For deeper discounts on predictable, long-term workloads.
Committed-Use Discounts When SUDs are not enough and you want to commit for a bigger discount.
Cost Optimization Overview of all cost levers on Google Cloud.