Why is my AWS bill so high? The usual causes, and how to find them

Start in Cost Explorer

1. Billing and Cost Management → Cost Explorer. Set the granularity to Monthly, the range to the last three months, and Group by to Service. Find the service that grew.
2. Filter Service to that one and change Group by to Usage type. Usage types are the units AWS meters. For example, BoxUsage:t2.micro is running hours of t2.micro instances. The usage type usually tells you what kind of resource is costing money. In us-east-1 most usage types have no prefix; in other regions they start with a region code.
3. Group by Region as well. Spend in a region you don't think you use is worth a look.

With the CLI (each Cost Explorer API request costs $0.01):

aws ce get-cost-and-usage --time-period Start=2026-07-01,End=2026-10-01 \
  --granularity MONTHLY --metrics UnblendedCost \
  --group-by Type=DIMENSION,Key=SERVICE

aws ce get-cost-and-usage --time-period Start=2026-09-01,End=2026-10-01 \
  --granularity MONTHLY --metrics UnblendedCost UsageQuantity \
  --group-by Type=DIMENSION,Key=USAGE_TYPE

Cost Explorer data can lag by up to 24 hours, so very recent usage may not show yet.

The usual causes, and where they show up

The free checkup looks for every cause in this table.

The checkup also looks for three things without a guide yet:

• Large S3 buckets with no lifecycle rules. It lists Standard-class buckets of 100 GB or more with no lifecycle configuration, and marks its saving estimate low-confidence.
• Lambda functions on x86. In us-east-1, Arm compute is $0.0000133334 per GB-second against $0.0000166667 for x86, 20% less. The checkup flags x86 functions when Lambda compute cost $20 or more last month.
• No commitment discount. Compute Savings Plans cut EC2, Fargate and Lambda costs by up to 66%, AWS says. The checkup reads AWS's own 1-year, No Upfront recommendation for your account, scaled down for the compute its rightsizing findings would remove. For RDS, it reads AWS's reserved instance recommendation.

Also worth checking

The checkup mentions these as observations, without a savings estimate:

• Data transfer out to the internet. In us-east-1 it's $0.09 per GB for the first 10 TB a month, after 100 GB a month free across all regions. Look for DataTransfer-Out-Bytes. The checkup notes it when it cost more than $50 last month.
• A big jump month over month. The checkup notes growth of more than 15% (and at least $50). Cost Explorer's monthly view grouped by service shows where it came from.
• Support plans and supporting services. The checkup notes a support plan, AWS Config, Amazon GuardDuty, CloudWatch, AWS CloudTrail or AWS Security Hub that's more than 8% of the bill.

General advice the checkup doesn't cover: to catch the next spike sooner, turn on AWS Cost Anomaly Detection. It uses machine learning models on your Cost Explorer data, runs about three times a day, and alerts you by email or through an Amazon SNS topic. Because Cost Explorer lags, it can take up to 24 hours to detect an anomaly.

Sources

• Filtering and grouping data in Cost Explorer (Service, Region and Usage type dimensions; the BoxUsage example), Detecting unusual spend with AWS Cost Anomaly Detection (AWS documentation)
• Cost Explorer API, $0.01 per request: AWS Cost Explorer pricing; AWS Price List API, Cost Explorer offer for us-east-1
• Lambda (us-east-1): AWS Price List API, Lambda offer for us-east-1, version published October 1, 2026. Lambda-GB-Second $0.0000166667 and Lambda-GB-Second-ARM $0.0000133334 per GB-second, first pricing tier.
• Data transfer out (us-east-1): AWS Price List API, data transfer offer for us-east-1, version published September 16, 2026. DataTransfer-Out-Bytes $0.09/GB for the first 10 TB a month; 100 GB a month free, aggregated globally.
• Compute Savings Plans, up to 66%, covering EC2, Fargate and Lambda: Savings Plans FAQs
• Usage types for each cause: see the Sources section of each linked guide.

Prices are AWS on-demand list prices for US East (N. Virginia), checked October 3, 2026. Other regions differ.