Egress fees are per gigabyte charges for data moving out of a cloud, and they come in three layers: traffic to the internet, traffic between regions, and on some providers traffic between availability zones inside one region. Data coming in is generally free; data going out is where the meter runs, which is why a workload that serves large volumes to users or replicates across regions can carry a transfer bill that rivals its compute. Among the four, OCI has deliberately priced egress lower, with a larger free allowance and cheaper per gigabyte rates, while AWS, Azure, and GCP price internet egress higher on tiered scales that fall with volume. The levers are the same everywhere: keep traffic local, cache at the edge, compress, replicate only what you must, and route on purpose rather than by default.
Here is how each cloud meters data out, where the hidden lines sit, and the moves that cut the transfer bill without breaking the architecture.
What exactly counts as egress, and why is it hidden?
Egress is metered wherever data crosses a boundary the provider charges for. The obvious boundary is the internet: every byte your application sends to a user, an API consumer, or another cloud is billed per gigabyte. Less obvious are the internal boundaries. Traffic between regions is almost always charged, so a multi region deployment pays to keep replicas in sync. Traffic between availability zones inside a single region is charged on some providers and configurations, so a microservice in zone A calling a database in zone B can rack up transfer cost on every request. The reason egress feels invisible is that it is never one line. It is a small charge attached to load balancers, gateways, replication, and inter service calls all over the bill, and only grouping the billing export by transfer usage type reveals the total.
How do the four clouds compare on egress?
The structural picture, holding traffic constant, looks like this. Exact rates are tiered and change, so treat these as indicative directional comparisons and verify against each provider's current pricing page before modelling.
| Cloud | Internet egress posture | Notable transfer mechanics |
|---|---|---|
| AWS | Higher per gigabyte, tiered down with volume; first allowance free (indicative) | Cross zone and NAT gateway processing add up quietly; review data transfer and gateway lines |
| Azure | Similar hyperscaler range; tiered down with volume (indicative) | Inter zone and inter region transfer billed; bandwidth can sit inside other service lines |
| GCP | Tiered internet egress; premium versus standard network tier changes the rate | Choosing the standard network tier where latency allows reduces egress cost |
| OCI | Materially cheaper, with a large free monthly allowance and lower per gigabyte rates (indicative) | Positioned as a deliberate differentiator for egress heavy workloads |
The headline is that for a workload dominated by outbound data, OCI can be dramatically cheaper on the transfer line alone, enough to change the cloud selection even when compute rates look similar. On GCP, the choice between premium and standard network tiers is a real lever most teams never touch. On AWS, NAT gateway processing and cross zone chatter are the quiet eaters rather than headline internet egress.
Where does egress quietly explode?
Three patterns produce most surprise transfer bills. The first is cross zone microservice chatter: a service mesh spread across zones for resilience pays transfer on every internal call, and at high request volume that dwarfs the internet egress. The second is cross region replication chosen for safety but never sized against its cost, where keeping a warm replica in a second region bills continuous transfer. The third is serving large static assets straight from object storage to the internet instead of through a content delivery network, paying full egress on every download instead of caching at the edge. None of these is wrong by default, but each should be a deliberate choice with its cost on the table, not an accident of the default architecture.
Group your billing export by transfer usage type and rank the lines. If cross zone or NAT gateway processing outranks your internet egress, the problem is internal architecture, not customer traffic, and it is fixable by colocating chatty services without changing what users experience.
How do you cut the transfer bill without breaking things?
The safe moves stack. Colocate services that talk constantly into the same zone so internal calls stop crossing a charged boundary. Put a content delivery network in front of static and cacheable content so the edge serves repeat requests instead of origin egress. Compress payloads so each byte carries more. Replicate across regions only what genuinely needs disaster recovery, and tier the rest. On GCP, test the standard network tier where latency budgets allow. For genuinely egress heavy workloads, model the transfer line on OCI or commit to network pricing, because at scale the per gigabyte rate is the dominant variable. Each of these reduces cost without reducing what the customer receives, which is the test of real optimization.
A European SaaS company serving large media files saw its transfer line approach its compute line. We grouped the export by transfer type and found three causes: object storage served directly to users with no edge cache, cross zone replication for a stateless service that did not need it, and uncompressed payloads. Adding a content delivery network, colocating the stateless service, and compressing responses cut the transfer bill substantially, and modelling the egress heavy tier on cheaper egress informed the longer term placement decision. This was part of the work that left the estate materially lighter. Figures are verified against billing data and anonymized.
Where this fits in your multicloud strategy
Egress is a major input to both cloud selection and architecture. For the full transfer picture, read multicloud networking costs explained, and for how transfer feeds the broader price comparison, read AWS versus Azure versus GCP versus OCI on price. The cross cloud framework sits in the cloud cost optimization guide.
Frequently asked questions
What are cloud egress fees?
Which cloud has the cheapest egress?
How do you reduce cloud egress costs?
Find the transfer cost hiding in your bill
We group your billing data by transfer type, expose the cross zone and egress lines, and turn them into a safe reduction plan, independent of any provider and taking zero provider commissions. Our guarantee: we reduce your cloud spend or we reimburse our service fee. Pricing is either a Fixed Fee scoped up front or Gainshare, a share of verified savings with no retainer and no risk. Start with our playbook.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.