It was just after 6:30pm GMT on February 20, 2026, when thousands of people across the United Kingdom and United States suddenly found themselves locked out of their favorite websites and apps. From sports betting platforms like Bet365 and Paddy Power, to food delivery giants such as Uber Eats and Just Eat, and even gaming services like Steam and Minecraft, a wave of digital silence swept across the internet. The culprit? A major outage at Cloudflare, the American internet infrastructure company that quietly powers much of the web.
According to both The Sun and The Telegraph, the outage was first detected by users who rushed to Downdetector, a platform that tracks online service disruptions. By 6:57pm GMT, over 5,200 reports had been filed about Bet365 alone, and Uber Eats had nearly 4,000 outage reports by 6:40pm. The problems weren’t limited to just a handful of platforms; the list of affected sites read like a who’s who of the digital world: Wikipedia, Sky Bet, PlentyOfFish, Workday, Dead By Daylight, Counter-Strike, ARC Raiders, Amazon Web Services, Dota 2, Overwatch 2, Call of Duty, Betfair, Uber, Sky Vegas, Microsoft 365, and more all suffered disruptions.
Cloudflare, which bills itself as a company that makes internet websites and apps "faster and more secure," quickly acknowledged the trouble. In a statement posted to its website, the company said: "We are investigating issues with our services and/or network. Users may experience errors or timeouts reaching Cloudflare’s network or services." Later, after identifying the root of the problem, Cloudflare clarified: "Cloudflare has identified impact to a subset of BYOIP prefixes. Cloudflare has identified and mitigated the underlying issue and is working to restore availability to impacted advertisements. Customers can self-mitigate by readvertising their prefixes in the dashboard."
For many, the outage was more than just an inconvenience. Social media was awash with frustrated users sharing their woes. One person wrote on X, "Imagine being hungry and Uber Eats app goes down." Another lamented, "So my Uber eats app just stopped working as my order is on its way." Gamers, too, were left in the lurch, unable to access their favorite titles or online services. One disgruntled Bet365 user vented, "Been down since 6pm, went to watch a race and been down since!" while another added, "Tried reinstalling it still no fix couldn’t even place a bet as the app crashed at 6, they should compensate all their customers for this problem imagine people with live bets at stake unfair."
Bet365, for its part, was quick to communicate with its customers. On X, the company posted: "We’re aware of an issue with our Website/App, and our Technical Team are working to resolve it as soon as possible. We apologise for the inconvenience." The message was echoed by a spokesperson who told The Sun, "We’re aware of an issue with our Website/App, and our Technical Team are working to resolve it as soon as possible. We apologise for the inconvenience."
While the outage was felt most acutely by end users, it also raised serious questions about the resilience of the internet’s underlying infrastructure. Jake Moore, a global cybersecurity adviser at ESET, put it bluntly to The Telegraph: "If a major provider like Cloudflare goes down for any reason, thousands of websites instantly become unreachable. The problems often lie with the fact we are using an old network to direct internet users around the world to websites but it simply highlights there is one huge single point of failure in this legacy design."
Cloudflare’s reach is enormous. Operating a vast network of servers across more than 330 locations in over 120 countries, the company connects some 13,000 internet networks. It claims that around 20% of all websites globally use its services in some form—whether directly or indirectly. That means when Cloudflare stumbles, the ripple effects are felt worldwide, as Thursday’s outage so clearly demonstrated.
The incident wasn’t the first of its kind. In fact, Cloudflare experienced similar disruptions in November and December of 2025, with the December event impacting major platforms like Zoom, Deliveroo, and LinkedIn. Each time, the outages have exposed the internet’s reliance on a handful of infrastructure providers, and the potential for widespread disruption when one of those providers encounters technical difficulties.
So what exactly went wrong this time? According to Cloudflare’s technical updates, the problem was tied to a subset of BYOIP (Bring Your Own IP) prefixes—a technical term referring to the way certain internet addresses are managed and advertised across networks. Cloudflare said it had "identified and mitigated the underlying issue" and was "working to restore availability to impacted advertisements." The company also advised that "customers can self-mitigate by readvertising their prefixes in the dashboard," a solution that, while helpful for IT professionals, likely meant little to the average user just trying to order dinner or place a bet.
For those caught up in the outage, the experience was a stark reminder of just how intertwined daily life has become with digital services—and how fragile that convenience can be. The list of affected platforms cut across industries: from online dating (PlentyOfFish), to productivity (Microsoft 365, Workday), to gaming (Minecraft, Dota 2, Overwatch 2), to essential cloud services (Amazon Web Services). Even the simple act of browsing Wikipedia or catching up on the news was interrupted for many.
Downdetector’s outage maps painted a vivid picture of the chaos, with dense clusters of complaints lighting up across the UK and the US. By 7:02pm GMT, there were 217 reports specifically about Cloudflare itself, and more than 5,500 reports for Bet365 by 7pm. The numbers likely understate the true impact, as not every affected user takes the time to file a report.
In the aftermath, the broader tech community is once again left to grapple with the implications. The internet, for all its complexity and reach, remains surprisingly vulnerable to failures at a handful of key chokepoints. As Jake Moore noted, "It simply highlights there is one huge single point of failure in this legacy design." While Cloudflare and other providers work tirelessly to shore up their systems and respond to incidents, the underlying architecture of the web continues to rely on central nodes—making resilience a pressing concern for businesses and individuals alike.
As Cloudflare restores normal service and companies like Bet365 and Uber Eats get back to business, the outage of February 20, 2026, will likely be remembered as another wake-up call. For users, it was a night of missed meals, lost bets, and interrupted games. For the industry, it’s a reminder that the backbone of the internet still has some serious growing pains to work through—and that the next outage might be just a click away.