GCP egress charges are one of the most underestimated line items on enterprise cloud bills. Understanding Google Cloud's data transfer pricing model — and knowing which costs can be negotiated or eliminated — can save six figures annually for large workloads.
This guide is part of the Google Cloud Contract Negotiation: Enterprise Buyer's Guide. Networking costs — particularly egress — are consistently the most surprising element of GCP bills for enterprises migrating from on-premises or other clouds. Unlike compute costs, which GCP optimises automatically through Sustained Use Discounts, egress pricing is billed per gigabyte and adds up fast when you have data-intensive workloads, microservices architectures, or multi-region deployments. See also our cross-cloud egress comparison guide for AWS and Azure benchmarks.
Google Cloud charges for data that leaves its network — specifically, data that crosses boundaries between regions, zones, or exits to the public internet. Understanding exactly where these boundaries fall is the first step to controlling costs.
Ingress is always free. Google charges nothing for data moving into GCP from the internet, from other cloud providers, or from on-premises environments. This makes GCP an attractive destination for data migration projects, but it also creates an asymmetry: moving data in is free, but moving it back out can be expensive.
Not all traffic types are equal. GCP applies different pricing depending on whether traffic stays within a zone, crosses zones within a region, moves between regions, or exits entirely to the internet or another cloud provider. The pricing jumps significantly at each boundary.
GCP uses a tiered pricing model for internet egress that decreases per-GB cost as monthly volume increases — but most enterprises hit the ceiling of those tiers quickly enough that they end up paying standard rates for the bulk of their egress. The real opportunity is in restructuring how your workloads generate egress in the first place.
The following table shows GCP's standard published pricing. Actual enterprise pricing may include negotiated discounts for high-volume commitments, particularly in GCP Committed Use Agreements (CUAs).
Want independent help negotiating better terms? We rank the top advisory firms across 14 vendor categories — free matching, no commitment.
| Traffic Type | Pricing | Notes |
|---|---|---|
| Ingress (all types) | Free | Data moving into GCP — always free |
| Intra-zone (same zone) | Free | Traffic between VMs in the same zone |
| Inter-zone (same region, different zones) | $0.01/GB | Each direction charged separately |
| Inter-region (within GCP) | $0.01–$0.08/GB | Varies by region pair; US regions cheapest |
| Internet egress (US/Canada to internet) | $0.08/GB | First 1 TB/month; tiered discounts above |
| Internet egress (Europe to internet) | $0.085/GB | Standard rate; network Premium tier |
| Internet egress (Asia/APAC to internet) | $0.12–$0.15/GB | Higher rates for APAC regions |
| Cloud Interconnect egress | $0.02–$0.05/GB | 40–75% reduction vs internet egress |
| Cloud VPN egress | $0.045–$0.065/GB | Varies by region; cheaper than raw internet |
| Google Cloud CDN origin egress | $0.04/GB | 50% reduction vs internet egress for cached content |
The inter-zone charge of $0.01/GB sounds trivial, but for microservices architectures generating hundreds of TBs of internal traffic monthly, this can become the largest networking line item. Co-locating communicating services in the same zone is the single highest-ROI architecture change for networking cost reduction.
Beyond the obvious internet egress charges, several GCP pricing mechanisms generate unexpected bills that many enterprises only discover after their first month of production workloads.
Google Cloud's load balancers charge a data processing fee in addition to hourly instance fees — typically $0.008 per GB processed. For high-traffic applications using HTTP(S) Load Balancers, this can add significantly to total networking costs. This fee applies to all traffic processed by the load balancer, including inbound traffic that itself is technically free as ingress.
Cloud NAT (Network Address Translation) for private VMs accessing the internet charges both data processing fees and standard egress rates. Enterprises that assume NAT is "free" because they're not directly allocating internet IPs are often surprised to find double-counting of the same egress traffic.
Services like Cloud Storage, Cloud Spanner, and AlloyDB with multi-region configurations automatically replicate data across regions. This replication traffic incurs inter-region egress charges that compound with the storage replication premium. A multi-region Cloud Storage bucket doesn't just cost more for storage — it continuously generates inter-region egress as data is replicated.
When BigQuery query results are returned to client applications, that data transfer incurs egress charges. Large result sets — common when analytics dashboards pull detailed data rather than pre-aggregated summaries — can generate significant egress fees that show up on billing as BigQuery network costs, not storage or compute. See our GCP cost optimization guide for BigQuery-specific tactics.
GKE clusters spanning multiple zones (recommended for high availability) generate inter-zone traffic charges whenever pods in different zones communicate. East-west traffic within a Kubernetes cluster — service mesh overhead, health checks, internal APIs — adds up quickly in multi-zone deployments. See our Kubernetes cost optimization guide for pod placement strategies.
Cloud Armor (DDoS protection) charges a data processing fee on top of load balancer fees. If you've enabled Cloud Armor policies on your load balancers, you're paying for each GB twice at the network layer. Review your Cloud Armor policy scope and ensure it's only applied to public-facing services that actually require it.
Get the IT Negotiation Playbook — free
Used by 4,200+ IT directors and procurement leads. Oracle, Microsoft, SAP, Cloud — all covered.
Cloud CDN is often evaluated purely as a performance product, but for high-egress workloads it's primarily a cost tool. The economics are straightforward: CDN edge nodes serve cached content to end users, reducing the volume of traffic that must transit Google's origin servers and incur standard internet egress charges.
The CDN pricing structure rewards this use case. CDN cache fill — the traffic from GCP origin to CDN edge nodes — is charged at CDN origin egress rates ($0.04/GB to North America, lower than standard internet egress). Traffic served from CDN to end users is charged at CDN egress rates ($0.0008–$0.02/GB depending on region and volume), which are dramatically lower than standard internet egress rates. For a workload with 80% cache hit rate, the effective egress rate drops from $0.08/GB to roughly $0.02/GB — a 75% reduction.
The key limitation: CDN only helps with cacheable content. Dynamic API responses, personalised content, and streaming media with unique sessions don't benefit from CDN caching. For these workloads, Cloud Interconnect or architectural changes (co-location of compute and storage) are the primary levers.
Edge compute for regional data processing. Rather than centralising all compute in a single region and exporting results globally, deploying lightweight compute in each region where your users are located eliminates cross-region egress for the final data delivery step. GCP's regional endpoints and Cloud Run deployments support this pattern well.
Event-driven data movement. Instead of continuous data synchronisation between regions or to external systems (which generates continuous egress), use Pub/Sub triggered workflows that batch-export data at scheduled intervals. This doesn't reduce total egress volume, but it reduces the frequency of TCP connection overhead and allows you to use cheaper Standard Tier network routing for non-time-sensitive transfers.
Data locality for analytics workloads. Where possible, run analytics against data in the same region where it resides. Moving raw data to a centralised analytics region for processing generates inter-region egress. BigQuery's federated queries can query external tables in Cloud Storage without moving data, preserving data locality while enabling cross-regional analysis.
For organisations managing complex multi-cloud or hybrid architectures, see our multi-cloud cost optimization guide and cloud BYOL guide for additional cost frameworks.
Many enterprises don't realise that Google Cloud's committed spend agreements (previously called GCP Committed Use Agreements or CUAs, now often structured as Google Cloud Commit) can include provisions that reduce effective egress costs. Negotiating egress terms is an advanced tactic that requires volume — typically $500K+ annual GCP spend — to get Google's enterprise team to engage on custom terms.
At sufficient spend levels, Google's commercial team has discretion on several networking pricing elements:
Start by generating a detailed egress analysis from your billing export: break down egress by traffic type (inter-zone, inter-region, internet), by project, and by service. This data is your opening position — it demonstrates to Google's commercial team exactly how much egress you're generating and which workloads are driving the spend.
Frame the negotiation around total committed spend growth, not just current egress. Google's commercial team cares about growing your total GCP wallet share. If you can credibly show a migration roadmap that will double your GCP spend over the next two years, the egress negotiation becomes part of a larger deal structure where Google has more commercial motivation to be flexible.
Need help benchmarking your GCP egress costs and negotiating cloud network pricing?
See our related GCP CUD negotiation guide and GCP credits negotiation guide for complementary tactics on structuring your overall Google Cloud commercial agreement. For the broader cloud negotiation picture, our enterprise cloud discount negotiation guide covers AWS EDP, Azure MACC, and GCP Commit structures side-by-side.
Download our Cloud Contract Negotiation White Paper for a comprehensive framework covering commitment structures, egress provisions, and contract terms across all three major hyperscalers.
Connect with an independent GCP cost expert who can benchmark your egress spend, identify architectural quick wins, and negotiate better network pricing in your cloud agreement.