Amazon Web Services (AWS) said late Monday that it had repaired a large-scale outage that disrupted access to thousands of websites and apps for much of the day.
More than 1,000 platforms — including Snapchat and major banks such as Lloyds and Halifax — went offline after technical problems hit Amazon’s core cloud network in the United States. Downdetector, a global outage tracker, registered over 11 million user complaints during the disruption.
Technology analysts said the incident underscored how dependent the modern internet and global economy have become on a few powerful companies.
Millions cut off as online services collapse
Professor Alan Woodward of the University of Surrey said the failure highlighted the fragility of global online infrastructure. He warned that most companies rely on external providers to host their services. “Even the biggest cloud systems are not immune to mistakes, and small human errors can cause global consequences,” he said.
The first reports of problems appeared around 07:00 BST on Monday. Users across the world struggled to reach services like Fortnite and Duolingo.
By midday, Downdetector had logged over four million issues across 500 sites — more than double the usual weekday average. That figure later soared to 11 million as more platforms, including Reddit and Lloyds Bank, went down.
By 23:00 BST, Amazon announced that all AWS systems had returned to normal. The company admitted it had throttled parts of its own network while engineers worked to repair the core issue.
Experts suspect a chain of system failures
Mike Chapple, an IT professor at Notre Dame University, compared the disruption to a power grid collapse. He said systems may have flickered back online before breaking again. “It’s likely Amazon fixed the surface problem first before finding the root cause,” he said.
Amazon has not yet given a full explanation of what went wrong. In a brief update, it said the issue appeared to involve DNS resolution for the DynamoDB API in its US-EAST-1 region.
DNS, or Domain Name System, acts like a directory for the internet by converting website names into numerical addresses computers can understand. When it fails, browsers can’t locate sites, leaving users unable to connect.
Concentrated cloud power alarms industry experts
Cloudflare chief executive Matthew Prince said the outage exposed how dependent the world has become on a handful of companies. “Everyone has a bad day, and today it was Amazon’s,” he said. “Cloud systems allow rapid growth, but when one fails, millions of users are affected instantly.”
Cori Crider, head of the Future of Technology Institute, compared the event to a “bridge collapsing” in the digital economy. She said that roughly 70% of the world’s cloud computing depends on Amazon, Microsoft, and Google — a dominance she called “unsustainable.”
“When a few major providers falter, vast parts of the economy collapse with them,” Crider warned. She urged governments and corporations to invest in more local and diversified systems to prevent similar crises in the future.
Businesses urged to build stronger digital defences
Cornell University professor Ken Birman said part of the responsibility lies with companies that depend on AWS. “Too many businesses fail to design proper backup systems,” he said. He noted that outages are common but rarely this severe.
Birman explained that the technology to prevent such failures already exists. “We know how to build resilient and secure systems,” he said. “But many firms choose speed and convenience over reliability.”
Legal battles could determine accountability
The question of who bears responsibility may soon reach the courts. After last year’s CrowdStrike failure, Delta Airlines is still trying to recover more than $500 million in losses. Even after the fix, Delta had to manually restart 40,000 servers, causing several days of flight disruptions.
The AWS outage could now reignite debate over the concentration of power among a few tech giants — and whether this dominance leaves the global economy dangerously exposed to single points of failure.
