A Single Fiber Cut Knocked X, Reddit and Teams Offline for Millions
A severed fiber line in Eastern North America rippled through Cloudflare on June 22, briefly taking dozens of major services down.
On June 22, 2026, large parts of the internet stuttered at the same moment. X, Reddit, Microsoft Teams, Zoom, Discord, Canva and even Amazon Web Services flashed errors for users across North America and beyond. The cause was not a clever cyberattack or a botched software push. It was a physical break in the cables that carry the web.
Quick answer
On June 22, 2026, a single severed fiber line in Eastern North America cascaded through Cloudflare, which fronts roughly a fifth of global web traffic, briefly knocking X, Reddit, Microsoft Teams, Zoom, Discord, Canva, and parts of AWS offline. Cloudflare saw elevated errors around 13:35 UTC and traced the cause to the fiber cut by 14:37 UTC; most services recovered the same afternoon once traffic rerouted. It was not a cyberattack, and restarting your router would not have helped, the failure was upstream and physical.
Key takeaways
- A severed fiber line in Eastern North America triggered cascading failures at Cloudflare, which handles roughly a fifth of global web traffic.
- DownDetector logged spikes for X, Zoom, Reddit, Canva, AWS, Microsoft Teams, Fortnite and Discord within minutes.
- Cloudflare engineers first saw elevated error rates around 13:35 UTC and traced the root cause to the fiber cut by 14:37 UTC.
- Some services recovered within about 20 minutes; a separate scheduled Newark datacenter maintenance overlapped in time but was not the cause.
- The episode is a reminder that much of the cloud still rests on a finite number of physical cables.
What happened
Cloudflare sits in front of a huge slice of the web, filtering traffic, blocking attacks and speeding up page loads. When its network has a bad day, the effects spread far beyond its own customers. That is what users felt on June 22. Around 13:35 UTC, Cloudflare's monitoring picked up climbing error rates and latency. Roughly an hour later, engineers identified the culprit: a cut fiber line somewhere in Eastern North America, a literal break in the underground or undersea strands that move internet traffic between regions.
Note
A fiber cut is exactly what it sounds like: the glass strands that carry data are physically severed, often by construction work, a ship anchor or damaged conduit. Traffic must then reroute over remaining paths, which can overload them and cause errors far from the break itself.
DownDetector, which tracks user-reported outages, lit up with complaints for a long list of platforms. The pattern was telling. The affected services had little in common except that many of them lean on Cloudflare or shared network paths in the region. When those paths degraded, the failures fanned out across unrelated apps at once.
Here is the rough timeline of how the incident unfolded, which is useful both for understanding the scale and for calibrating your expectations the next time several apps fail together:
| Time (UTC) | Event |
|---|---|
| ~13:35 | Cloudflare monitoring detects climbing error rates and latency |
| ~13:40 | DownDetector spikes for X, Reddit, Zoom, Teams, Discord, Canva, Fortnite |
| ~13:55 | First services begin recovering as traffic partially reroutes |
| ~14:37 | Root cause confirmed as a fiber cut in Eastern North America |
| Same afternoon | Normal routing restored, most platforms back to normal |
Why it matters
Outages like this puncture a comforting illusion. People talk about "the cloud" as if it were weatherless and weightless, but it runs on concrete data centers and bundles of glass fiber strung across continents and oceans. A single damaged line, in the wrong place, can briefly disconnect millions of people from services they treat as always-on.
It also highlights how much consolidation now shapes reliability. A handful of providers, Cloudflare among them, sit in the critical path for an enormous share of the web. That concentration brings real benefits, including strong security and fast performance, but it also means one provider's bad day becomes everyone's bad day. The same concentration risk shows up in software supply chains, as our breakdown of the Klue Salesforce OAuth supply-chain attack makes clear: when one trusted node fails, the blast radius is everyone downstream.
What is next
In the short term, Cloudflare restored normal routing and most services recovered the same afternoon. The longer story is about resilience. Expect renewed attention to a few areas:
- Path diversity. Networks will review whether enough independent fiber routes exist to absorb a major cut without cascading failures.
- Faster rerouting. Providers will look at how quickly traffic can shift away from a damaged segment before errors pile up.
- Transparency. Detailed post-incident reports help customers understand shared risk and plan their own redundancy.
- Physical security of cables. Construction near critical conduits and undersea landing sites remains a quiet but real vulnerability.
For everyday users, the practical takeaway is modest but worth keeping: when several unrelated apps fail at once, the problem is rarely your device or your home network. It is often something much larger, and usually fixed within the hour.
What to do during an outage
When a wave of apps goes dark at once, resist the urge to troubleshoot your own gear. Run this quick triage instead:
- Check whether the failure is widespread before touching settings: a status page or DownDetector tells you in seconds.
- If multiple unrelated apps are down together, assume an upstream provider, not your device or router.
- Do not factory-reset anything; a provider-side fiber cut cannot be fixed from your end.
- Switch to a different network (mobile data versus Wi-Fi) only to confirm the problem is not your local ISP.
- For anything time-critical, have an offline or alternate-channel fallback ready before the next outage, not during it.
The single useful lesson for businesses is path and provider diversity. Relying on one CDN or one transit route turns a regional cable fault into a total outage, while a second independent path turns it into a blip.
Frequently asked questions
Was this a cyberattack?
No. Cloudflare attributed the disruption to a physical fiber cut in Eastern North America, not a hack or malware. The damaged line forced traffic onto other paths, which caused the cascading errors.
Why did so many different apps go down together?
Many of the affected services rely on Cloudflare or shared network routes in the region. When those routes degraded, the failures spread across otherwise unrelated platforms at the same time.
How long did the outage last?
Some services began recovering within about 20 minutes of the first reports, and most were back to normal the same afternoon after traffic was rerouted.
What can I do during an outage like this?
Very little, and that is the honest answer. Restarting your router or device will not fix a provider-side fiber cut. Checking a status page or outage tracker is the fastest way to confirm the problem is not on your end.
The web held up better than it might have, but June 22 was a useful jolt: the internet is sturdier than a single cable, yet not nearly as abstract as we like to pretend.
