A status web page showed various parts of the Wikimedia network as suffering performance issues.
Wikipedia ruled out any suggestion of malicious intent being behind problem.
The two cables, which stretched between Tampa and Virginia, were broken for an hour and six minutes, the site said.
After the cables were repaired, it took another hour for basic service on Wikipedia to be restored.
Its mobile site appeared to be unaffected, although the service’s API – application programming interface – continued to suffer problems even when the main site had been restored.
David Gerard, a UK spokesman for the Wikimedia Foundation – the charity which owns Wikipedia and other similar sites – told the BBC that the problem had been fixed.
“Things appear to have been patched up, services are being brought back and things are getting to OK now,” he said.
“Someone cut the cables going to the Tampa, Florida data centre. We have two big [centres], one in Florida and one in Virginia, and some network proxies in Amsterdam.
“Everyone in that data centre was affected!”
He said it was not yet clear what had caused the cut, or where exactly it had taken place.
Mr Gerard joked that due to the site’s limited financial resources, some of its infrastructure relied on “gaffer tape and string”.
In an error message posted to the site, the Wikimedia Foundation reiterated its reliance on donations to fund its continued operation.
“The Wikimedia Foundation is a non-profit organisation which hosts some of the most popular sites on the internet,” the message read.
“It has a constant need to purchase new hardware. If you would like to help, please donate.”
Despite its limited funding, the site is considered to have impressive reliability. Its last significant down time was deliberate – the site went “offline” for 24 hours in protest at proposed anti-piracy bills in the US.