A massive outage in Cloudflare's network, a key provider of cybersecurity and content delivery services for the internet, left dozens of popular websites and applications inoperable worldwide on Tuesday.
The company, which protects and accelerates 20% of the world's web traffic – including banks, governments, and e-commerce – had announced sporadic maintenance throughout the day, but a failure in server response amplified the problem. The incident affected millions of users, preventing access to platforms like X (formerly Twitter), OpenAI's ChatGPT, League of Legends, and other essential services.
The company confirmed it is investigating an 'anomaly' during scheduled maintenance that generated server errors and blocked content loading in regions across the Americas, Europe, and Asia. The impact was felt immediately on social media and digital tools. Initially confined to technical support, the incident escalated due to an 'undisclosed combination of factors,' according to cybersecurity experts.
The magnitude of the outage highlights the fragility of modern digital infrastructure. Cloudflare, founded in 2009 and headquartered in San Francisco, handles petabytes of data daily and defends against cyberattacks, but past incidents – such as the one in February 2024 due to a configuration error – have exposed vulnerabilities.
In Europe, EU regulators are already demanding explanations under the Digital Services Act, while in the U.S., the FTC is monitoring potential impacts on consumers. Frustrated users turned to alternatives: Telegram and Discord for chatting, or browsers in incognito mode, though with limited success. Cloudflare updated at 10:00 ET that 70% of services were recovering, but residual spikes persist in Asia. In parallel, the Spanish news aggregator Menéame showed internal errors directly linked to the failure, leaving its community without access to its usual discussions.
The entertainment and video gaming sector did not escape the chaos. Downdetector analysts estimate at least 10 million direct users affected, with ripple effects in digital economies: freelancers lost hours of work, players abandoned tournaments, and outlets like The New York Times reported glitches on their sites.
'It's like the internet went back to the 90s, with eternal dial-up,' quipped a user on Twitter from Madrid. 'Thousands of players kicked out in the middle of ranked games,' complained a Twitch streamer from Los Angeles. On X, thousands of users reported failures to view posts, update feeds, and play multimedia like videos and images, especially on the desktop version.
'We are aware of an issue impacting multiple customers, causing errors in the normal operation of websites and apps,' stated Cloudflare on its status dashboard, promising a resolution 'as soon possible.'
Other services like Spotify suffered streaming pauses, with users unable to load playlists, while Canva – a graphic design tool – blocked collaborative editing, impacting remote teams worldwide. Additional services such as Letterboxd (for movie reviews), Perplexity (AI search engine), Uber (in some regions for trip tracking), and forums like Forocoches and Resetera also went down, evidencing the global reliance on Cloudflare.
OpenAI, which uses Cloudflare for its distribution network, experienced interruptions to its APIs, affecting not only the main chatbot but also integrations in business applications.
This event, the largest since the Fastly collapse in 2021, underscores the need for diversification: experts recommend companies have backups with providers like Akamai or AWS. As the company advances in its investigation, the digital world holds its breath. In an era where a click moves fortunes, failures like this not only disrupt routines but also question the resilience of the global network. By the time the lights come back on, the debate on cloud monopolies will gain urgency.
League of Legends, Riot Games' battle royale with over 180 million monthly players, experienced massive disconnections: servers in North America, Europe, and Oceania reported 'high latencies' and complete outages, frustrating ongoing matches and matchmaking. 'Impossible to load anything, seems like a total blackout,' tweeted a user from New York, while in London, another described 'white screens and 503 errors' – the typical code for server overload on Cloudflare.
The monitoring platform Downdetector, ironically dependent on the same infrastructure, registered a peak of 50,000 global complaints in an hour, but soon collapsed itself, complicating real-time verification.
The wave of problems extended to artificial intelligence with ChatGPT, where users attempted queries but faced 'server unavailable' messages or eternal delays in responses.
The collapse, which began around 6:00-6:30 AM Eastern Time (US), paralyzed millions of users worldwide. 'I was in the middle of writing an important document and everything froze,' related a freelance professional from São Paulo.
