99% by cable.
63 ships to fix it.
Internet fragmentation layers a hard physical chokepoint problem on a policy one. The Red Sea–Egypt corridor carries ~17% of global traffic through a single route, repaired by a global fleet of ~63 aging vessels.
Internet fragmentation has a hard physical chokepoint problem layered on a policy one. VegaReady tracks two layered risks: physical (submarine-cable chokepoints, repair capacity, LEO backup limits) and policy (data localization, internet controls, kill-switch risk). The chokepoint problem mirrors the maritime one — the Red Sea–Egypt corridor carries ~17% of global internet traffic and >90% of Europe–Asia bandwidth through a single corridor, while the global repair fleet is just ~63 aging vessels. Submarine-cable sabotage has become a normalized grey-zone tactic with a verifiable incident record.
For the stablecoin layer of digital-dollar infrastructure - reserves, the GENIUS Act T-bill bid and emerging-market dollarization - see Stablecoins: the dollar's private plumbing →
The global internet runs on ~600 submarine cables (~1.5M km) carrying ~99% of intercontinental data, repaired by a global fleet of only ~63 aging vessels at $1–3M per repair and months of lead time. Roughly 150–200 faults occur annually, 70–80% accidental. The Red Sea–Egypt corridor carries ~17% of global internet traffic and >90% of Europe–Asia bandwidth; the September 6, 2025 Jeddah event cut four cables (SEA-ME-WE-4, IMEWE, FALCON GCX +1), three needing ~5 months and one still out 7 months later.
LEO satellite is not a backup at scale. A single-blog estimate puts total global satellite capacity at ~50 Tbps by 2026 versus ~8,750 Tbps subsea, but the exact ratio is T3 and should be treated as illustrative. The usable monitoring signal is simpler: subsea cables remain the primary global-capacity layer, while LEO systems support emergency prioritization, remote access and outage triage rather than full traffic substitution.
Global internet freedom declined for the 14th consecutive year in Freedom House's Freedom on the Net 2024 (72 countries); China and Myanmar tied as worst, Iran 3rd most repressive, Iceland freest, and 48% of internet users live in countries where authorities disconnected networks. Data localization is proliferating: ~100 measures across 40 countries by early 2023, >two-thirds combining storage requirements with flow prohibition (OECD).
| Region | Control | Localization | Cable redund. | Sat backup | Kill-switch | Composite | Notes |
|---|---|---|---|---|---|---|---|
| China | 5 | 5 | 2 | 2 | 4 | 3.6 | Domestic-internet isolation; own LEO build |
| Russia | 5 | 4 | 3 | 3 | 5 | 4 | RuNet kill-switch demonstrated |
| Egypt | 4 | 3 | 1 | 4 | 4 | 3.2 | ~17% global traffic transits; chokepoint |
| Pakistan | 4 | 3 | 2 | 4 | 4 | 3.4 | Frequent shutdowns; few landing points |
| India | 3 | 4 | 3 | 3 | 4 | 3.4 | World-leading shutdown count (J&K) |
| Gulf (Saudi/UAE) | 4 | 3 | 2 | 3 | 3 | 3 | [PROVISIONAL-2026] cable force majeure |
| EU | 1 | 4 | 4 | 3 | 1 | 2.6 | Localization is regulatory, not coercive |
| US | 1 | 2 | 5 | 1 | 1 | 2 | Lowest fragmentation risk |
| Singapore | 2 | 3 | 3 | 3 | 2 | 2.6 | Critical Asia interconnection node |
| East Africa | 3 | 2 | 1 | 4 | 3 | 2.6 | High Bab el-Mandeb cable exposure |
Scored 1–5, 5 = highest fragmentation/kill-switch risk. Control and localization columns are anchored to Freedom House and OECD data; redundancy/backup/kill-switch are analyst estimates (T3). Russia (RuNet) and China (Great Firewall) score highest on coercive control; Egypt and East Africa are the physical-chokepoint extremes; the US scores lowest overall.
| Incident | Date | Detail | Source |
|---|---|---|---|
| Baltic — C-Lion1 / BCS East-West | Nov 17–18, 2024 | Two cables severed; restored Nov 28; anchor-drag attribution amid NATO scrutiny | Wikipedia |
| Taiwan Strait — Matsu Islands | 2023 (5-yr record) | >20 cable cuts in five years; Matsu left 50+ days offline in 2023 | The Diplomat |
| Red Sea — Jeddah | Sep 6, 2025 | 4 cables cut (SEA-ME-WE-4, IMEWE, FALCON GCX +1); 1 still down 7+ months later | TeleGeography/ICPC |
Submarine-cable sabotage has become a normalized grey-zone tactic with a verifiable incident record — anchor-drag and deliberate cuts across the Baltic, Taiwan Strait and Red Sea. With ~150–200 faults annually (70–80% accidental), the tracking signal is the concentration of deliberate cuts at chokepoints, not the raw fault count.
01 Red Sea–Egypt Cable ChokepointIndex~17% of global internet traffic and >90% Europe–Asia bandwidth via a single corridor; ~63 repair vessels globally; Sep 2025 Jeddah cut left 1 of 4 cables down 7+ months. medium
02 LEO Satellite vs Subsea Capacity GapIndexGlobal satellite capacity is directionally far smaller than submarine-cable capacity; one T3 estimate puts satellite at ~50 Tbps by 2026 vs ~8,750 Tbps subsea. LEO is useful for emergency prioritization, not full substitution. low
HIGH — Freedom House FOTN (14th-year decline, 48% disconnect figure); OECD data-localization count; TeleGeography cable/vessel counts; CSIS Jeddah cable-cut record.
MODERATE — Red Sea ~17% traffic share (denominator differs from the >90% Europe–Asia bandwidth figure — both reported with denominators); cable-count revisions (~530/570/600).
Quarantined — LEO-vs-subsea capacity ratio (~175×) and Starlink subscriber counts (single-blog T3, illustrative only); scorecard redundancy/backup/kill-switch columns (analyst T3).
Related: The cable-weaponization investment angle (Starboard, repair-ship scarcity, mine-clearing) lives on /markets/defense; the chokepoint geography on /structural/chokepoints; the cable_severance scenario across the corpus.