Meta to Construct Undersea Cable Linking Five Continents

The parent company of Facebook and Instagram, Meta, has announced plans to lay an extensive undersea cable that will span five continents. This new cable will facilitate data transmission, particularly for developing artificial intelligence (AI).

The undersea cable will stretch over 50,000 kilometers, connecting regions including the United States, South Africa, India, Brazil, and several additional areas, as detailed in a blog post by Meta.

Undersea cables form a crucial part of global digital communication, with an estimated 1.2 million kilometers of cable already in place, according to a report by the U.S.-based Center for Strategic and International Studies (CSIS) published in 2024. These cables can provide short links between countries or long systems that span across multiple continents.

Each cable consists of several pairs of fiber-optic wires housed in armored sheaths, which might be buried a few meters under the seabed to safeguard against damage.

In recent years, major tech companies like Meta have entered the subsea cable arena, traditionally dominated by telecommunications providers. Alan Mauldin, research director at Telegeography, explained, "When your growth is significant and your demand exceeds that of others, you are motivated to invest directly and eliminate the middleman."

Meta has stated that this cable project involves a "multi-billion-dollar, multi-year investment." While this is a significant commitment, it pales in comparison to the tens of billions that are typically invested in the artificial intelligence sector.

Joint ventures and independent projects to build new cables have ramped up, especially from companies like Google and Meta, due to the enormous data requirements of their platforms, such as YouTube and Facebook.

This forthcoming cable, dubbed "Project Waterworth," marks the third cable Meta is building on its own, in contrast to Google's impressive tally of 16 cables. Meta's initial cable, named "Anjana," which connects the U.S. to Spain, is expected to go live early this year.

The name "Waterworth" honors the late Gary Waterworth, who worked at Meta for five years after a notable career at Alcatel Submarine Networks (ASN), a leading firm for laying submarine cables.

Other firms capable of such installations include SubCom from the U.S., Japan's NEC, and China's HMN. One of the key motivations for major platforms to engage in cable construction is to create direct links to their global infrastructure and enhance network resilience. Mauldin noted that having just one large, high-capacity cable is not sufficient; a network needs multiple cables to ensure reliability if one or two become non-functional.

On average, about 200 incidents damaging undersea cables occur each year, posing risks to significant economic activities. These damages can be caused by natural events like underwater landslides and tsunamis or by human activities such as fishing or ship anchor incidents.

Deliberate acts of sabotage and espionage also threaten undersea cables. In January, NATO began conducting patrols in the Baltic Sea following suspected attacks on telecom and power cables, believed to be linked to tensions with Russia.

Meta's proposed "Waterworth" route is strategically planned to avoid high-risk geopolitical areas, such as the South China Sea and the Red Sea.

The company emphasized that this cable project aims to deliver the essential, high-speed connectivity required for AI development. Data demand continues to grow, particularly for AI, which is currently a focal point in the tech industry. Mauldin remarked that although the exact impact of AI on the need for cables and bandwidth remains uncertain, training AI models typically demands fast data transfers to computing centers distributed around the globe.

AI inference, which consists of the AI's responses to user queries, also comes with its own transmission needs. Mauldin pointed to potential newcomers in the AI sector, highlighting that established tech firms, referred to as "hyperscalers," have significant advantages due to their existing data centers and networks. Examples include OpenAI, which is strongly supported by Microsoft, and Anthropic, financially backed by Google and Amazon.

Meta, cable, AI