To reduce Cloud Storage egress and operation costs on GCP, split the bill into egress, operations, and storage; co-locate compute with the bucket so reads stay in-region; put a CDN in front of hot reads; batch and cache small requests to cut Class A and Class B operations; and match the storage class to real access frequency. Internet egress runs around 0.12 US dollars per GB and is usually the largest line, while chatty small-object workloads quietly rack up per-operation charges. The rule: attack egress and operations before storage-at-rest.
Last updated: June 2026. Written by Fredrik Filipsson and reviewed by Morten Andersen, built on our See, Cut, Lock, Run method.
This article sits in our complete guide to Google Cloud cost optimization, the cluster pillar it links up to, and is part of the broader cloud cost optimization playbook. Cutting storage and network waste is a Cut step move. Because the network tier you serve from changes the egress rate, pair this with our companion Network Service Tiers Premium vs Standard cost comparison.
Attack egress and operations before storage-at-rest. The per-GB storage rate is rarely what blows up a Cloud Storage bill; cross-region and internet egress, plus high-volume small-object operations, almost always are. Co-locate readers with the bucket, cache hot reads behind a CDN, and batch small requests. The expensive mistake is shaving the storage rate by moving data to a colder class while leaving the egress and operation patterns that actually drive the cost untouched.
How much does Cloud Storage egress cost on GCP?
Cloud Storage egress to the public internet is on the order of 0.12 US dollars per GB, while reads from compute in the same region as the bucket are free or low cost, per the Cloud Storage pricing documentation. That gap is the whole game: the same object read by an in-region service costs almost nothing and read by a user over the internet costs real money per GB. Inter-region reads sit in between. This is why egress, not storage-at-rest, is usually the largest and most surprising part of a Cloud Storage bill, and why co-location and caching are the first levers. Confirm the current per-GB egress rates for your regions in that documentation, because Google Cloud network pricing changes.
What are Class A and Class B operations, and what do they cost?
Class A operations are writes and lists, such as inserting an object or listing a bucket; Class B operations are reads, such as getting an object. On Standard storage, Class A operations cost about 0.05 US dollars per 10,000 after a free monthly tier of 50,000, and Class B operations about 0.004 US dollars per 10,000, per the Cloud Storage pricing documentation. Those rates look tiny, but a workload that polls, lists, or reads millions of small objects makes millions of operations, and the cost adds up. Colder classes charge more per operation: Nearline Class A operations are around 0.30 US dollars per 10,000, an order of magnitude above Standard, which is why moving chatty data to a colder class can backfire.
| Line item | Typical rate | What drives it |
|---|---|---|
| Internet egress | About 0.12 USD per GB | Serving objects to users or other clouds |
| In-region reads | Free or low cost | Compute in the same region as the bucket |
| Standard Class A ops | About 0.05 USD per 10,000 | Writes and list operations |
| Standard Class B ops | About 0.004 USD per 10,000 | Reads (get object) |
| Nearline Class A ops | About 0.30 USD per 10,000 | Writes and lists on colder data |
The figures above reflect Google Cloud pricing structure as we read it in June 2026. Confirm exact rates for your region and storage class in the linked documentation, because prices change.
How do you cut egress and operation costs?
Cut them by separating the bill, co-locating compute, caching hot reads, batching small requests, and matching the storage class to access. The steps below are the Cut sequence from our method.
- Split the bill into egress, operations, and storage. Use the billing export to separate network egress, Class A and Class B operations, and storage-at-rest so you know which line to attack.
- Co-locate compute with the bucket. Keep readers in the same region as the bucket so internal traffic is free or cheap and you avoid internet and inter-region egress.
- Put a CDN in front of hot reads. Serve frequently read objects through Cloud CDN so repeat reads hit the cache instead of generating bucket egress and Class B operations.
- Batch and cache small requests. Combine many small objects, cache listings, and avoid per-object polling so you make fewer Class A and Class B operations.
- Match the storage class to access. Move genuinely cold data to Nearline, Coldline, or Archive with lifecycle rules, but watch the higher per-operation and retrieval costs on colder classes.
Does moving data to a colder class always save money?
No. Colder classes lower the storage rate but raise per-operation and retrieval costs and add minimum storage durations, so they only save money for data you rarely touch. Nearline, Coldline, and Archive cut the per-GB-month storage charge sharply, which is exactly right for backups, logs, and archives that are written once and read almost never. But each class also charges more per operation and adds a data-retrieval fee and a minimum storage duration, so frequently accessed data placed in a cold class can cost more than it did in Standard once the retrieval and operation charges land. Match the class to real access frequency, measured from the billing export, not to a guess. The full storage cost playbook lives in our Google Cloud guide.
Is egress or operations quietly running up your storage bill?
Our Google Cloud cost audit splits your Cloud Storage spend into egress, operations, and storage, finds the cross-region reads and chatty workloads, and quantifies the saving from co-location, CDN, and the right class. On the performance model, you pay only from realized savings. No savings, no fee.
Book a Google Cloud cost audit →How do you stop egress surprises?
Stop egress surprises by keeping reads in-region, caching hot objects, and never serving large public downloads straight from the bucket. Most egress shocks come from one of three patterns: a service in one region reading a bucket in another, an application serving large files to the public internet object by object, or an analytics job pulling a full dataset across regions repeatedly. Co-locating compute with the bucket removes the first, a CDN in front of public reads removes the second, and caching or regional replicas remove the third. Set a budget alert on the egress line specifically so a new cross-region reader is caught in days, not at month end. For the budgets and guardrails that keep this from drifting back, see the Lock step in our Google Cloud guide.
The Google Cloud Cost Optimization Field Guide includes the egress-profiling worksheet and the storage-class decision tree referenced above. It is the downloadable companion to this guide.
Frequently asked questions
How much does Cloud Storage egress cost on GCP?
Cloud Storage egress to the public internet is on the order of 0.12 US dollars per GB, while reads from compute in the same region as the bucket are free or low cost. Egress, not storage-at-rest, is usually the largest and most surprising part of a Cloud Storage bill. Confirm current per-GB rates in the Cloud Storage pricing documentation.
What are Class A and Class B operations?
Class A operations are write and list operations such as inserting objects and listing buckets; Class B operations are reads such as getting an object. On Standard storage, Class A operations are about 0.05 US dollars per 10,000 after a free monthly tier, and Class B operations about 0.004 per 10,000, so chatty, small-object workloads can run up real operation costs.
Does moving data to Nearline always save money?
No. Nearline, Coldline, and Archive lower the storage-at-rest rate but raise per-operation and data-retrieval costs and add minimum storage durations. They save money only for data you rarely access; for frequently read data the higher operation and retrieval charges can exceed the storage savings, so match the class to real access frequency.
How do I stop Cloud Storage egress surprises?
Co-locate compute with the bucket so reads stay in-region, put a CDN in front of frequently read objects, and avoid serving large files to the public internet directly from the bucket. Most egress surprises come from cross-region reads or public internet serving that a CDN or co-location would have absorbed.
The short version
On a Cloud Storage bill, attack egress and operations before storage-at-rest: internet egress runs around 0.12 US dollars per GB, and small-object workloads rack up Class A and Class B operations. Co-locate, cache, batch, and match the class to access. When you want your storage spend split and the egress drivers found, that is what our Google Cloud cost optimization service delivers.
Cloud pricing and service behavior change frequently. Verify the specifics in this guide against the providers’ own current documentation and the FinOps Foundation: Google Cloud pricing ↗, Google Cloud documentation ↗ and FinOps Foundation Framework ↗. This article also reflects Cloud Cost Room’s hands-on, vendor-neutral engagement experience.