AWS EC2 Pricing by Region: Where to Run for the Lowest Cost
Why EC2 Pricing Varies by Region
Not all AWS regions cost the same. The exact same instance type — identical vCPUs, RAM, and storage — can cost 10–50% more depending on where you run it. Three factors drive regional price differences:
- Infrastructure costs — Real estate, power, cooling, and construction costs vary dramatically. Building a data center in Stockholm is cheaper than building one in Singapore. AWS passes these costs through to pricing.
- Local demand — Popular regions with high utilization (like us-east-1) benefit from economies of scale, while smaller regions spread fixed costs over fewer customers.
- Regulatory & operational overhead — Some regions have higher compliance costs, local staffing requirements, or import tariffs on hardware that increase operational expense.
The cheapest region isn't always the best choice. Infrastructure cost is just one variable — you also need to consider latency to your users, data residency requirements, and cross-region data transfer fees.
Regional Price Comparison
Here's how On-Demand pricing compares across major AWS regions for three common instance types. All prices are per hour in USD.
| Region | t3.micro | m5.large | c6i.4xlarge | Premium vs us-east-1 |
|---|---|---|---|---|
| us-east-1 (N. Virginia) | $0.0104 | $0.096 | $0.680 | Baseline |
| us-east-2 (Ohio) | $0.0104 | $0.096 | $0.680 | +0% |
| us-west-2 (Oregon) | $0.0104 | $0.096 | $0.680 | +0% |
| eu-north-1 (Stockholm) | $0.0108 | $0.100 | $0.714 | +4–5% |
| eu-west-1 (Ireland) | $0.0116 | $0.107 | $0.762 | +11–12% |
| ap-northeast-1 (Tokyo) | $0.0136 | $0.124 | $0.878 | +29–31% |
| ap-southeast-1 (Singapore) | $0.0132 | $0.121 | $0.856 | +26–27% |
| ap-south-1 (Mumbai) | $0.0116 | $0.107 | $0.762 | +11–12% |
| sa-east-1 (São Paulo) | $0.0160 | $0.147 | $1.040 | +53–54% |
| me-south-1 (Bahrain) | $0.0143 | $0.132 | $0.935 | +37–38% |
The spread is significant: running a c6i.4xlarge in São Paulo costs over 50% more than in N. Virginia. For a single instance running 24/7, that's an extra ~$260/month.
Cheapest AWS Regions Ranked
For On-Demand EC2 pricing, the most affordable regions are consistently:
✅ Lowest-Cost Regions (On-Demand)
The three US regions are almost always tied for lowest price. If you're in Europe and need data residency, Stockholm is your best bet for cost. Ireland is slightly more expensive but offers the broadest set of AWS services in Europe.
Most Expensive Regions
Some regions carry a hefty premium due to higher operational costs or limited scale:
💰 Highest-Cost Regions
A modest fleet of 10 m5.large instances running 24/7 in sa-east-1 instead of us-east-1 costs an extra ~$3,700 per year. For larger deployments, the regional premium can represent tens of thousands of dollars annually.
Data Transfer Costs Can Negate Region Savings
Choosing a cheap region is only half the equation. If your application transfers data across regions — or to users far from that region — data transfer fees can wipe out your savings.
- Cross-region data transfer: $0.01–$0.02/GB between AWS regions. Moving 1 TB/month between us-east-1 and eu-west-1 costs ~$20/month.
- Internet egress: $0.05–$0.15/GB depending on region and volume tier. Serving 10 TB/month to end users costs $500–$1,500.
- Same-region transfer: Free between services in the same Availability Zone. $0.01/GB across AZs within the same region.
Run your compute close to your data. If your database is in us-east-1, running application servers in eu-north-1 to save on compute will cost more in cross-region data transfer than you save on instances. Co-locate compute and data in the same region, ideally the same AZ.
Latency Tradeoffs
A cheaper region is meaningless if it degrades your user experience. Physical distance directly impacts network latency:
| User Location | Nearest Region | Latency | Cheapest Region | Latency |
|---|---|---|---|---|
| New York | us-east-1 | ~5 ms | us-east-1 | ~5 ms ✅ |
| London | eu-west-2 | ~10 ms | us-east-1 | ~75 ms |
| Tokyo | ap-northeast-1 | ~5 ms | us-east-1 | ~170 ms |
| Sydney | ap-southeast-2 | ~5 ms | us-east-1 | ~200 ms |
| São Paulo | sa-east-1 | ~10 ms | us-east-1 | ~130 ms |
For API-heavy applications, each additional 100 ms of latency can degrade throughput and user experience significantly. For batch processing or background jobs, latency rarely matters — optimize for cost.
Spot Price Regional Variation
Spot pricing is driven by supply and demand within each region, and the differences can be dramatic. While On-Demand pricing is fixed, Spot prices fluctuate:
- Popular regions (us-east-1, eu-west-1) have higher Spot demand, which can drive prices up and increase interruption rates.
- Less popular regions (ca-central-1, eu-north-1, ap-south-1) often have surplus capacity, resulting in lower Spot prices and fewer interruptions.
- Price swings can be significant — Spot prices for the same instance type can differ by 2–3x between regions at any given moment.
✅ Spot Pricing Strategy
For batch workloads, CI/CD pipelines, and data processing, running Spot instances in a less popular region can cut compute costs by 80–90% compared to On-Demand in a premium region.
Decision Framework: Cost vs Latency
Use this framework to decide whether to optimize for the cheapest region or the closest one:
- Your workload is batch processing, ETL, or background jobs with no real-time user interaction
- You're running CI/CD pipelines, build servers, or testing environments
- Your application uses a CDN (CloudFront) for user-facing content, so origin latency is less critical
- Your data already resides in the cheap region
- You're running Spot workloads that can tolerate region flexibility
- You're serving real-time APIs, gaming, or trading applications where latency is critical
- Your users are concentrated in a specific geography and expect sub-50ms response times
- You have data residency or sovereignty requirements (GDPR, CCPA, etc.)
- Your application makes frequent database calls — co-locate compute with data
- Cross-region data transfer would cost more than the compute savings
Many organizations split workloads: user-facing services run in the region closest to customers, while batch processing, analytics, and CI/CD run in the cheapest region. Use CloudBench to compare pricing across regions and find the optimal split for your architecture.
Frequently Asked Questions
Which AWS region is cheapest?
us-east-1 (N. Virginia) is consistently the cheapest for On-Demand EC2 pricing. It was AWS's first region and benefits from massive scale. us-east-2 (Ohio) and us-west-2 (Oregon) match its pricing for most instance types. In Europe, eu-north-1 (Stockholm) is typically the most affordable, running just 4–5% above US pricing.
Does region affect Spot pricing?
Yes, significantly. Spot prices are set by supply and demand within each region. Less popular regions like ca-central-1 or eu-north-1 often have more spare capacity, resulting in lower Spot prices and fewer interruptions. Always check Spot pricing across multiple regions — the differences can be 2–3x for the same instance type.
Is us-east-1 always cheapest?
For On-Demand pricing, yes — us-east-1 is cheapest for virtually all instance types. For Spot pricing, no — regions with less demand may have lower Spot prices at any given time. And when you factor in data transfer costs, running in us-east-1 while your users or data are in another region can actually cost more overall.
Compare EC2 Prices Across Every Region
CloudBench shows real-time On-Demand and Spot pricing for every EC2 instance type across all AWS regions. Find the cheapest region for your workload in seconds.
Explore Instance Pricing →