Users heading for X, ChatGPT, DoorDash, IKEA and New York City’s Metropolitan Transportation Authority were greeted with Cloudflare-related complaints after an odd surge in traffic hit one of the company’s services at about 6.20 am ET. A Cloudflare spokeswoman said the fault was fully sorted by 9.30 am ET.

For several hours, the online world was a dog’s dinner, with retail, social media, finance, transport and e-commerce services either sputtering or refusing to load. Cloudflare’s shares dipped on Tuesday as the company scrambled to explain itself.

The fiasco looked rather like last month’s Amazon Web Services outage, although this one was smaller. These cloud outfits often remain invisible to users, but their kit props up a hefty slice of the modern internet.

AWS’s cock up last month took down swathes of sites for millions of people and last year a bug in a CrowdStrike security tool crippled computer systems worldwide, causing thousands of flight delays and playing havoc with government and corporate operations.

Check Point Software cybersecurity expert Graeme Stewart said the mess showed how dangerously dependent everything has become on a few infrastructure providers.

Stewart said, “Many organisations still run everything through one route with no meaningful backup. When that route fails, there is no fallback. That is the weakness we keep seeing play out.”

Cloudflare chief technology officer Dane Knecht apologised for the collapse: “We failed our customers and the broader Internet when a problem in @Cloudflare's network impacted large amounts of traffic that rely on us”

He then said the impact and sluggish fix were unacceptable and that work had already started to prevent a repeat.