This Friday, December 5, 2025, Cloudflare recorded a global outage that affected the company’s control panel and APIs. The problem started at 8:56 am UTC (5:56 am Brasília time) and lasted for about two hours. Serviços like Zoom, Shopify, Notion and Medium presented access errors and slowness in different regions.
The failure resulted from internal degradation during scheduled maintenance in data centers in Estados Unidos. Esta is the second such incident in less than three weeks, following the November 18 incident.
- Zoom recorded error spikes in logins and calls;
- Shopify faced blocks on checkout pages;
- Notion and Medium displayed error 500 messages.
Technical origin of the failure
Engineers at Cloudflare identified the issue in data centers at Chicago and Detroit. Manutenções scheduled between 7am and 11am UTC coincided with an error in the Cloudflare Workers scripts.
The combination generated degradation that reached approximately 20% of the traffic managed by the company. Correções partials began at 9:12 UTC and restored 70% of operations within an hour.
Most affected services
Zoom has confirmed disruptions to active sessions on Europa from 9am CET (5am on Brasília). The platform recorded a 40% increase in error reports during the peak of the failure.
Shopify reported blocking payment processes for around 45 minutes. Lojas in Estados Unidos and Canadá were the most impacted during morning business hours.
Notion presented page loading and collaborative synchronization failures. Usuários on Ásia and Oceania faced almost total unavailability between 2pm and 4pm local time.
Recovery and immediate measures
Technicians activated alternative routes in European and Asian data centers at 9:20 UTC. The redirect covered 60% of affected traffic in less than 20 minutes.
The company isolated the problem scripts and expanded capacity on nodes Chicago and Detroit. Normalization reached 95% of domains at 11am UTC (8am Brasília).
Customers received direct alerts on the dashboard with guidelines for using local cache. Desenvolvedores were advised to implement automatic retries on API calls.
Recent incident history
On November 18, 2025, a bug in bot rules caused a four-hour outage. At the time, platforms such as Discord and ChatGPT were inaccessible for up to 90 minutes.
Another episode occurred in September, with 5% of traffic slowing down due to overload on edge servers. The three incidents add up to more than seven hours of partial degradation throughout the year.
Cloudflare currently processes traffic from around 20% of global websites. The monthly volume reached 100 exabytes in 2025, an increase of 30% compared to 2024.
Expansion and investments in infrastructure
The company operates in 310 cities and covers 95% of the world’s connected population. In 2025, 50 new points of presence were added to reinforce the network.
Investments of US$500 million expanded capacity in edge computing. The measure reduced average latency by 40% for users in the Ásia-Pacific region.
Partnerships with local operators increased coverage on América Latina, including new nodes on Bogotá and Santiago. Atualizações in edge encryption maintains protection for 99.99% of connections.
Monitoring and next actions
Cloudflare plans a full report in the next 24 hours on its official blog. The document will detail root cause and adjustments implemented after the incident.
Automated testing in dynamic configurations will be expanded later in December. The company reinforces real-time monitoring via status.cloudflare.com for future incidents.