Meta is building one of the world’s biggest undersea cable networks—it’s just one piece of a growing seafloor empire spanning 6 continents

Meta announced Petal, a 4,300-mile U.S.-France subsea cable targeted for 2029 with 1 petabit-per-second capacity—roughly double current leading transoceanic systems. The project supports Meta's expanding AI and data-center network, alongside its 31,000-mile Waterworth cable system, which could cost more than $10 billion. Meta expects $130 billion-$145 billion in 2026 capital expenditures, nearly double its $72.2 billion spend in 2025, underscoring the scale of its infrastructure investment.

Analysis

The investment relevance is not near-term bandwidth revenue but META’s vertical integration of an AI distribution bottleneck. Owning scarce long-haul capacity lowers its marginal cost of serving inference and video-heavy products internationally, reducing dependence on carrier pricing and improving the economics of moving workloads among geographically dispersed GPU clusters. The advantage compounds only after 2029; before then, the market should treat this as evidence that META’s capex cycle is broadening beyond GPUs and data centers, which raises the hurdle for incremental ad-revenue and AI-monetization returns.

ORA is a modest beneficiary through landing-station, backhaul and enterprise-connectivity opportunities, but the larger second-order winners are cable-system vendors NEC (6701 JP) and Sumitomo Electric (5802 JP), where a concentrated supplier base can support backlog visibility and pricing. MSFT faces a mixed read: industry-wide capacity expansion reduces congestion risk for its cloud and AI products, yet hyperscaler-owned routes weaken the strategic position of traditional telecom intermediaries and may reduce incentives for future shared ownership. AAPL has little direct financial exposure; faster global AI delivery could improve app engagement, but it does not alter the company’s core earnings drivers.

Consensus may over-credit this as an immediate AI monetization catalyst. The financial payoff is contingent on sustained cross-border inference demand and utilization high enough to offset a growing fixed-cost base; cable assets can become underutilized if on-device AI, model compression, regional data localization, or lower-cost terrestrial alternatives reduce traffic intensity. Over the next 1-3 months, META’s valuation response should be governed by capex guidance and evidence of advertising or agent monetization, not engineering milestones; over 6-18 months, the key risk is further upward capex revisions without corresponding operating-margin durability.

Key Decisions for Investors

• Maintain META as a long-term core long, but do not add solely on this announcement. Add on post-earnings weakness only if management holds the capex envelope and demonstrates stable or improving operating margin; the 12-18 month upside is multiple support from a defensible AI-cost moat, while the principal downside is another capex step-up without revenue conversion.
• Watch-list long NEC (6701 JP) and Sumitomo Electric (5802 JP) on disclosed subsea-cable backlog or margin commentary rather than pre-emptively buying the project narrative. A 6-18 month supplier trade is attractive only if order growth converts to higher-margin systems revenue; falsify on flat backlog, competitive bid losses, or fixed-price cost overruns.
• Avoid treating ORA as a direct proxy for META’s infrastructure spend. Consider ORA only if it discloses material landing, wholesale-capacity, or backhaul economics; absent that disclosure, the likely revenue contribution is too small to justify a standalone position.
• For a relative-value expression over the next 1-3 months, favor long META versus MSFT only if META’s next results show AI-driven engagement or ad-load gains while capex remains within guidance. Exit the spread if META cuts expense discipline, raises full-year capex materially, or MSFT shows accelerating Azure AI growth that re-establishes its infrastructure monetization lead.