Meta’s Petal Subsea Cable: Multi-Core Fiber Targets 1 Petabit per Second Across Atlantic

Twice the Capacity of Anjana: Meta’s September 21 Transatlantic Blueprint

On September 21, 2026, Meta announced Petal, a new transoceanic subsea cable system connecting France and the United States that spans approximately 7,000 km (4,300 miles) and is expected to enter service in 2029. According to Meta’s announcement, Petal is designed to deliver 1 petabit per second (Pbps) of capacity, which the company states is about twice the capacity of today’s best transoceanic cables. As reported by Baird Maritime, Meta compared that 1 Pbps figure to streaming music simultaneously for 75 per cent of the world’s population, noting that the design transfers twice the data without a proportional increase in electrical power or physical cable size.

To build and land the transatlantic link, Meta partnered with NEC and Sumitomo Electric Industries, while Orange is supporting the cable landing on France’s Atlantic coast. As Interesting Engineering detailed, NEC serves as the turnkey system supplier for the project and Sumitomo Electric Industries is supplying the specialized optical fiber. According to the IEEE ComSoc Technology Blog’s September 28, 2026 analysis of Meta’s statement, the company has now invested in more than 20 subsea cable projects worldwide.

Independent optical-networking analysts are careful to distinguish between theoretical design ceiling and initial lit throughput. As the IEEE ComSoc Technology Blog stresses, the 1 Pbps figure is a design target rather than immediately deployed capacity, and one-way optical propagation across the roughly 7,000 km Atlantic route is about 35 milliseconds. Even with that engineering caveat, moving the design envelope of a single transoceanic cable to the petabit threshold marks a major milestone in undersea telecommunications.

Inside the Engineering: Two-Core Fiber, 96-Amp Repeaters, and the 18 kV Power Ceiling

For the past decade, subsea cable engineers increased capacity primarily through space-division multiplexing—packing more traditional single-core fiber pairs into the same copper-sheathed marine conduit while operating at lower optical power per pair. Eventually, however, physical conduit diameter and high-voltage subsea power delivery hit hard mechanical and electrical limits. According to Engineering at Meta, Petal will be the first subsea cable system to deploy multi-core fiber technology at scale to break through that bottleneck.

Specifically, as Interesting Engineering explains, the Sumitomo fiber fits two independent light-guiding cores into a single standard glass strand. By deploying 2-core fiber technology inside a 24 fiber-pair system, Engineering at Meta notes that Petal achieves the functional equivalent of 48 fiber pairs, doubling capacity compared with Meta’s earlier Anjana subsea cable, which reached 0.5 Pbps across 24 single-core fiber pairs.

Amplifying 48 optical pathways across 7,000 km of ocean floor without exceeding safe marine voltage tolerances required a parallel redesign of the undersea repeaters. According to Engineering at Meta, the transatlantic route requires about a hundred repeaters spaced along the seabed. Petal uses a newly engineered single-body 96-amp repeater equipped with optical fan-in/fan-out interfaces that separate and recombine the two cores at each amplification stage, all while operating within existing 18 kV power-feed limits. Keeping the system within the 18 kV envelope allows Meta and NEC to double optical path density without requiring exotic new shore-end power feeding equipment or thicker marine armor.

Why AI and Data Center Replication Drive Transoceanic Cable Design

The push toward petabit-class cables is frequently linked to the rise of artificial intelligence, though network engineers point out that the relationship is structural rather than tied to a single model run. As the IEEE ComSoc Technology Blog argues, AI is not a single monolithic workload that continuously consumes a full petabit per second on its own; rather, moving enormous amounts of data between global data centers becomes increasingly important as both AI workloads and cloud services expand.

Because hyperscalers replicate training datasets, synchronize distributed storage clusters, and serve inference traffic between North American and European cloud regions, inter-data-center backbone traffic grows far faster than ordinary consumer web browsing. By doubling the fiber-pair equivalence per cable without doubling repeater housings or shore power, multi-core systems like Petal lower the long-term unit cost of moving bulk data across the Atlantic.

What This Means for Algerian Telecom Planners, Cloud Architects, and Developers

Petal is a transatlantic cable between the United States and France’s Atlantic coast—it does not land in North Africa or directly serve Algeria, and its relevance for the Algerian tech ecosystem is strictly indirect. Even so, because Algerian traffic to Western European cloud regions and content delivery networks (CDNs) relies on French and European transit hubs where Orange and global hyperscalers interconnect, local telecom engineers and cloud architects can draw four grounded takeaways from Meta’s 2029 roadmap.

1. Recognize How French Atlantic Landings Shape Downstream European Cloud Transit

With Orange supporting Petal’s landing on France’s Atlantic coast and the cable scheduled to enter service in 2029, Western Europe’s primary transatlantic gateway is gaining a 48-equivalent fiber-pair corridor directly to the United States. Algerian developers and enterprises hosting workloads in French or Western European cloud regions—or pulling AI model weights, container registries, and CDN caches hosted in North America via European transit—benefit indirectly when transatlantic backbone congestion and per-bit transport costs decline between the US and France.

2. Separate Bandwidth Capacity From Propagation Latency in Architecture Design

As the IEEE ComSoc analysis points out, even on a petabit-class cable like Petal, one-way optical propagation across the roughly 7,000 km Atlantic route remains fixed by the speed of light in glass at about 35 milliseconds. Doubling a cable’s capacity from 0.5 Pbps to a 1 Pbps design target allows vastly more concurrent data transfers between US and European data centers, but it never reduces the physical round-trip latency between continents. Algerian cloud architects designing latency-sensitive applications must continue to place real-time compute and caching nodes close to end users in Algeria or nearby Mediterranean regions rather than expecting higher-capacity transoceanic cables to speed up individual packet round trips.

3. Track Multi-Core Fiber and 96-Amp Repeater Maturity for Future Subsea Upgrades

Petal represents the first at-scale subsea deployment of 2-core fiber technology paired with single-body 96-amp repeaters operating within existing 18 kV power-feed limits, supplied by NEC and Sumitomo Electric Industries. In the subsea telecom industry, optical technologies proven first on high-volume transatlantic routes by hyperscalers typically become standard commercial options for regional and Mediterranean systems in subsequent procurement cycles. Algerian telecom planners and subsea infrastructure engineers should monitor the field performance of fan-in/fan-out interfaces and 2-core splicing during Petal’s deployment ahead of 2029, as these techniques allow future cables to double pair counts without exceeding standard 18 kV power feeds.

4. Read Hyperscaler Cable Announcements With Engineering Realism

The distinction highlighted by IEEE ComSoc—that 1 Pbps is a design target rather than initial deployed capacity—is a useful rule of thumb for evaluating all international connectivity announcements. When global cloud providers or telecom consortia announce headline terabit or petabit figures, enterprise infrastructure buyers should inspect the underlying physical parameters: fiber-pair count, repeater spacing, power-feed voltage, and expected ready-for-service dates.

The Structural Lesson: Hyperscale Data Centers Now Dictate Optical Physics

Meta’s investment in more than 20 subsea cable projects—culminating in the September 21, 2026 announcement of Petal—shows how thoroughly hyperscale cloud and AI operators now drive the physical layer of the global internet. When a single company commissions a 7,000 km transatlantic system with NEC, Sumitomo Electric Industries, and Orange to double the capacity of its 0.5 Pbps Anjana cable using 2-core glass strands and 96-amp repeaters, undersea engineering is no longer paced by traditional voice and consumer ISP traffic.

For engineers in Algeria and across the Mediterranean, Petal will not change local last-mile access, nor will it enter service until 2029. What it does signal is that the core optical corridors linking European cloud hubs to North American AI data centers are transitioning to multi-core fiber—ensuring that the upstream transatlantic backbone feeding European transit hubs can absorb the next decade of inter-data-center cloud and AI replication.

Frequently Asked Questions

Sources & Further Reading

• Announcing Petal: Meta’s Petabit Transoceanic Cable — Meta
• Petal: Engineering a Petabit Transoceanic Subsea Cable — Engineering at Meta
• Meta’s Petal Subsea Cable to Bring Petabit-Class Optics to the Atlantic — IEEE ComSoc Technology Blog
• Meta to Develop Transoceanic Subsea Cable Connecting US and France — Baird Maritime
• Meta Unveils Petal, the World’s First Petabit-Class Transatlantic Cable — Interesting Engineering