Google & Meta Submarine Cables: Global Capacity, Future Investments and Carrier Partnerships

Open any map of the world's submarine cables from 2014 and the names along the lines are the ones you'd expect — KDDI, NTT, BT, Tata, Singtel, Telefónica, the great consortia of state and private telcos that have laid the wires between continents since the era of the telegraph. Open the same map for cables that lit up in 2024 or 2025, or for the ones in survey today, and a different set of names dominates: Google, Meta, Microsoft, Amazon. The hyperscalers didn't quietly become significant subsea operators. They became the largest.

The shift took about ten years. Today the four of them sit on something close to three-quarters of global subsea fibre capacity, and on the trans-Atlantic route they're closer to nine-tenths. Google alone is part-owner or sole funder of around thirty cable systems. Meta is in roughly fifteen, including the largest cable ever built (2Africa) and the longest one ever attempted (Project Waterworth, announced in early 2025 and now in survey). This is the story of how the company you log into to watch YouTube or scroll Instagram has, in the same decade, become a meaningful operator of the physical internet — and what that means for the telcos who used to own those cables alone.

How carriers became co-stars

For most of the past century and a half, intercontinental wires belonged to carriers. State telegraph monopolies first, then state telcos, then licensed private telcos organised into consortia. A cable like SEA-ME-WE 3 — Southeast Asia, Middle East and Western Europe, lit in 1999 — was a 39-carrier project. Ownership was diffused across continents, capacity was sold off in slices to anyone willing to sign a 25-year IRU, and the economics worked because demand was rising fast enough to absorb whatever the consortium poured in.

That model has not gone away — there are still consortium cables being built in 2026 — but it is no longer the centre of gravity. As of this year:

• Google is now part-owner or sole funder of around 30 subsea cable systems.
• Meta is part-owner of around 15 cable systems.
• Microsoft participates in five, and Amazon in around four.

The drivers are simple. AI training, cloud-region replication, social media video, and global SaaS all generate enormous east-to-west traffic that hyperscalers want to control end-to-end. Buying wavelengths from carriers exposes them to consortium pricing, oversubscription, and other operators' priorities. Funding a new cable — even at USD 300 million to over a billion dollars per system — pays back in lower long-term cost-per-bit and tighter control over physical diversity. That control matters more in an era where a single cable cut between continents can affect production traffic for major services.

From carriers to clouds — the structural shift

The hyperscaler involvement evolved in three stages:

1. Indefeasible Rights of Use (IRUs) — early 2010s. Google, Microsoft and Facebook bought long-term IRUs on carrier-built consortium cables. They were anchor tenants but did not own the cables.
2. Consortium membership — mid 2010s. Hyperscalers became formal consortium members of new builds — typically alongside a half-dozen carriers — and contributed fibre-pair-level investment. Examples include FASTER (US–Japan, 2016) and MAREA (US–Spain, 2018, with Microsoft, Meta and Telxius).
3. Sole ownership — late 2010s onwards. Google built Curie (US to Chile, 2019) as the first wholly-owned hyperscaler cable. Dunant, Equiano, Grace Hopper, Firmina, Topaz, Nuvem and others followed. Meta is now executing on its own wholly-owned Project Waterworth announced in 2025.

Today the typical pattern is a hybrid: hyperscalers anchor the cable, one or two carriers handle landings and backhaul in countries where the hyperscaler has no operating licence, and a wider consortium of regional carriers takes some of the remaining capacity.

Google's submarine cable portfolio

Google's portfolio spans every major ocean. Below is the headline picture — owned, co-owned or sole-funded systems with broad geographic role. (System counts and routes are subject to revision as new builds light up and older ones retire.)

This is not exhaustive — Google also participates in several legacy consortium systems and continues to add new builds. The full picture is documented by subsea-industry analysts; the table above captures the systems most visible in 2026.

Google's future investments

Google's near-term pipeline gives a fair sense of where the money is pointing:

• Sol (RFS 2027). Trans-Atlantic, connecting the US, Bermuda, the Azores, and Santander in Spain. Wholly-funded by Google, landed in Spain by Telxius. Sol complements Dunant and Grace Hopper with a southern trans-Atlantic path, and pairs with Nuvem at the Bermuda hub.
• Nuvem (RFS 2026). Sole-funded by Google, connecting the US East Coast, Bermuda, and Portugal — adding diversity to the US–Iberian peninsula corridor.
• America – India Connect initiative. Announced in 2025; Google committed to landing three additional cables in India over the next several years, complementing existing capacity from Equiano-east (via Indian Ocean) and Apricot. The cables will support a US–India subsea corridor for AI workloads and cloud-region growth.
• South Pacific Connect. A multi-cable initiative — including Tabua and Honomoana — to materially expand subsea fibre into the Pacific Islands and provide redundant US–Australia paths. Reflects both commercial demand and an emerging strategic interest in diversifying Pacific connectivity.
• Umoja (Africa to Australia). First-of-its-kind cable system combining a terrestrial trans-African fibre route with a subsea hop to Australia. Builds on Equiano's African landings to extend Google Cloud's reach into a continent that has historically been over-reliant on Mediterranean and South Atlantic transit paths.
• Bosun. Announced trans-Caribbean cable system from South Carolina serving the Caribbean basin.

The overall trajectory: Google is willing to fund (and in many cases solely own) trans-oceanic cables on every major route where AI and Cloud growth requires more deterministic capacity, while still partnering with carriers for in-country landings and last-mile/backhaul where local licensing or operational expertise is required.

Meta's submarine cable portfolio

Meta's portfolio is smaller in number than Google's but punches above its weight on capacity per cable, with several flagship trans-Atlantic and African systems:

2Africa deserves a closer look. With 45,000 km of fibre when the Pearls extension lights up in 2026, it is the world's longest subsea cable system. It interconnects three continents — Europe, Africa, Asia — touching 33 countries with 46 cable landings, and serving an addressable population of approximately three billion people. Its capacity per fibre pair (~21 Tbps) and the use of switched branching units throughout mean that traffic can be flexibly routed and re-allocated as African markets mature. Planned Red Sea landings (in Sudan and on the west coast of Saudi Arabia) have been delayed by the ongoing Yemen-related disruption to maritime traffic; Meta and Google's parallel Blue-Raman cable in the same waters has faced similar challenges.

Meta's future investments — Project Waterworth

The headline future-investment story is Project Waterworth, announced by Meta in February 2025. It is, on paper, the most ambitious single subsea cable project ever attempted:

• Length: Over 50,000 km — longer than Earth's circumference.
• Reach: Five continents. Confirmed landing regions include the US, Brazil, South Africa, India, and Australia, with multiple intermediate landings.
• Technology: The first subsea cable to use 24 fibre pairs (existing high-end systems use 8–16). This unlocks dramatic per-system capacity gains.
• Engineering: Routing at depths of up to 7,000 m in deep ocean segments, with reinforced burial techniques in shallow / high-traffic coastal zones to mitigate anchor and fishing-trawler cuts.
• Cost: Multi-billion-dollar investment over a multi-year build. Industry sources reported a USD 10 billion+ budget at the project's pre-announcement stage.
• Strategic purpose: Meta has positioned the project as foundational infrastructure for AI workload distribution and the next generation of services. It introduces three new oceanic corridors that complement, rather than replicate, existing hyperscaler routes.

Project Waterworth, combined with 2Africa Pearls (2026), Anjana (already RFS), and Meta's continuing participation in Pacific consortia, gives Meta a credible end-to-end physical footprint matching its software footprint. As of mid-2026, Waterworth is in survey and contracting; final RFS is expected over the next several years.

Pearls, the 2Africa extension, was scheduled to light in 2026 and pushes the system into the Middle East and India. Beyond that, Meta continues to upgrade Echo and Bifrost capacity, and picks its spots in Pacific Island and Atlantic routes that line up with where its data-centre footprint is heading.

How the carrier partnerships work

Even on cables marketed as "wholly-owned" by a hyperscaler, telcos do much of the work that ends up on the press release without their name on it. A few different roles, depending on the project:

Read enough press releases and a few patterns become hard to miss:

• Telxius (Telefónica's infrastructure subsidiary) is now Google's most consistent partner — Dunant, Grace Hopper, Junior, Tannat, Firmina, and the upcoming Sol all involve Telxius landings or co-investment. The relationship has effectively made Spain a primary European termination point for Google's Atlantic traffic.
• Vodafone is a significant Meta partner — capacity on Amitié, consortium membership on 2Africa, and a long history of consortium cables across Asia and Africa.
• Aqua Comms — a specialist subsea infrastructure operator — provides operational and landing services for several Meta and Microsoft cables in the Atlantic basin.
• African carriers' coalescence around 2Africa — MTN's Bayobab, Vodacom, Orange Africa, and a handful of national operators — has reshaped African connectivity economics in a way that previous WIOCC- and SEACOM-era cables could not.
• Asian carrier consortia remain the norm for trans-Pacific cables: NTT, KDDI, China Mobile International, China Telecom Global, Singtel, Telstra, PLDT, Telin and Chunghwa Telecom each appear repeatedly on hyperscaler-backed Asian cables.

What this concentration means

None of this is academic. For cloud buyers, telecom carriers and the governments that regulate subsea infrastructure, the shift to hyperscaler ownership shows up in a few places:

• Cheaper, faster cloud egress and inter-region replication. Hyperscalers can route traffic on their own cables at marginal cost. This is one reason inter-region bandwidth in AWS, Google Cloud and Azure has fallen so much over the past decade.
• Better resilience for hyperscaler customers. If you're using Google Cloud or AWS, you're effectively riding on a private, multi-path subsea fabric with engineered diversity — usually better than the public internet's average path between the same endpoints.
• Narrower commercial market for capacity wholesale. Traditional capacity wholesale is squeezed; carriers increasingly monetise landing, backhaul and dark-fibre services rather than IRUs.
• Strategic dependency on a small number of operators. Several governments — including in Europe, the UK, Australia and parts of Asia — have begun reviewing subsea cable concentration as a critical-infrastructure question, particularly after high-profile cable cuts in the Baltic and Red Sea.
• National-security review intensity. US CFIUS reviews led to the partial cancellation of Pacific Light's Hong Kong leg; similar reviews now affect routing decisions in the Mediterranean, the Indian Ocean and around the South China Sea. New cables are increasingly designed to avoid politically sensitive landfalls.

The Singapore picture

Singapore is one of the most densely-connected subsea cable nodes in the world — comparable only to Marseille, Fortaleza and a small number of US East/West Coast clusters. Every major Asia-Pacific hyperscaler cable lands here, either as an origin or transit point:

• Echo, Bifrost, Apricot, INDIGO West, SJC2 — all land in Singapore.
• Google operates region in Jurong (asia-southeast1) and has multiple landings at Tuas and Changi cable stations.
• Meta uses Singapore as the Asian anchor for Echo, Bifrost and Apricot, and operates a data centre on the island.
• Local carriers — Singtel, StarHub, M1, NetLink Trust — operate landing-station and backhaul services for these and other consortium cables.
• Singapore's regulator IMDA issues the FBO licences that govern cable landings — see FBO Licences in Singapore for the framework.

The country's combination of regulatory clarity, deep landing-station infrastructure (Tuas, Changi, Katong), and proximity to the Indonesian archipelago — through which most Pacific-bound hyperscaler routes now run — makes it structurally hard to bypass for any Asian cable build. Project Waterworth, while routed primarily across the Indian Ocean and Pacific basins, is expected to interconnect with the Singapore subsea fabric at some level given the country's gateway role.

Where to go next

• How the global internet is plumbed: Global Connectivity: Submarine Cables, IXPs, and BGP — the macro picture that the hyperscaler cables sit inside.
• What it takes to land a cable in Singapore: FBO Licences in Singapore: IMDA's Telecom Licensing Framework.
• Adjacent connectivity context: The Modern Wireless Stack — Wi-Fi, 5G, satellite and where each fits.
• The data centres at each end: Singapore data centre operators in our directory.
• The carriers that partner with the hyperscalers: Telecom and connectivity providers in Singapore.

Anyone using these numbers in a procurement or due-diligence decision should bear in mind that the subsea map moves quickly. Project Waterworth is still in survey. 2Africa's Pearls extension was scheduled to light in 2026 at the time of writing but had not. Several Google cables in the South Pacific and across the Indian Ocean — Tabua, Honomoana, Bulikula, Proa, Taihei among them — are at various stages between announcement and final assembly. Operators publish updates on their own engineering channels; the wider industry tracks them through TeleGeography and the SubmarineCableMap. Treat any specific number here as a snapshot of the moment, not a permanent value.

Find Singapore telecom and infrastructure partners

Looking for a Singapore carrier, landing-station operator, or data centre to anchor your own subsea or cloud connectivity?

Browse Telecom & Connectivity →