Amazon Web Services (AWS) appears to have suffered a partial outage of its CloudFront content delivery network service.

According to the hyperscaler, increased "5xx" errors occurred between 12:45 AM and 4:18 AM PDT for CloudFront VPC Origins users.

The root cause appears to have been an internal constraint on the fleet that manages connections to private virtual private cloud (VPC) origins. When reached, the hyperscaler’s remediation team found the system responsible for distributing routing configuration to its network processors failed to load the updated configuration data correctly, affecting routing of VPC Origin connections.

AWS said the issue has since been resolved, and customers who temporarily changed their origin type can safely revert. Meanwhile, customers utilizing other origin types were not impacted.

The incident marks the first major outage for AWS since its mammoth web outage last October. In the time since, the hyperscaler has bolstered its domain name system (DNS) service while also unveiling an automated outage tool mere days after inadvertently taking down a good portion of the internet.

APIContext CEO Mayur Upadhyaya noted the incident itself wasn’t the story, but the “concerning trend” of provider consolidation, meaning that when there’s an outage, it impacts a greater number of users.

“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. We're seeing the same pattern across cloud platforms, identity providers, AI foundation models, and DNS,” Upadhyaya wrote in an email comment. “As digital infrastructure centralizes, the blast radius of individual failures grows. This isn't an argument against consolidation. The operational and economic benefits are enormous. But it does change the resilience conversation.

“Organizations need to understand not only whether a supplier is reliable, but how much of their business depends on that supplier, what the failure modes look like, and how critical workflows behave when those dependencies degrade," Upadhyaya continued. "The question isn't whether outages will happen. It's whether your most important transactions continue to work when they do."