In Brief: Cloudflare’s global outage
- On Tuesday, November 18, 2025, Cloudflare suffered a major outage that rippled across the internet and disrupted websites worldwide.
- The incident began around 11:30 UTC and hit e-commerce, banking, social media, and AI platforms.
- Cisco ThousandEyes pointed to a failure in Cloudflare’s backend services rather than a network path issue or attack.
- Many businesses were offline for hours, renewing concerns about the fragility of critical online infrastructure.
Cloudflare outage on November 18: What actually happened?
A widespread Cloudflare outage on November 18, 2025 knocked thousands of sites and apps offline for hours. Network telemetry from Cisco ThousandEyes flagged a surge in HTTP 5XX errors inside Cloudflare’s environment, which points to a backend services failure. In short, the network routes looked fine. The origin systems behind the edge were not responding.
The trouble started at roughly 11:30 UTC. Traffic flowed to Cloudflare’s edge, but requests often died on arrival. That left businesses in key sectors scrambling to explain outages to customers while engineers waited for services to recover.
How did the outage unfold in real time?
ThousandEyes observed the outage as global and service wide, paralyzing content delivery for many Cloudflare customers. Big consumer platforms felt it quickly. X, OpenAI, and Anthropic saw degraded performance or outright failures that blocked users from core features.
Cloudflare posted updates on its status page as teams worked through remediation. Recovery was gradual. For many customers, normal operations did not return for several hours.
Which businesses were hit hardest?
E-commerce took a direct hit. Retailers that rely on Cloudflare for DNS, CDN, and security saw a sharp drop in availability and conversions. As CNET reported, several large storefronts experienced total traffic loss during peak shopping windows, which translated to immediate revenue impact and a torrent of support requests.
One retail leader put it plainly to CNET. The first hour alone cost thousands in sales while customers could not load the site and support lines backed up. The moment echoed other large scale failures such as the 2021 AWS outage, a reminder that single points of failure still lurk in modern stacks.
Recommended tech
When tickets spike, keeping customers informed is half the battle. AI assistants can deflect common questions so human agents focus on tougher issues. If you need an easy on ramp, Tidio offers a capable chatbot and inbox that can help during incidents.
What was the impact on banks and financial services?
Financial institutions that use Cloudflare for secure web access also reported disruptions. Some online banking portals and mobile apps timed out, which locked users out of routine tasks.
According to the BBC, several major banks in Europe and North America faced temporary outages that slowed or blocked customer access. One European bank said it shifted users to alternate channels, but delays and confusion were hard to avoid. The episode adds to a growing body of cautionary tales, much like Australia’s Optus outage, about the fragility of essential digital services.
Were social and AI platforms affected too?
Yes. X struggled to serve timelines for many users, which quickly drove conversation about the outage onto other networks. Meanwhile, AI providers OpenAI and Anthropic reported degraded performance and intermittent errors. The takeaway is familiar. Even cutting edge platforms share the same underlying plumbing as everyone else, and when that foundation stumbles the effects cascade.
Get the latest tech updates and insights directly in your inbox.
How did Cloudflare respond and what was the root cause?
Cloudflare posted regular status updates during the event and later explained that the root cause stemmed from backend services inside its platform rather than a network routing issue or a malicious DDoS campaign. That distinction matters. The internet could reach Cloudflare. The failing components were the ones meant to process and serve requests.
What should businesses take away from this?
This outage is a stark reminder that even trusted providers can have bad days. Teams should assume dependencies will fail and design for resilience. That can mean multiple DNS and CDN options, more aggressive edge caching for critical pages, circuit breakers, and clear customer messaging. The broader concern echoes themes we have covered before about the dangerous reality of big tech dependence.
Recommended tech
Measure what matters during an incident. With Databox, teams can track uptime, conversion rates, support volume, and estimated revenue loss in one place. Those metrics help steer recovery and justify future resilience work.
Quick playbook to reduce outage risk
Start by mapping your dependencies, then add failover paths and graceful degradation. Practice incident drills so your team can switch modes fast. Finally, quantify impact so you can make smart tradeoffs on where to invest next.
FAQ
Was the Cloudflare outage a cyberattack?
No. Telemetry and analysis from Cisco ThousandEyes point to a backend services failure inside Cloudflare, not an attack or a routing meltdown.
When did the outage start and how long did it last?
It began around 11:30 UTC on November 18, 2025 and lasted several hours, with recovery arriving in stages for different customers.
Which services were affected?
A wide mix of e-commerce sites, banking portals, social platforms like X, and AI providers including OpenAI and Anthropic experienced disruptions.
How can businesses prepare for similar outages?
Adopt multi vendor DNS and CDN, cache critical pages at the edge, build read only modes, implement clear status communications, and rehearse failover.
Is Cloudflare still safe to use?
Cloudflare remains a widely used provider with strong performance overall, but no platform is immune to failure. The right move is to design for failure and reduce single points of dependency.
1 comment