An outage in Amazon Web Services (AWS) infrastructure caused widespread instability in Mercado Pago on Monday, October 20, 2025, impacting millions of users across Brazil. The digital payment platform, part of the Mercado Livre ecosystem, experienced disruptions in services like Pix, balance inquiries, and mobile app operations. The issue began around 4 a.m. Brasilia time due to a DNS failure in the DynamoDB API in the US-East-1 region in Virginia, United States, with effects spreading to Latin America.
Users reported difficulties completing transactions from early morning, with complaint peaks at 5:40 a.m. AWS identified the cause and began implementing fixes, restoring most services by noon. The incident underscored the reliance of financial platforms on shared cloud infrastructure.
Mercado Pago, with over 50 million active accounts in Brazil, faced rapid error propagation due to its deep integration with Amazon’s storage services.
- Pix errors accounted for 26% of initial complaints.
- Unavailable balances made up 28% of reports.
- Mobile app failures comprised 46% of cases.
Technical origin of the disruption
The failure stemmed from a domain name resolution issue in the DynamoDB API endpoint, increasing error rates in data requests. AWS teams applied mitigations, including DNS cache clearing, to address the underlying issue. Dependent platforms like Mercado Pago experienced recovery delays due to complex integrations.
The impact extended beyond payments, temporarily halting purchases and shipments on Mercado Livre. Global services in Latin America faced similar latencies, with partial recovery by 9 a.m.
Financial services affected by the outage
Mercado Pago automatically processed pending transactions after restoration, crediting balances within 24 hours for failed operations. Users of rival wallets like PicPay reported minor delays, but complaints surged 300% in the early hours. Banks like Stone and Nubank saw spikes in mobile banking errors, prompting reliance on traditional transfers.
E-commerce platforms integrated with Mercado Livre paused seller logins for about three hours, coinciding with a 25% increase in online sales volume in 2025 compared to the previous year. Stone activated redundancy protocols to mitigate losses in receivables advances.
User reports during the outage
Users shared experiences of blocked daily payments, such as bill settlements and in-store purchases using Mercado Pago’s virtual cards. Freelancers highlighted disruptions in online sales, with paralyzed platforms affecting critical deliveries.
A digital product seller noted that balance releases appeared processed but failed to complete, causing delays in transfers. Complaint volume exceeded normal levels by 300%, with a focus on morning access issues.
Comments indicated many turned to alternatives like traditional banks to bypass the issue. Full recovery occurred by 11 a.m., with tests confirming stability in subsequent processing.
Recovery measures implemented
AWS issued periodic updates via its status dashboard, recommending retries for failed requests and load distribution across multiple availability zones. Mercado Pago sent push notifications on restoration progress, prioritizing accounts with pending transactions.
Engineers adjusted shared API settings to prevent error propagation in future instances. Platforms activated redundancies, testing loads to avoid relapses, with a promise of a detailed report within 48 hours.
Alternative transaction options
During the outage, users opted for wallets like PicPay, which maintained operations with reduced delays, or traditional bank TED transfers. iFood, also affected, suspended delivery driver logins, impacting a sector with 25% order growth this year.
Monitoring tools aggregated real-time reports, helping isolate local connection issues from systemic failures. App updates were suggested for post-recovery compatibility.
History of similar cloud outages
AWS incidents have occurred annually, with durations ranging from minutes to hours in 2024, affecting streaming in South America for two hours. Mercado Pago prioritized security updates post-outages, restoring 95% of operations in under six hours in this case.
Internal reports indicate a focus on diversifying regions to reduce single-point dependencies. In 2024, a similar failure caused errors in e-commerce services due to latency propagation.
Full normalization ensured transaction continuity, with emphasis on resilience testing.
