The next major leap in the internet may not come from a new smartphone, a faster Wi-Fi standard or another AI model.
It may come from something most people will never see.
A glass fiber cable sitting on the floor of the Atlantic Ocean.
Meta has announced Petal, a new 7,000-kilometre subsea cable that will connect the United States and France and is designed to carry 1 petabit per second (1 Pbps), or 1,000 terabits per second when it enters service in 2029.
Meta says that would make Petal the first transoceanic subsea system to reach petabit capacity and roughly double the capacity of today's most advanced transoceanic cables.
But the bigger story isn't simply the number.
Petal is a glimpse of the internet's next infrastructure layer.
Because as AI, cloud computing, video, autonomous systems and billions of connected devices consume more bandwidth, the global internet increasingly needs something that sounds almost contradictory:
More data through the same physical space.
And that's what Petal is designed to do.
The Internet Is Not “In the Cloud”
We often talk about the internet as if it exists somewhere in an invisible digital space.
Your photos are “in the cloud.”
Your Netflix stream comes “from the cloud.”
Your AI conversation happens “in the cloud.”
But the cloud is ultimately physical infrastructure.
Servers sit inside data centres.
Data centres connect to terrestrial fibre networks.
And when information crosses an ocean, much of it travels through submarine fibre-optic cables laid across the seabed.
Meta says approximately 99% of intercontinental data traffic travels through subsea cables.
That means the seemingly invisible digital economy is sitting on a very physical foundation.
The next time someone in Lagos sends a WhatsApp message to someone in London, an American company trains an AI model using data distributed across continents, or a European user accesses a service hosted in another region, the underlying infrastructure may ultimately involve fibre running beneath an ocean.
The internet has always had a physical geography.
Petal is part of the effort to expand that geography's capacity.
What Makes Petal Different?
The headline number is 1 Pbps.
But capacity isn't the only innovation.
Petal is designed around multi-core fibre, specifically a two-core fibre architecture.
Traditional optical fibre essentially uses a single light-guiding core within the fibre.
Petal puts two optical cores inside one fibre strand.
That allows the cable to increase the amount of information carried without simply doubling the number of physical fibres.
Meta says Petal will use 24 fibre pairs with two cores per fibre, effectively providing the equivalent of 48 fibre pairs in terms of the number of optical cores.
That is important because subsea cables aren't simply bundles of glass.
They are complex engineering systems containing optical fibres, protective structures, repeaters, power systems and other components that have to survive thousands of kilometres underwater.
More fibre isn't free.
More physical infrastructure means more material, manufacturing complexity and potentially more power requirements.
Petal's approach is therefore essentially:
Increase the information density of the cable rather than simply making the cable physically larger.
Think of It Like Expanding a Highway Without Building a Second Highway
Imagine a motorway carrying traffic between two cities.
If traffic doubles, the obvious solution is to build another motorway.
But that requires more land, construction, materials and maintenance.
Now imagine engineers discover a way to make each existing lane carry significantly more traffic without proportionally increasing the physical footprint.
That's roughly the principle behind multi-core fibre.
Instead of simply asking:
“How do we put more fibres into the cable?”
engineers are asking:
“How much more information can each fibre physically carry?”
That distinction could become increasingly important as global bandwidth demand continues to grow.
The AI Boom Is Changing the Meaning of Bandwidth
This is where Petal becomes bigger than a telecommunications engineering story.
The world is entering an AI infrastructure race.
AI models require enormous data-centre capacity.
Data centres require electricity.
They require chips.
They require cooling.
And they require networks capable of moving enormous quantities of data between facilities, regions and users.
The internet infrastructure that was originally designed around websites, email and relatively modest data transfers is increasingly being asked to support:
- generative AI;
- video streaming;
- cloud computing;
- autonomous systems;
- real-time communications;
- massive software platforms;
- industrial IoT;
- gaming;
- virtual and augmented reality;
- machine-to-machine communication.
Meta itself has explicitly linked its subsea investments to the growing demands of AI and high-bandwidth applications. Its earlier Project Waterworth initiative was announced as a more than 50,000-kilometre subsea cable project connecting major regions including the United States, India, Brazil and South Africa.
So Petal should not be viewed as an isolated cable.
It is part of a much larger infrastructure race.
The Internet Is Becoming an Infrastructure Arms Race
For years, the biggest internet companies competed primarily on software.
Search.
Social media.
Advertising.
Cloud platforms.
Operating systems.
Now, the largest technology companies are increasingly investing directly in the physical infrastructure underneath their services.
Meta has become one of the world's largest private investors in subsea cable infrastructure, with more than 20 subsea cable projects, according to the company.
Google, Microsoft, Amazon and other major technology companies have also invested heavily in subsea connectivity and data-centre infrastructure.
The reason is straightforward.
If your business depends on moving enormous quantities of data, owning or controlling critical pieces of the network becomes strategically important.
The cable is no longer merely a telecom operator's problem.
It is increasingly a technology company's problem.
Petal Is Also About Power Efficiency
There is another interesting part of Meta's announcement that deserves more attention.
Petal isn't simply trying to increase capacity.
It is trying to increase capacity without proportionally increasing power consumption and physical infrastructure.
According to Meta's engineering team, the two-core design allows Petal to reach its capacity target while remaining within existing subsea power-feeding equipment limits, including systems rated up to 18 kilovolts.
That's significant because energy is becoming one of the biggest constraints on digital infrastructure.
AI data centres already consume enormous quantities of electricity.
Adding more servers means adding more power.
Adding more network capacity can also require more equipment and energy.
So the future of the internet isn't simply:
More bandwidth.
It is:
More bandwidth per unit of physical infrastructure and energy.
That is a much more important engineering problem.
And Then There Are the Repeaters
There's another piece of Petal's design that most people will never think about.
A 7,000-kilometre optical cable can't simply send a signal from France to America and hope it survives.
The optical signal weakens as it travels.
Subsea cables therefore use repeaters along the route to amplify the signal.
Meta says a cable of this distance typically needs around 100 repeaters.
Petal introduces a specialised repeater architecture capable of handling 96 fibre cores in a single repeater body, using a fan-in/fan-out interface to transition between the two-core fibre and single-core amplification system.
In other words, the breakthrough isn't just the glass.
The entire ecosystem around the glass has to evolve.
Fibre.
Repeaters.
Power.
Manufacturing.
Installation.
Landing stations.
Terrestrial networks.
Everything has to work together.
From Terabits to Petabits
The scale of the transition is worth understanding.
A terabit is 1,000 gigabits.
A petabit is 1,000 terabits.
So:
1 Pbps = 1,000 Tbps = 1,000,000 Gbps.
That's an extraordinary amount of bandwidth concentrated into a single transoceanic system.
Meta says the capacity is roughly enough for around 6.25 billion simultaneous 160-kbps audio streams, which it describes as approximately 75% of the world's population streaming music simultaneously.
That comparison isn't intended to suggest everyone will suddenly receive a petabit internet connection.
They won't.
The capacity is shared across networks and services.
But it helps demonstrate the scale.
The real purpose is to create enormous underlying capacity that can be divided among cloud services, telecom operators, businesses, platforms and users.
This Is What the Next Internet Looks Like
The first generation of the internet was about getting computers connected.
The next phase was about getting people connected.
Then came smartphones.
Then cloud computing.
Now we're entering an era where machines themselves are becoming enormous consumers and producers of data.
AI systems generate and process massive datasets.
Robots need low-latency connectivity.
Autonomous vehicles will exchange information.
Factories will become increasingly connected.
Data centres will communicate with other data centres.
AI inference will happen across distributed infrastructure.
And billions of humans will continue consuming video, music and social media.
The network underneath all of this has to scale.
That's why cables like Petal matter.
Africa Should Be Watching Closely
There is an important African dimension to this story.
Meta's subsea infrastructure strategy already extends into Africa through projects such as 2Africa.
Meta says the completed core 2Africa system spans approximately 45,000 kilometres, connects 33 countries and provides connectivity across Africa, Europe and Asia.
The West African segment includes landing points in countries including Senegal, Ghana, Côte d'Ivoire, Nigeria, Gabon, the Republic of Congo, the DRC and Angola. Meta says that segment supports 21 Tbps per fibre pair, with eight fibre pairs on the trunk.
This matters because Africa's digital future isn't determined only by how many people own smartphones.
It is also determined by whether the continent has enough international capacity to support:
AI → cloud → fintech → streaming → data centres → software → digital businesses.
More international capacity can create opportunities.
But capacity alone does not automatically create affordable internet or digital prosperity.
The cable still needs landing infrastructure, terrestrial fibre, data centres, local networks, competitive service providers, electricity and supportive policy.
That is where the real economic battle begins.
The Hidden Infrastructure Beneath the AI Revolution
There is a fascinating irony happening right now.
Public attention is overwhelmingly focused on AI models.
Who has the smartest model?
Who has the biggest GPU cluster?
Who will build the next supercomputer?
Who will dominate AI agents?
But underneath all of it is a less glamorous infrastructure stack.
Chips → data centres → electricity → fibre → subsea cables.
Without those layers, AI doesn't scale globally.
Petal therefore represents something bigger than another Meta infrastructure announcement.
It represents the continuing transformation of the physical internet to accommodate an increasingly data-intensive world.
The Next Internet May Be Defined by Density
For decades, technological progress often meant making things smaller.
Smaller chips.
Smaller transistors.
Smaller devices.
But network infrastructure has another problem:
How much information can you move through a fixed physical environment?
Petal's answer is to increase the density of information inside the fibre itself.
That is why the two-core architecture is arguably more interesting than the headline number.
Today it is two cores.
Tomorrow, the industry could explore more sophisticated multi-core designs.
And if those technologies become commercially practical at scale, the amount of information that can move through existing physical corridors could increase dramatically.
That's the larger technological trajectory.
The internet is becoming denser.
And There Is a Strategic Side to This
Subsea cables are also critical infrastructure.
They connect economies, governments, financial systems, technology platforms and ordinary consumers.
That makes their physical routes strategically important.
A cable failure can disrupt connectivity.
A damaged cable can force traffic onto alternative routes.
A region with insufficient redundancy can become vulnerable to a single point of failure.
Meta's own 2Africa engineering work highlights the importance of routing, burial depth, redundancy and protection from seabed hazards.
So the future internet isn't simply about speed.
It is also about:
capacity + redundancy + resilience + geography + energy + security.
2029 Is the Interesting Date
Petal is expected to enter service in 2029.
By then, today's AI landscape could look completely different.
The number of AI users could be dramatically larger.
AI agents could be interacting with thousands of services.
Video could be increasingly generated rather than merely streamed.
Data centres could be distributed across continents.
And businesses could rely on AI systems operating continuously in the background.
All of that requires one thing:
A network capable of carrying the consequences.
Petal is being designed for that world.
The Internet Isn't Floating in the Sky
The next time someone says that “everything is moving to the cloud,” remember what that actually means.
The cloud has buildings.
The buildings have servers.
The servers have chips.
The chips need electricity.
And the buildings need fibre.
Some of that fibre runs thousands of kilometres beneath the ocean.
Meta's Petal is therefore more than a new cable between France and America.
It is a statement about where the internet is going.
More AI.
More data.
More cloud.
More machines.
More connectivity.
More demand.
And underneath it all, an increasingly sophisticated network of glass stretching across the planet.
The next internet revolution may not be something you download.
It may be something engineers lay on the ocean floor.
Comments
Post a Comment