Cloud egress costs sit in a category of their own: a charge that grows silently in the background, invisible in architecture reviews, and rarely discussed until someone stares at a bill and asks where an extra $40,000 went. For Australian organisations, the problem is sharper than it is for customers in the United States or Europe. Local data transfer pricing from AWS, Azure, and GCP carries a meaningful distance premium, and the typical enterprise workload generates far more outbound traffic than architects anticipate at design time.
What egress actually means and why it's expensive
Ingress, moving data into a cloud provider's network, is almost always free. Egress, moving data out, is where the charges begin. The cost varies depending on where the data is going: from a cloud region to the public internet, from one availability zone to another, or from one cloud provider to a different one. Each path carries a different rate, and the rates are not always published clearly enough to model upfront.
Australian regions add a layer of complexity. AWS charges roughly USD $0.114 per GB for internet egress out of its Sydney region, which is higher than the equivalent US rate of around $0.09 per GB. Azure and GCP follow a similar pattern, with Asia-Pacific egress pricing sitting above North American baselines. That gap matters less at small scale and considerably more once a workload is streaming telemetry, serving media, or replicating databases across regions regularly. A workload that transfers 100 TB per month out of an Australian region can face an egress line item of more than USD $11,000 before any other costs are counted.
Cross-region transfer costs compound this further. An application that spans AWS Sydney and AWS Melbourne for resilience purposes, for example, incurs inter-region data transfer charges on top of internet egress charges. Teams building multi-region cloud strategies for Australian businesses often discover this pricing reality only after deployment, when the architecture is already committed.
The workloads that generate the most egress
Not all workloads are equal. A few categories generate disproportionate egress volume and deserve specific attention during architecture review.
- Analytics and business intelligence. Query results, report exports, and dashboard data pulled to client browsers or on-premises data warehouses add up quickly, especially when running tools like Power BI or Tableau against cloud-hosted data sources.
- Backup and disaster recovery. Replicating backup sets across regions or pulling recovery data from cloud storage during a restore event can trigger egress charges at the worst possible time. Teams planning cloud disaster recovery in Australia should model egress costs into their RTO/RPO economics, not just storage and compute.
- Log and telemetry streaming. Shipping logs from cloud workloads to a SIEM or a third-party observability platform is a common pattern and a consistent egress generator. A modest 50-node environment pushing detailed telemetry can produce several terabytes per month of outbound transfer.
- API-heavy SaaS integrations. Microservices calling external APIs return data inbound for free and ship responses outbound for a fee. In high-throughput integration pipelines, this can accumulate faster than monthly reviews catch it.
How the three major clouds price Australian egress
AWS, Azure, and GCP each publish their egress rates, but the structures differ enough that direct comparison is not straightforward. AWS prices internet egress on a tiered scale, with rates falling slightly after the first 10 TB per month. Azure includes a free tier of 100 GB per month for internet egress and then charges approximately AUD $0.165 per GB for the next 10 TB out of its Australian East (Sydney) and Australia Southeast (Melbourne) regions. GCP applies a similar tier structure but also offers committed use discounts on egress in some configurations, which AWS and Azure don't replicate in the same way.
All three providers offer reduced or free transfer between services within the same availability zone. Moving data between availability zones within the same region costs a small per-GB fee on AWS and Azure. Moving data between regions costs significantly more. Multicloud architectures that pass data between providers over the public internet pay full internet egress rates on the sending side, which is one reason why multicloud strategy in Australia requires careful data flow modelling before commitments are made.
Practical ways to reduce egress costs
Reducing egress charges is mostly an architectural discipline rather than a procurement one. There are no discounts to negotiate in the way you might push back on compute reserved instances. The spend has to be reduced by changing where processing happens and where data moves.
Processing data at source is the most effective lever. If your analytics workload pulls raw data out of cloud storage to transform it on-premises, every byte of that raw data triggers an egress charge. Running the transformation inside the cloud and only exporting the result cuts transfer volume significantly. This is sometimes described as "compute close to the data," and the principle applies at multiple scales, from SQL queries run in-region to machine learning inference served from within the cloud boundary.
Content delivery networks reduce egress charges for publicly facing content. AWS CloudFront, Azure CDN, and Google Cloud CDN each apply lower egress rates to traffic served through their own CDN networks than to direct internet egress from storage or compute. For applications serving static assets, media, or API responses to Australian end users, routing through a CDN is almost always cheaper than serving directly from origin.
Compressing data before transfer sounds basic but is frequently skipped. A logging pipeline that ships uncompressed JSON at 5 GB per hour will cost roughly three times more to egress than the same pipeline compressing to gzip before transfer. The compute cost of compression is negligible compared to the transfer savings at scale.
Audit your data flows before assuming the problem is the cloud bill. Many organisations discover during egress cost reviews that they're sending data to third-party vendors they no longer actively use, replicating datasets to multiple destinations where one would suffice, or running test pipelines against production-scale data with full egress switched on. A quarterly data flow audit, mapping source to destination for every significant transfer, often reveals 20 to 30 per cent of egress volume that can be cut without any architectural change.
Negotiate where you can, but know the limits
Large Australian enterprises with significant cloud commitments can negotiate custom egress pricing as part of enterprise discount agreements with AWS, Azure, and GCP. These arrangements are not publicly advertised and require dedicated spend commitments, typically in the millions of dollars annually, before providers will discuss them. For mid-market organisations, the standard published rates apply in full.
Australia's geographic isolation also matters here. The Australian Competition and Consumer Commission examined cloud pricing in its 2023 cloud services inquiry, noting that data transfer costs disproportionately affect Australian customers due to limited local interconnect options and the distance from major global network hubs. Megaport's network interconnect fabric, which connects directly to cloud on-ramps in Australian data centres, can cut egress costs for organisations that are moving significant data between their own infrastructure and cloud providers. The savings come from avoiding public internet egress rates by using private interconnect, which carries lower transfer pricing on AWS Direct Connect, Azure ExpressRoute, and GCP Cloud Interconnect.
Egress costs won't disappear from cloud bills, but they don't have to grow unchecked. Architectural discipline, compression, CDN routing, and quarterly flow audits are the four controls that consistently move the number. The providers have no incentive to make this easy. That's the job of the team managing the environment.