5 MIN READ
Cloud storage egress in 2026 still surprises the budget. The hyperscalers charge $0.05 to $0.12 per gigabyte for the data leaving the region, the typical mid market enterprise moves between 100 and 400 terabytes a month, and the line item lands at $5,000 to $50,000 a month before anyone asks a question. The shock is not the rate. The shock is the rate times the volume the platform team never measured.
The market has a shape. AWS, Azure, and Google Cloud charge an egress fee to keep workloads from roaming. Cloudflare R2, Backblaze B2, and Wasabi charge zero for egress and recover their margin on the storage side. R2 lists storage at $0.015 per gigabyte per month against S3 at $0.021 to $0.023, so the trade is real. B2 charges $0.006 for storage and $0.01 for egress, which beats R2 at smaller scale, with the catch that the feature set is narrower and the regional footprint is smaller. The math depends on how much data leaves, how often, and how much of it is sitting cold anyway. Most procurement teams run the comparison once at contract renewal and never update it, which is the part that costs the most.
Where the egress actually comes from
Four sources account for the bulk of the line item at most enterprises. Backup traffic, where the production cloud snapshots into a different cloud or a different region, is the single largest contributor for organisations that run cross cloud disaster recovery. Analytics traffic, where the production warehouse exports to Snowflake, Databricks, or BigQuery in a different provider, runs as a daily or weekly transfer of tens to hundreds of terabytes for any company that keeps the warehouse outside the production cloud. Partner and regulator sharing, where data flows to a customer, a vendor, or a supervisory body on a contract schedule, is a smaller line for most buyers and a much larger one for financial services and healthcare. The one off migrations, the project where the data centre moves to a new cloud, land as a single event in the middle of the project and are easy to budget because they are visible, but the recurring paths underneath are the ones that compound.
What the typical buyer has tried
Three moves show up over and over, and all three deliver less than the buyer expects. Negotiation is the most common one, and it is real only at the top of the spend curve. AWS, Azure, and Google Cloud will discount egress for customers who spend north of a million dollars a year on the platform, with documented cuts of 30 to 60 percent for the largest accounts. For everyone below the line, the answer is a polite no. Compression, applied before the transfer, saves 30 to 50 percent on bandwidth and the egress cost falls by the same ratio, but the engineering effort to put compression on every backup, every export, and every partner feed is significant, and most teams compress the easy 20 percent and leave the rest. Multi cloud, on the theory that spreading the data across two providers will let the buyer shop the price, works on paper and fails in practice, because the operational cost of keeping two clouds consistent eats the egress saving within a year for most mid market workloads. None of the three moves is wrong. All three overstate what they deliver.
How to actually reduce the bill
Three moves that work, in roughly the order of how much they cost to land. Audit what is actually moving, because the data the team assumes is being transferred is rarely the data the network is actually transferring. A 30 day packet capture on the egress gateway, run through a tool like CloudHealth or Kubecost, will surface workloads the platform org did not know were syncing, and a meaningful share of them will turn out to be dev environments, log shippers, or backup jobs that nobody owns anymore. Match the tier to the access pattern, because the cold tier on S3, Azure Blob, and GCS has a lower egress rate than the hot tier on the same provider, and a one time job to move aged datasets from hot to cold pays back in months. Negotiate the enterprise discount at the right spend level, because once the account crosses the threshold the discount kicks in automatically on the next renewal, and the procurement lead has more pull than the engineering lead on this one. The three moves together cut the egress bill in half at most buyers that have not done them yet, which is still the majority of the market.
The bottom line
Cloud storage egress in 2026 is a measurement problem before it is a price problem. Audit the traffic, move the cold data to the cold tier, and renegotiate the enterprise discount when the account clears the threshold. The buyer who treats egress as a line item to track each month will save more than the buyer who tries to architect around it. The line item is not the enemy. The line item nobody owns is.
Sources & Further Reading
All claims in this article are sourced from primary documentation, vendor advisories, and reputable security researchers.
Spotted an error? Email the editor. Corrections are issued with a visible correction note.
Editorial standards. Every article on humanrequired.org is reviewed by a human editor before publication. AI may assist with drafting or research; final editorial control is human. Read the full standards.
