Massive AWS Glitch Knocks Popular Apps Offline — Recovery Underway

CNBC, CNN, the New York Times, and Bloomberg contributed to this report.

A bad Monday for the cloud turned into a rough morning for the internet. Amazon Web Services, the world’s largest cloud provider, stumbled overnight with what it called an “operational issue” that rippled through dozens of household-name apps and websites. The trouble kicked off around 2:01 a.m. PDT, when AWS acknowledged elevated errors across “multiple services” — nearly 70 by its own count — and said engineers were pursuing “multiple parallel paths” to fix it. By 3:03 a.m. PDT, Amazon was talking about “significant signs of recovery,” but warned some systems would lag while it chewed through a backlog of queued requests.

If you were locked out of your crypto wallet, couldn’t stream your shows, or your favorite game wouldn’t load, you weren’t alone. Outage tracker Downdetector lit up with reports tied to Amazon’s own properties as well as Snapchat, Disney+, Reddit, Canva, and more. Coinbase told users many couldn’t access accounts, while gamers flagged hiccups in Roblox and Fortnite. Canva said error rates spiked because of “a major issue with our underlying cloud provider.” Perplexity AI’s CEO, Aravind Srinivas, said bluntly the root cause was an AWS issue. Even airlines felt the wobble: some United and Delta customers griped about check-in and bag-drop snafus, though flight operations themselves held steady, according to early airport and FlightAware checks.

The hot spot appeared to be AWS’s US-EAST-1 region — its busiest — and the company later said global services and features dependent on that region had recovered. The guidance to customers was simple but unsatisfying: retry failed requests and expect extra latency while the backlog clears. Amazon eventually added that it had identified a root cause and continued to push toward full resolution.

If this all felt uncomfortably familiar, that’s because it is. The episode echoed last year’s CrowdStrike fiasco, when a faulty Windows update underscored how fragile the digital plumbing can be. AWS alone powers an enormous slice of the internet — roughly a third of global cloud infrastructure by some counts — and in the first half of this year accounted for about 60% of Amazon’s operating profit. When a platform that central hiccups, the effects move fast and wide, from banking apps and shopping carts to secure messaging and gaming lobbies.

Markets barely blinked — Amazon shares were little changed in premarket trading — suggesting investors see this as a nuisance, not an existential threat. But for users who woke up to spinning wheels and error pages, it was another reminder that the cloud’s promise of resilience still hinges on a few very big hubs working exactly as designed, exactly on time. Today, they didn’t — and the internet felt it.