Cloud region choice sets the list price of compute and storage, the egress you pay when workloads cross regions, and whether a workload is even legal to run in a given country. The same virtual machine can cost noticeably more in a small or remote region than in a large home region, and on every major provider the cheapest published rates tend to sit in the biggest North American regions while Asia Pacific, South America, and some European regions run higher. But the headline rate is only the start. A cheaper region that forces data to travel back to where it lives can cost more once cross region transfer is counted, and a region that breaches data residency rules costs you a fine rather than a saving. The decision that holds up is total delivered cost per workload, region by region, with egress and compliance priced in.
Here is how the economics actually break down, and the placement rules that turn region choice into a repeatable saving rather than a one off guess.
Why do cloud prices differ between regions?
A region is a cluster of data centres in one geography, and each one carries its own cost base: local power prices, land, construction, tax regime, and how much spare capacity the provider holds there. Providers pass that through as a per region price list, so identical resources carry different rates depending on where you place them. As a rule the largest, oldest regions are the cheapest, because scale and competition have driven the rate down, while newer or remote regions sit at a premium. On AWS the US East and US West regions anchor the low end; on Azure the same pattern holds with the large United States regions; GCP and OCI follow the same logic. The practical point for a buyer is simple: the region selector in a deployment template is a price lever, and most teams never treat it as one.
How big is the region price delta in practice?
The gap is workload specific, but it is large enough to matter. Across the four providers, premium regions commonly run 10 to 30 percent above the cheapest equivalent region for the same compute and storage, and certain constrained regions sit higher still. Multiply that by a steady production footprint and the difference is a real line on the bill. The mistake is to assume the delta is noise. For a stateless tier that can run anywhere, a 20 percent regional premium is pure avoidable cost. For a latency sensitive tier pinned near users, the premium may be the price of the product working at all. The skill is knowing which workloads are which.
A European SaaS company ran its entire platform, including stateless batch and analytics, in a single premium region chosen years earlier for proximity to one office. Modelling the workloads showed the batch and analytics tiers had no user facing latency requirement and could move to a lower cost region in the same regulatory zone. The latency sensitive application tier stayed put. Repricing only the movable workloads at the cheaper region's rates, net of the modest cross region transfer the split introduced, cut the relevant compute and storage spend by roughly a fifth. The figures are verified against billing data and anonymised. The saving came not from a discount but from placing each workload where its economics, not its history, pointed.
What does cross region data transfer add?
This is the cost that turns a clean looking move into a loss. When a workload in a cheaper region talks constantly to data that lives in another region, every byte that crosses the regional boundary is billed as data transfer, and on the hyperscalers that egress is not cheap. A compute tier moved to save 20 percent on instance rates can hand the saving straight back if it now pulls terabytes a month across regions to reach its database. The rule is to move compute toward data, not data toward compute, unless the data itself is being relocated. Note also that OCI prices egress materially below the hyperscalers, which changes the maths for transfer heavy designs and is a legitimate input to a multicloud placement decision. Always model the transfer the new topology creates before you commit to the move.
When does latency or compliance override the cheaper rate?
Two forces can veto a cheaper region. The first is latency. A user facing tier placed far from its users either delivers a worse experience or forces you to overprovision to compensate, and overprovisioning erases the regional saving. The second is data residency. Regulated data, personal data under regional law, or contractual residency commitments can make a region legally off limits regardless of price, and the cost of getting that wrong dwarfs any rate saving. Both belong in the model as hard constraints, applied before price. Region choice is a constrained optimisation: satisfy latency and residency first, then minimise cost within what remains.
A placement rule set you can apply this week
Sort every workload into three buckets. Latency bound tiers stay near users and are priced wherever that lands. Residency bound data stays in its required jurisdiction. Everything else, the stateless, the batch, the analytics, and the non production environments, is free to chase the lowest delivered cost, which means the cheapest region in the permitted set net of any transfer the placement creates. Score each candidate region on instance rate, storage rate, and the egress the topology would generate, then pick the lowest total. Re run the exercise when traffic patterns or provider pricing change. Done once across a real estate, this typically frees a meaningful share of spend on the movable tiers without touching architecture.
Frequently asked questions
Why do cloud prices differ between regions?
Does moving a workload to a cheaper region always save money?
What is the biggest hidden cost in region choice?
Put a region strategy on your estate
We help enterprises model region choice across AWS, Azure, GCP, and OCI, pricing in egress, latency, and residency so each workload lands where its economics point. We take zero provider commissions and answer only to you. Our guarantee: we reduce your cloud spend or we reimburse our service fee, on either a Fixed Fee or a no risk Gainshare basis. Book a strategy call to scope it for your footprint, and follow more in The Cloud Spend Navigator.
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.