Meta's Petal cable promises a petabit across the Atlantic

Meta has announced Petal, a subsea cable that will run roughly 7,000 km between France and the United States and, when it enters service in 2029, carry one petabit per second - the first transoceanic cable to do it. Meta says the system will double the capacity of the best cables in the water today, and it plans to get there without doubling the amount of glass, copper or power the job normally demands.

What Petal actually changes

Every big jump in subsea capacity for the past decade has come from packing more fibre pairs into a cable. Meta's own run goes from Marea at 8 pairs, to Amitie at 16, to Anjana at 24 - the first 0.5 Pbps transatlantic system. Petal's trick is to change what a single strand is: it uses 2-core fibre, two light-carrying cores inside one strand, in a 24-pair cable. Meta reckons that is equivalent to 48 pairs, which is how it doubles Anjana's capacity to 1 Pbps, or 1,000 Tbps. Sumitomo Electric's version of that fibre is called 2C Z-PLUS ULL, and the strand is still 125 micrometres wide - about the width of a hair.

Meta's own headline comparison, complete with its working, is that 1 Pbps is roughly enough network capacity for about 75% of the world's population to stream music at the same time, which it calculates as around 6.25 billion simultaneous streams at 160 kbps. The company also puts Petal at 5.5 times the capacity of Marea, its first transatlantic investment, and calls it the single largest generational increase in capacity of any repeatered subsea system.

Why the engineering is the story

A 7,000 km cable needs about a hundred repeaters to keep the signal alive. Petal's repeaters are a single-body, 96-amp design built around a Fan-In/Fan-Out interface that splits each 2-core strand into two single-core fibres for amplification, then recombines them. That lets Meta keep using ordinary single-core amplifiers instead of inventing a new one, and the whole system stays inside existing power-feeding limits rated up to 18 kV - which matters, because going above them would force the wider subsea industry to requalify its equipment.

The hard part is the glass. Turning 2-core fibre into a 7,000 km product means holding low attenuation without widening the strand, and keeping the two cores from talking to each other. Sumitomo says its manufacturing keeps crosstalk to a level that leaves optical performance nearly identical to single-core fibre. NEC is the turnkey supplier: it engineered and qualified the first petabit transoceanic system and will build and install it. Orange will handle the French landing on the Atlantic coast and the onward link into the European network.

"Achieving petabit-per-second capacity across a transoceanic submarine system represents a major technological milestone in the history of global telecommunications," said Eduardo Mateo, chief strategy officer for NEC's submarine network division. Sumitomo's Takehiko Okada framed the fibre as the enabler: "As a pioneer with nearly four decades of experience in ultra-low loss submarine fiber manufacturing, we are committed to supporting the global network expansion essential to realizing a highly digital future."

Why you should care

Roughly 99% of intercontinental data traffic travels through cables on the seabed, which is why every hyperscaler treats ocean glass as core infrastructure. Meta has now put money into more than 20 cable projects, including Project Waterworth. What makes Petal worth watching is that the old escape route is running out: optical engineers have been pushing against the Shannon limit for years, which is why the industry switched to adding fibre pairs rather than squeezing more out of each one. Petal attacks the fibre itself, and if NEC and Sumitomo can manufacture a 2-core ecosystem at scale, that approach becomes a template for everyone else rather than a one-off.

Our opinion

A petabit cable will never trend on social media, and Meta knows it, which is exactly why the announcement leads with a music-streaming analogy rather than the Shannon limit. But the interesting decision here is a commercial one dressed up as physics. Adding fibre pairs is expensive, well understood and boring; changing the fibre so that 24 pairs behave like 48 is a bet that the supply chain can be dragged along with you, and Meta is buying that bet with an 18 kV power envelope specifically chosen so nothing else in the subsea ecosystem has to be recertified. The three-year wait to 2029 is the tell: this is not an answer to a capacity crunch, it is capacity booked in advance, largely for Meta's own AI-era traffic. Europe gets a fatter pipe and a landing site in France; the rest of the industry gets a working example to copy, or a cautionary tale about how expensive it is to be first.