Cloudflare, one of the internet’s most important infrastructure providers, suffered another major outage on Friday, December 5, 2025, briefly knocking offline or slowing a huge range of websites and apps around the world. Users trying to reach services from social networks and AI tools to trading platforms and banks were met with “500 Internal Server Error” messages and stalled dashboards — just weeks after a similar global incident in November. [1]
The company says it has now implemented a fix and is monitoring systems, but the back‑to‑back outages are intensifying questions about how much of the modern web depends on a handful of central players such as Cloudflare.
How the December 5 Cloudflare outage unfolded
Reports of problems began early Friday, when BleepingComputer noted that “Cloudflare is down” and that many websites were crashing with 500 Internal Server Errors instead of loading normally. [2]
On Cloudflare’s own status page, the company acknowledged “issues with Cloudflare Dashboard and related APIs”, warning that customers using those interfaces could see failing requests and error messages. [3]
Key points in the timeline based on status updates and media reports:
- Around 08:56 UTC – Cloudflare posts that it is investigating problems affecting the Dashboard and APIs, but says its edge network and cached content delivery should continue to operate. [4]
- Shortly after – Users worldwide start seeing 500 errors on numerous sites. Even outage‑tracker Downdetector itself becomes unavailable for many visitors. [5]
- Late morning (around 09:00–09:15 UTC) – Monitoring data and media outlets report that Cloudflare has deployed a fix and that services are starting to recover, with error rates dropping as traffic stabilizes. [6]
- Early afternoon – Regional outlets citing Cloudflare’s status page say the fix is in place and the company has moved into a “monitoring” phase to ensure there are no lingering issues. [7]
Bloomberg later reported that Cloudflare said it had fixed the problem, which had temporarily taken down websites for several banks as well as services like Shopify, Zoom and LinkedIn. [8]
Which apps, websites and sectors were hit?
Because Cloudflare sits in front of millions of websites as a content delivery network (CDN), security layer and DNS provider, even relatively short disruptions ripple across the wider internet.
Across multiple regions and outlets, the following services were reported as impacted or slowed during the December 5 incident:
- Social & communication
- X (formerly Twitter), LinkedIn, Discord and parts of OpenAI‑powered services. [9]
- Video, creativity & productivity
- Zoom, Canva, Notion and other collaboration tools experienced errors or timeouts. [10]
- AI platforms
- ChatGPT, Claude and other AI assistants were intermittently unreachable, timing out, or returning server errors according to user reports compiled by global outlets. [11]
- Finance, trading & crypto
- Indian brokers such as Zerodha, Groww, Angel One and Upstox went offline during market hours; crypto exchange Coinbase and several digital banks also reported issues. [12]
- E‑commerce & retail
- Shopify‑powered stores and admin dashboards suffered intermittent failures, while sites like Etsy, Wayfair, Vinted and Ikea were also named among affected platforms. [13]
- Monitoring & infrastructure
- Downdetector — usually the go‑to destination for checking if “the internet is broken” — itself showed 500 errors for many users, a strong sign of how deeply the issue cut into Cloudflare’s own ecosystem. [14]
British tabloid and regional outlets went as far as to describe the event as “half the internet” going down — a rhetorical flourish, but one that reflects just how many high‑traffic sites were affected simultaneously. [15]
What went wrong? Cloudflare’s CTO points to logging changes and a React CVE
In a detailed explanation on X, Cloudflare CTO Dane Knecht stressed that the December 5 outage was not a cyberattack. Instead, he linked the disruption to a change made to help mitigate a newly disclosed vulnerability in React’s server components. [16]
According to the Times of India and other outlets summarising his comments:
- Cloudflare engineers disabled certain logging features as part of their response to a React CVE (a security vulnerability in React’s server‑side components).
- That change had unintended side effects on the availability of parts of Cloudflare’s network.
- Once the impact was understood, Cloudflare rolled back and added further mitigations, restoring services and pledging a fuller technical blog post on the incident. [17]
This fits with Cloudflare’s public status line for the incident, which framed the issue as affecting the Dashboard and APIs, not the core edge network that serves cached content. In practice, though, many complex apps that rely on real‑time API calls or dynamic content still saw 500 errors and login failures while the disruption was underway. [18]
A replay of November’s chaos — but with a different trigger
Friday’s outage is especially sensitive because it came less than three weeks after a separate, large‑scale Cloudflare incident on November 18, 2025. That earlier event also produced widespread “widespread 500 errors” and knocked out sites such as X, ChatGPT, Spotify, Canva, Shopify and many others for hours. [19]
Cloudflare has already published a detailed post‑mortem on the November 18 outage:
- A database permissions change caused a feature file used by Cloudflare’s Bot Management system to suddenly double in size.
- The oversized configuration file was pushed to machines across the network, exceeding built‑in limits and causing the proxy software responsible for routing traffic to crash.
- Because the bad file kept getting regenerated, the network repeatedly flipped between “failing” and “partially recovered” until engineers halted propagation and rolled back to a safe version. [20]
In its own words, Cloudflare called that November disruption its “worst outage since 2019”, acknowledging that for several hours its network was unable to route core traffic for a large share of customers. [21]
By contrast, the December 5 outage appears shorter in duration and tied to the React‑related logging change rather than Bot Management configuration files. However, from a user’s perspective, the symptoms — 500 errors, inaccessible dashboards, trading apps going down, and AI tools failing to respond — look strikingly similar.
Impact on banks, brokers and critical services
While outages at social platforms and entertainment apps tend to drive the loudest memes, Friday’s incident also exposed the reliance of financial and critical services on Cloudflare.
- Bloomberg reported that the disruption temporarily affected websites for several banks, along with Shopify, Zoom and LinkedIn. [22]
- Economic Times and Indian business media documented downtime at major stockbroking platforms including Zerodha, Groww, Angel One and Upstox in the middle of trading hours. Some brokers resorted to advising clients to use backup channels such as WhatsApp to manage orders. [23]
- The National and other outlets highlighted downtime for Coinbase and other fintech apps, illustrating how an issue in a single vendor’s network can ripple through payments and crypto markets. [24]
For banks and financial platforms, even a 15‑ to 30‑minute outage during peak times can cause:
- Failed or delayed trades
- Customer frustration and reputational damage
- Extra load on call centres and support teams
- Potential regulatory scrutiny if outages repeat
It’s no coincidence that analysts were already flagging reliability risks: just two days before this latest incident, an investment note pointed out that Cloudflare’s stock (ticker: NET) had fallen about 20.5% over 21 trading days, citing worries about network reliability after the November outage. [25]
Friday’s problems are likely to sharpen that conversation.
How big is Cloudflare’s footprint on the internet?
Outlets covering the December 5 outage repeatedly emphasised a now‑familiar theme: the internet has a concentration risk problem.
Estimates from infrastructure analysts and media reporting on recent outages suggest that:
- Cloudflare connects roughly 13,000 networks worldwide.
- Its services sit in front of around 20% of all websites, and an even higher share of the world’s top‑traffic sites. [26]
When a provider with that footprint hits trouble, it doesn’t just affect one sector. The December 5 incident simultaneously impacted:
- social networks
- AI tools and developer platforms
- e‑commerce and payment providers
- news sites and media companies
- government and transit‑related portals in some regions
That breadth is why even a relatively short‑lived disruption can “feel” like a much larger internet outage to ordinary users.
What Cloudflare and customers are doing now
By Friday afternoon, Cloudflare’s public communication and coverage from multiple outlets painted a picture of partial normalisation:
- Cloudflare’s status updates indicated that a fix had been implemented and services were operating normally, though engineering teams were continuing to monitor for elevated errors or latency. [27]
- Customers were advised that it was safe to re‑enable Cloudflare services that may have been temporarily disabled as a precaution during the incident. [28]
- CTO Dane Knecht promised a full technical breakdown in a blog post, continuing a pattern of relatively transparent post‑mortems from Cloudflare after major incidents. [29]
For developers, IT teams and site owners, the immediate steps are mostly reactive:
- Review error logs and analytics
Check for spikes in 5xx errors, failed transactions and retries during the outage window and quantify any business impact. - Communicate with users and customers
Many businesses now proactively explain when external infrastructure problems are to blame, rather than leaving users guessing about “site bugs.” - Stress‑test failover and redundancy plans
The past month has shown that relying entirely on a single CDN or DNS provider can be risky. Some organisations are now fast‑tracking multi‑CDN strategies or at least ensuring they have direct DNS or backup routing options. - Watch for the post‑incident report
Cloudflare’s deeper analysis of the December 5 event — especially around the React CVE and logging changes — will likely inform how other companies roll out security mitigations without compromising availability.
Why this matters beyond today: centralisation and resilience
Two massive Cloudflare incidents in under a month, plus other recent outages at hyperscale providers like Amazon Web Services and Microsoft Azure, are fuelling a broader debate: Has the internet become too centralised for its own good? [30]
From a resilience perspective, Friday’s outage underscores a few uncomfortable truths:
- Single points of failure are now global – A configuration mistake or rushed mitigation at one vendor can instantly affect banking, streaming, news, AI, gaming and transit systems across continents.
- Security vs. availability is a delicate balance – In this case, a change made to help address a React security vulnerability ended up impairing availability. Getting that trade‑off wrong, even briefly, can have real‑world consequences. [31]
- Market perception is shifting – Investors and customers alike are beginning to bake in “infrastructure outage risk” when evaluating vendors, as seen in Cloudflare’s recent share‑price volatility and heightened scrutiny of its network decisions. [32]
Yet despite these concerns, most businesses are unlikely to abandon providers like Cloudflare; the performance, security and cost benefits are too significant. Instead, the next phase of the conversation is likely to be about architecting around such providers: multi‑CDN setups, regional failover, better incident communications, and more rigorous change‑management practices.
Outlook: What to watch after the December 5 outage
In the coming days and weeks, key questions for Cloudflare customers, regulators and the broader tech ecosystem will include:
- Technical root cause:
How exactly did the React‑related logging change cascade into such visible user‑facing errors, and what safeguards will be added to prevent similar issues? [33] - Process improvements:
After November’s database‑driven outage, Cloudflare promised tighter checks on configuration rollouts. How will Friday’s incident reshape that roadmap? [34] - Customer risk management:
Will more high‑value organisations — especially in finance and government — invest in redundant providers so that a single Cloudflare issue can’t take their services offline? - Regulatory scrutiny:
As outages begin to hit critical infrastructure more frequently, regulators in some jurisdictions may revisit guidance around uptime, transparency and “systemically important” internet services.
For now, most affected sites appear to be back online, and the December 5 outage may ultimately be remembered as a short, sharp reminder of just how interconnected the web has become. But paired with November’s longer breakdown, it also feels like a turning point: a warning that the backbone of the internet needs not just more speed and security, but more resilience.
References
1. www.bleepingcomputer.com, 2. www.bleepingcomputer.com, 3. www.tradingview.com, 4. dataconomy.com, 5. www.bleepingcomputer.com, 6. www.red94.net, 7. www.dawn.com, 8. www.bloomberg.com, 9. www.thescottishsun.co.uk, 10. www.thenationalnews.com, 11. m.economictimes.com, 12. m.economictimes.com, 13. www.red94.net, 14. www.bleepingcomputer.com, 15. www.thescottishsun.co.uk, 16. timesofindia.indiatimes.com, 17. timesofindia.indiatimes.com, 18. dataconomy.com, 19. thenewstack.io, 20. blog.cloudflare.com, 21. blog.cloudflare.com, 22. www.bloomberg.com, 23. m.economictimes.com, 24. www.thenationalnews.com, 25. www.trefis.com, 26. www.thescottishsun.co.uk, 27. www.dawn.com, 28. timesofindia.indiatimes.com, 29. timesofindia.indiatimes.com, 30. www.red94.net, 31. timesofindia.indiatimes.com, 32. www.trefis.com, 33. timesofindia.indiatimes.com, 34. blog.cloudflare.com
