Skip to content

Inside Petal: Building the World's First Petabit-Class Transoceanic Subsea Cable

Summary

Meta announced Petal, the first transoceanic subsea cable to deliver petabit capacity (1 Pbps = 1,000 Tbps) and the first to deploy multi-core fiber at scale. Spanning ~7,000 km (4,300 mi) between France and the United States and expected in service in 2029, Petal doubles what today's most advanced cables carry at transatlantic distance — Meta's own Anjana (24 fiber pairs, 0.5 Pbps) was the prior high-water mark. The core bet is a 2-core fiber in a 24 fiber-pair system, "equivalent to 48 fiber pairs," which Meta calls the single largest generational capacity increase of any repeatered subsea system ever (5.5× Marea, Meta's first transatlantic cable). The article is a clear primer on how subsea capacity has actually scaled: EDFA amplification (1980s) → coherent optical transmission + dispersion-uncompensated cable (2010s, 10×+ gains) → the Shannon-limit wall → spatial division multiplexing (SDM) (more fibers per cable). Meta had scaled SDM from Marea's 8 fiber pairs → Amitié's 16 → Anjana's 24, and faced three ways to double again: (1) push to 48 fiber pairs conventionally; (2) expand the optical band into the L-band (24-pair C+L, as done on PLCN); or (3) adopt 2-core fiber. Petal chose option 3. The two hard engineering problems are keeping attenuation low while holding the standard 125 µm outer fiber diameter (solved with ultra-pure synthetic silica) and minimizing crosstalk between the two cores (solved with high-refractive-index cores in a lower-index medium plus counter-propagating signals → "nearly immeasurable crosstalk"). Because single-core amplification is more efficient and reliable, Petal's single-body 96-amp repeater uses a Fan-In/Fan-Out (FIFO) interface to split 2-core fiber into two single-core fibers for amplification and recombine them, with an SDM pump-sharing architecture — letting Petal double capacity without a proportional power increase and stay within existing ≤18 kV power-feeding-equipment limits (avoiding a costly requalification of the subsea ecosystem). Built with NEC (turnkey system supplier), Sumitomo Electric Industries (2-core "2C Z-PLUS ULL" fiber), and Orange (French landing + European terrestrial interconnection).

Key takeaways

  1. Subsea capacity has scaled through three regime shifts, and Petal is the fourth. EDFA amplifiers (1980s) enabled long-haul optical transmission; coherent optical transmission + dispersion-uncompensated cable designs (2010s) drove "10x and more" per-fiber capacity gains until the Shannon limit pushed back; the industry then pivoted to SDM — increasing the number of fibers rather than per-fiber bits. Petal extends SDM into the fiber core itself (Source: sources/2026-09-21-meta-inside-petal-building-the-worlds-first-petabit-class-transoceanic-subsea-cable).

  2. Petal delivers 1 Pbps over ~7,000 km — double Anjana's 0.5 Pbps and 5.5× Marea. Meta's SDM scaling ladder is explicit: Marea (8 fiber pairs) → Amitié (16) → Anjana (24, the first 0.5 Pbps transatlantic system) → Petal (24 pairs of 2-core fiber = 48-fiber-pair equivalent, 1 Pbps). Framed as "the single largest generational increase in cable capacity of any repeatered subsea system, ever."

  3. Three ways to double again — Meta picked 2-core fiber. The options were: (1) conventional 48 fiber pairs; (2) C+L band expansion (24-pair C+L, as on PLCN in the Pacific); or (3) 2-core fiber in a 24-pair system. Option 3 concentrates a petabit into one cable, which "reduces materials, resources, and carbon footprint compared to building two 0.5 Pbps systems."

  4. The two fiber problems: low attenuation at fixed 125 µm diameter, and inter-core crosstalk. Sumitomo holds the standard 125 µm outer diameter (about a human hair) using ultra-pure synthetic silica in the preform, and suppresses crosstalk by pairing high-refractive-index cores against a lower-index surrounding medium and counter-propagating the two signals — yielding "nearly immeasurable crosstalk" and optical performance "nearly identical to single-core fiber." A preform 2–3 m long / 20 cm wide is drawn to thousands of km at 125 µm. (Source: systems/sumitomo-2c-zplus-ull-fiber|2C Z-PLUS ULL Fiber)

  5. The repeater keeps single-core amplification via Fan-In/Fan-Out (FIFO). A ~7,000 km cable needs "about a hundred repeaters." Petal's single-body 96-amp repeater uses FIFO to transition each 2-core fiber into two single-core fibers, amplify with an SDM pump-sharing architecture, then recombine — retaining "the highest efficiency and reliability of single-core amplification" while carrying 2-core capacity. (Source: Petal 96-amp FIFO repeater)

  6. Capacity doubles without a proportional power increase — the load-bearing systems constraint. FIFO + highly efficient amplification + low-loss fiber let Petal stay within existing power-feeding equipment rated ≤18 kV. Going above that voltage would "trigger a requalification of the subsea ecosystem" — a large, expensive step Meta deliberately avoids. This is the article's central engineering-trade-off: capacity/watt, not just capacity.

  7. A three-party partner ecosystem builds it. NEC is the turnkey supplier (cable + repeaters + FIFO + system powering; manufacturing + installation) and "engineered and qualified the world's first petabit transoceanic system." Sumitomo Electric supplies the 2-core fiber. Orange lands Petal on France's Atlantic coast and handles the European terrestrial interconnection — the terrestrial segment's "security and resilience." Nearly 99% of intercontinental data traffic travels through subsea cables, making capacity here a foundational Internet-backbone investment.

Systems, concepts, and patterns extracted

  • Systems: Petal (the cable system); Sumitomo 2C Z-PLUS ULL 2-core fiber; NEC single-body 96-amp FIFO SDM repeater. Prior Meta cables named as SDM-ladder history: Marea (8 fiber pairs), Amitié (16), Anjana (24, first 0.5 Pbps transatlantic), and PLCN (Pacific C+L-band reference) — recorded as mentions, not dedicated pages.
  • Concepts: Spatial division multiplexing (SDM) — the central primitive, extended here into multi-core fiber; Submarine cable — the infrastructure class; Digital sovereignty — France–US routing + Orange as European landing party. The Shannon limit, EDFA amplification, coherent optical transmission, dispersion-uncompensated cable design, C-band / L-band, and crosstalk are recorded as tags/prose (foundational optical-networking terms, single-source here — below the page-creation bar).
  • Patterns: none minted. The turnkey-supplier + fiber-supplier + landing- partner ecosystem (NEC / Sumitomo / Orange) is an implementation detail of a single project, not a reusable named pattern — left as prose.

Operational numbers

Metric Value
Total capacity 1 Pbps (1,000 Tbps)
Length ~7,000 km (4,300 mi), France ↔ United States
Fiber design 2-core fiber, 24 fiber pairs (= 48-pair equivalent)
Fiber outer diameter 125 µm (≈ human hair)
Repeaters ~100 over the route; single-body 96-amp repeater
Power-feeding limit ≤18 kV (stays within existing equipment)
In service 2029
Relative capacity 2× Anjana (0.5 Pbps); 5.5× Marea
Illustrative scale ≈ 75% of world population streaming music simultaneously (~6.25B streams at ~160 kbps)
Subsea traffic share ~99% of intercontinental data traffic

Caveats

  • Petal is an announcement: the cable enters service in 2029, so the capacity/power figures are design targets, not measured production numbers.
  • Crosstalk is described qualitatively ("nearly immeasurable", "nearly identical to single-core") — no dB figures are disclosed.
  • Fiber count semantics: Petal is 24 fiber pairs of 2-core fiber, which Meta describes as "equivalent to 48 fiber pairs." The 48-pair conventional path and the 24-pair C+L path were the alternatives not taken.

Source

Last updated · 766 distilled / 2,225 read