On June 12, 2025, Cloudflare suffered a major service outage lasting 2 hours and 28 minutes. This impacted critical services and users worldwide, including Workers KV, WARP, Access, Gateway, Images, Stream, Workers AI, Turnstile and Challenges, AutoRAG, Zaraz, and parts of the Cloudflare Dashboard. 

The incident was caused by a failure in the Workers KV service’s storage infrastructure, which supports configuration, authentication, and asset delivery across Cloudflare’s ecosystem. The root cause was an outage at a third-party cloud provider. Although an external fault, Cloudflare acknowledged full responsibility for the architectural decision to rely on the provider. 

Furthermore, there was no data loss or evidence of a cyberattack. Key services such as Magic Transit, Magic WAN, DNS, caching, proxy, and WAF remained operational during the incident.

Workers KV Centralized Dependency Exposed

Workers KV is designed as a “coreless” service, operating independently in every Cloudflare data center to avoid centralised points of failure. However, the system still depends on a data store for operations like cold reads and writes. This became the single source of failure during the outage. 

At the time, Cloudflare was transitioning Workers KV to a more resilient architecture using its R2 storage platform. This will improve data residency support, and eliminate dependency on the third-party storage provider. Unfortunately, the migration was still underway, and the resulting infrastructure gap was exposed by the outage. This had a cascading effect across the platform because many Cloudflare products rely on KV.

Immediate Remediation and Infrastructure Reforms Underway

The company’s engineers acted fast during the outage, backfilling critical KV namespaces into its infrastructure. This aimed to reduce impact if the outage continued. Afterwards, Cloudflare accelerated its existing resilience initiatives. Hence, products like Access, Gateway, and WARP can operate independently from Workers KV in the event of future failures. The company is also developing tools to allow controlled re-enablement of KV namespaces, reducing the risk of system strain during service recovery.

Commitment to Service Resilience

Cloudflare acknowledged that the immediate cause was external, but responsibility lies in their architectural decisions. The company remains committed to building most of the infrastructure on its platform, limiting exposure to external risks. 

In an official statement, Cloudflare apologised for the disruption and promised to improve reliability. The company stated that it is taking aggressive steps to avoid a repeat incident. This will help rebuild trust with customers and partners depending on its global network.