The private sector is already routing around the risk
On 7 October 2026, Rest of World reported that Google and Meta have begun sending some live traffic over a land route across Iraq that they had held in reserve. Google also bought a pair of fiber strands along Turkish state pipelines in September. A source told the outlet it paid two to three times the roughly 6 million euros ($7 million) that newer lines crossing Turkey are expected to cost. Microsoft said on 23 September that it plans to invest more than $400 million in subsea and land links across the Middle East by 2030.
The trigger is the fighting at Bab al-Mandab. Rest of World reports that more than 90% of data capacity between Europe and Asia runs through cables under the Red Sea, and that the strait is less than 32 kilometers wide. The Houthis seized Yemen's side of it in September, including the island of Mayyun. On 5 October, Saudi-backed government forces said they had retaken key positions near the strait and the port of Mocha. The Houthis denied this.
The case for treating this as a public-law problem
The strongest argument for heavier government intervention is that a cable cut is not just a commercial loss. Banks, hospitals and governments in Gulf, South Asian and East African states depend on these routes. Private operators will under-invest in redundancy that mostly benefits others. The risk is not hypothetical. In September 2025, The Register reported that Microsoft warned Azure customers of higher latency after undersea fiber cuts in the Red Sea, and that NetBlocks identified the SMW4 and IMEWE systems as affected. Microsoft rerouted traffic over alternate paths. Rest of World adds that three Red Sea cables were damaged in 2024, apparently by the anchor of a ship disabled by a Houthi missile, and that repairs took months.
If that logic holds, one would expect international law to supply a backstop. It does not do so for this scenario.
What the law actually covers
The main treaty provision is Article 113 of the UN Convention on the Law of the Sea. It requires each state to make it an offence for its flagged ships or nationals to break or injure a high-seas cable, "wilfully or through culpable negligence". It is a flag-state criminal-law duty, so it depends on each state's own enforcement. It does not give anyone a way to protect a cable from a non-state armed group controlling a coastline, and it does not make repair ships safe in a war zone. Academic commentary notes that Article 113 is derived from the 1884 Cable Convention, that many states have implemented it weakly, and that it creates no obligation to protect cables in the territorial sea. That is where much of the damage occurs.
The diplomatic track is further along but still soft. The International Telecommunication Union's 2025 International Submarine Cable Resilience Summit in Abuja endorsed the Abuja Declaration, a collective commitment by governments, industry and international organizations. It also set up three working groups, on timely deployment and repair, on risk identification and monitoring, and on connectivity and geographic diversity. The ITU's own page flags a need for better data on cable disruptions. That is a useful admission: the field lacks even a shared fault dataset. A declaration and working groups are worth having, but nothing in them stops a missile or compels a repair crew to sail.
Why the market response is the right one
The hyperscalers' move shows how the problem is actually being fixed. Rest of World reports that Google, Meta, Microsoft and Amazon use about three-quarters of the world's international internet bandwidth, according to TeleGeography. They are among the most exposed parties and have the strongest incentive to pay for diversity. They are paying a 2-3x premium for Turkish pipeline fiber because the alternative is an outage with a months-long repair window. Rest of World also notes that a normal repair takes one to two weeks if a ship is available, but that crews in contested waters need assurances they will not be fired upon.
The lesson for policy is to avoid mandates that crowd out this behavior. Three principles follow.
- Make land and terrestrial transit easy to buy. Iraq's 10 September contract with Qatar's Ooredoo to carry data toward Europe from the port of Al-Faw is the kind of deal governments should encourage. Transit states should publish clear, non-discriminatory terms for crossing pipelines, rail corridors and borders, and should avoid surcharges that turn a backup route into a tollbooth.
- Speed up permits and repair authorizations. The Abuja working group on deployment and repair points at the real bottleneck. Streamlined landing licences, cross-border repair-ship access and pre-agreed safe-passage procedures cost little and shorten outages.
- Do not force localization or route control. Sovereignty-driven rules that dictate which cables carry which traffic would reduce the flexibility operators need when a corridor closes. Redundancy comes from many independent paths, not from national routing mandates.
The limits of the land-route fix
Land routes are not a complete substitute. They carry much less capacity than the Red Sea corridor and cross territory with their own political risks, from Iraq's security conditions to dependence on state-owned pipeline operators. Rest of World reports that experts expect these routes to stay dormant until a sea cable is cut. That makes them insurance, not a replacement. The insurance premium is also uneven: operators with the capital to buy diverse capacity are protected, while smaller regional carriers and the countries that depend on them are not.
That gap is where public policy can add the most value. Development banks and regional bodies can co-finance diverse landings and terrestrial links for smaller markets. Governments can also share fault and threat data, which the ITU itself identifies as a gap. Both measures leave routing decisions with the operators who understand the traffic. They also avoid the costly, slow and often counterproductive step of trying to regulate a war zone.
Bottom line
The Bab al-Mandab crisis shows that cable resilience is a geography and logistics problem before it is a legal one. Treaties and declarations set norms, but diversified routes and fast repair determine whether the internet stays on. Policy should lower the cost of building and buying that diversity.