Meta's 1-Petabit Petal Cable Doubles Transatlantic Capacity — But Not Until 2029

Meta Announces 1-Pbps Transatlantic Cable

Meta announced Petal, a France-to-U.S. subsea cable system designed to deliver 1 petabit per second across approximately 7,000 kilometers — roughly double the capacity of the most advanced transoceanic cables in operation today. The system will use multi-core fiber, a technology that embeds multiple optical cores in a single strand to increase throughput without laying additional physical cable.

The announcement, made September 21, provides no service date, construction budget, wholesale pricing, committed customers, or carrier-access model. That makes Petal a future supply signal, not a budgetable connectivity option for enterprise network planners working on 2027 or 2028 capacity contracts.

What 1 Pbps Means in Context

Petal's announced capacity sits above two other hyperscaler-backed transoceanic projects. AWS is building Sta'O'Nuk, a 20-fiber-pair Japan-to-Washington State system rated at 420 Tbps, targeted for 2029. Exa Infrastructure's Meridian system, a 6,552-km transatlantic route, is reported at up to 500 Tbps. Conventional carrier-operated and consortium cables typically deliver between 200 and 400 Tbps.

The capacity gap reflects AI workload economics. Training runs and inference clusters generate persistent east-west traffic between U.S. data centers and European cloud regions. Hyperscalers are increasingly funding proprietary subsea routes to avoid bandwidth auctions, eliminate third-party transit costs, and control path diversity. Meta's decision to announce 1 Pbps suggests internal traffic models justify the capital expenditure even if no third-party revenue materializes.

For enterprise buyers, the question is whether Meta will sell capacity to other operators or enterprises. If it does, Petal could reduce dependence on congested carrier-consortium routes and create pricing pressure. If it does not, the cable adds nothing to purchasable transatlantic inventory.

Digital Realty Plans Los Angeles Cable Landing Station at LAX12

Digital Realty announced plans September 22 to develop a cable landing station at its LAX12 data center, in partnership with Assured Communications. The facility is expected to begin supporting subsea cables in 2028.

The project targets enterprises using Los Angeles as an Asia-to-North America gateway. Current landing-station workflows require separate contracts for subsea spectrum, backhaul, colocation, and carrier or cloud interconnects across multiple facilities and providers. Digital Realty's proposal consolidates landing infrastructure, colocation, and interconnection on one campus.

The announcement does not disclose construction cost, rack or power capacity, carrier count, or customer commitments. Buyers cannot yet calculate cost savings against existing multi-facility workflows or confirm whether their preferred subsea operators have signed landing agreements. The 2028 target also places the project outside most current procurement windows.

Digital Realty competes with established Los Angeles landing and interconnection hubs operated by Equinix, Cologix, CoreSite, and carrier-owned infrastructure. Its differentiator is integration: eliminating physical handoffs between landing point, colocation, and interconnection. Whether that integration commands a price premium or reduces total procurement cost depends on unannounced commercial terms.

Subco Announces Three Asia-Pacific Cables; AWS Adds Japan-Washington Route

Australian subsea developer Subco announced three Asia-Pacific systems — APX North, APX-2, and APX West — with a stated 2029 completion target. The announcements cite rising AI workload demand but provide no route-by-route capacity, financing, cable length, contracted customers, or wholesale prices.

The projects add to a crowded Asia-Pacific field that includes AWS's Sta'O'Nuk cable, a 20-fiber-pair, 420-Tbps Japan-to-Washington State system also targeted for 2029. Subco's portfolio competes with hyperscaler-controlled routes, carrier consortia, and specialist infrastructure operators.

For enterprise buyers, the development signals future route diversity for Australia and Asia-Pacific cloud traffic. It does not yet support immediate sourcing decisions. Buyers should track whether Subco converts the announcements into binding capacity contracts and disclosed landing-station plans. A 2029 target without published commercial terms is insufficient to reserve budget or negotiate service-level commitments.

What to Watch

The strongest pattern is hyperscaler acceleration toward proprietary, ultra-high-capacity subsea routes. Meta's 1-Pbps Petal, AWS's 420-Tbps Sta'O'Nuk, and Exa Infrastructure's 500-Tbps Meridian all reflect internal traffic models that justify bypassing carrier-consortium procurement.

For enterprise network buyers, that creates a bifurcated market: hyperscaler-controlled capacity that may or may not become available for third-party purchase, and carrier-operated capacity that remains subject to consortium governance and auction-based pricing. The immediate procurement question is not "how much capacity exists" but "how much capacity is contractually available, geographically diverse, and priced in published wholesale terms."

Buyers planning 2027 or 2028 transatlantic or transpacific capacity should ask providers whether they offer dedicated spectrum, managed wavelengths, or only best-effort IP transit. They should also confirm the number and ownership of physical landing points to assess path diversity and avoid single-provider concentration risk. Announced capacity is not purchasable capacity until commercial terms, landing-station access, and service dates are disclosed.