Undersea internet cables underpin our way of life—but they’re vulnerable, as Red Sea incident shows

This picture taken on March 7, 2024 shows the Rubymar cargo ship partly submerged off the coast of Yemen.
An attack on the Rubymar cargo ship pff Yemen may have led to the damage of undersea internet cables.
KHALED ZIAD—AFP via Getty Images

The internet is in some ways an abstract layer that augments our physical world, but it’s important to remember that it, too, is physical—a network of networks or, as the late Sen. Ted Stevens memorably described it nearly two decades ago, a “series of tubes.”

Some of the most important data conduits lie buried in the seabed, connecting continents. The vulnerability of these cables was emphasized in late February, when a few were damaged in the Red Sea, markedly degrading people’s online experience in parts of East Africa and South Asia.

At the time, there was speculation that Yemen’s Iran-aligned Houthi rebels had caused the damage after earlier suggesting on social media that the nearby infrastructural hotspot—there are 16 crucial submarine cables in the Red Sea—was a tempting target. The Houthis ultimately denied responsibility, claiming that U.S. and U.K. naval forces were behind the incident.

But now it increasingly looks like the rebels were responsible, albeit unwittingly. Backing up one theory about how the incident played out, Wired yesterday published an analysis that strongly suggests the cables were cut by a sinking cargo ship called the Rubymar, which the Houthis had hit with a ballistic missile. (Attacks such as this one are of course the reason why the U.S. and U.K. have sent naval vessels to the area.)

The evidence isn’t entirely conclusive, but satellite imagery, and data marking the timing of the cables going offline, indicate that the Rubymar’s anchor may have cut or severely bent the fiberoptic cables while dragging along the seabed, as the vessel drifted north from the spot where it was attacked (known as the Gate of Tears). The crew, all of whom were saved, had dropped the anchor during the rescue operation.

If and when the cables are repaired, it should be possible to get a clearer picture of what caused the damage. This would be a dangerous mission, thanks to those Houthi rebels.

The incident serves as a reminder of something positive: the internet’s resilience. The damage to these cables may have had a noticeable impact on connection speeds, but it didn’t cut anyone off. As it’s designed to do, the internet routed around the damage, making sure people’s data packets got where they needed to go.

However, there are limits to this resilience when many cables drop out. Just look at the internet outages in countries like Ivory Coast, Ghana, and Liberia a couple weeks ago, when multiple West African coastal cables failed. In this case, seismic activity is the chief suspect; again, the repair operation will reveal more, as it progresses over the coming weeks.

At this point it’s worth remembering that, last year, Taiwan suspected China’s hand behind the severing of two cables that serve its main island. Also last year, Russian hawks threatened to attack Western undersea cables—Russia may or may not have the capacity to actually make good on such threats, but NATO has certainly taken notice, recently establishing a new monitoring station in the U.K. to keep an eye on Russian activity that may point to potential cable sabotage.

After all, that series of tubes has become fundamental to our lives, making it a worthwhile target—and worth defending. More news below.

David Meyer

Want to send thoughts or suggestions to Data Sheet? Drop a line here.

NEWSWORTHY

Altman hands over fund reins. OpenAI CEO Sam Altman has given up control of the OpenAI Startup Fund. As Axios reports, the company claims the fund’s initial general partner structure was “a temporary arrangement, and involved no personal investment or financial interest from Sam.” Ian Hathaway, who has led the fund’s investments in companies like Cursor (a code editor) and Speak (AI-powered language tuition), now has formal control over it.

U.S./U.K. AI agreement. The U.S. and U.K. have signed an agreement around the testing of advanced AI systems, the BBC reports. Under the terms of the bilateral memorandum of understanding, the countries will jointly develop tests and perform at least one testing exercise on a publicly accessible model. They might even swap some staffers between their respective AI safety institutes.

U.S. internet subsidies. Millions of U.S. households are at risk of losing internet access due to the looming demise of a subsidy program that helped get them online. That government initiative, the Affordable Connectivity Program, doesn’t have the funds to provide full subsidies beyond this month. But, according to Reuters, the White House will shortly make a fresh push to get Congress to extend the scheme.

ON OUR FEED

“[The flood of AI funding] brings with it a whole attendant bunch of hype and maybe some grifting and some other things that you see in other hyped-up areas, crypto or whatever…In a way, AI’s not hyped enough but in some senses it’s too hyped. We’re talking about all sorts of things that are just not real.”

DeepMind cofounder Demis Hassabis tells the Financial Times that hype is obscuring real AI advances.

IN CASE YOU MISSED IT

U.S. pushes South Korea to follow its lead and tighten export controls on chips for China, by Bloomberg

Amazon hits pause on controversial seller fee rollout that had caused a revolt, by Jason Del Rey

Warren Buffett-backed EV maker BYD surprises with stellar first quarter sales as Tesla continues to slide, by Marco Quiroz-Gutierrez

Xiaomi warns potential customers that they may to wait over half a year to get the company’s new EV, as its shares jump by as much as 16%, by Lionel Lim

Trump’s social media platform says it’s above ‘traditional key performance indicators.’ An auditor has substantial doubt it can continue as a going concern, by Jason Ma and Amanda Gerut

BEFORE YOU GO

Ad-funded AI search. If AI is the future of search, and the conversational nature of AI-powered search favors single answers rather than the listed results where ads have traditionally appeared, what might the future connection between search and advertising look like? Perplexity, a hot new AI search service, is experimenting to find out. The company told AdWeek that it will let brands influence the suggested follow-on questions that it gives users alongside the main response.

“Advertising was always part of how we’re going to build a great business,” said chief business officer Dmitry Shevelenko—despite the fact that Perplexity officially claims search should be “free from the influence of advertising-driven models.”

This is the web version of Data Sheet, a daily newsletter on the business of tech. Sign up to get it delivered free to your inbox.