AWS Billing Bug Shows Some Customers Trillions in Phantom Charges

Quick Facts

• A bug in AWS’s estimated billing computation subsystem showed some customers phantom charges as high as $7.1 trillion starting July 16.
• No actual invoices, payment records, or real charges were affected — only the estimated figures displayed in the Billing Console and Cost Explorer.
• Amazon identified the root cause and began correcting billing data, with all customers expected to see accurate figures by July 18 at noon Pacific time.

Some Amazon Web Services customers opened their billing dashboards on Friday to find they owed billions — or trillions — of dollars for cloud services they had never used. Amazon confirmed the bug and said it traced the problem to a unit pricing error inside the estimated billing computation subsystem.

The issue began at approximately 7:38 p.m. PDT on July 16. By 1:33 a.m. PDT on July 17, AWS engineers had opened a formal incident response. They pinpointed the root cause by 3:03 a.m. PDT: an incorrect pricing rate applied to real usage data inside the forecasting engine that powers Cost Explorer and the Billing and Cost Management Console.

The scale of the phantom figures was extreme. One user whose AWS charges totaled $0.19 the prior month received an estimated bill of nearly $2.5 billion. A UK charity that normally pays less than one British pound per month saw a projected charge of roughly $7.8 billion. One account’s console showed an estimated $7.1 trillion owed since the start of the month — more than twice Amazon’s entire market capitalization. Another reported month-over-month growth of 745,728,201,771%.

AWS said on its Health Dashboard: “We have identified the root cause as an issue with unit pricing within the estimated billing computation subsystem. The displayed billing estimates do not reflect actual usage and charges.”

An initial rollback attempt did not resolve the issue. AWS paused estimated bill updates to stop the inflated figures from climbing further. The company then identified the root cause, mitigated the underlying issue, and began backfilling corrected data across accounts.

No real invoices or payment systems were touched. But the operational fallout was immediate. At several companies, engineers assumed a security breach before anyone confirmed the numbers were wrong. Teams began rotating access keys. Finance departments flagged the figures as real liabilities. Automated tools — including Budget Actions and Cost Anomaly Detection alerts — fired on the false data, requiring teams to audit every alert that triggered during the bug window.

“I nearly had a heart attack when my two S3 buckets with a few MBs of data generated a half-billion-dollar forecast,” wrote one user in the AWS subreddit. Another posted that they had deleted their entire account in a panic. Bill Radjewski, who runs CollegeFootballData.com and had never spent more than two cents in a month on AWS, was quoted billions in estimated charges.

APIContext CEO Mayur Upadhyaya pointed to a broader industry concern. “A fault that might once have affected a handful of organizations can now impact thousands simultaneously because so many businesses depend on the same infrastructure,” he said. “As digital infrastructure centralizes, the blast radius of individual failures grows.”

Cost estimation tools sit at the center of financial operations for most cloud-dependent businesses. They feed budget enforcement systems, compliance workflows, and executive reporting. Even a short-lived display error can force manual audits, delay financial closes, and raise questions about the reliability of cloud-native cost controls.

Amazon expected all customers to see corrected billing estimates by noon Pacific time on Saturday, July 18.

Read more: Amazon fixing bug that billed some AWS customers billions of dollars