Infrastructure
When the cable breaks: designing for connectivity that will fail
6 August 2026 · 7 min read · 2 public sources

On 14 March 2024, four submarine cables — WACS, ACE, MainOne and SAT-3 — failed in close succession, and ten countries lost or badly degraded their international connectivity. Carnegie’s analysis of the incident records the repair economics plainly: a single cable repair costs between $1 million and $3 million, the four-cable event came to roughly $8 million, and only three repair vessels serve the region, with the first responder arriving about three weeks after the fault.
A landlocked position is a different risk, not a smaller one
Uganda reaches the sea through terrestrial fibre crossing neighbouring countries to landing stations on either coast. That means a fault far offshore is felt here, and it also means the transit route itself — the cross-border link, the power at an intermediate site, the commercial relationship with a transit provider — is part of your availability story. Buying two circuits from two vendors that quietly share the same transit corridor is a common and expensive way to discover this.
Questions that establish whether you actually have diversity
- Which submarine cable systems does each of my providers ultimately ride on, and do any of them overlap?
- Do my two terrestrial routes leave the country through different borders, or through the same one on different fibres?
- What is contracted: a restoration target, or best effort with an apology?
- When the primary path fails, does failover happen automatically, and has it been tested with traffic rather than on paper?
- Which systems must keep working on a degraded link, and have they been tested at that bandwidth?
Design for the degraded day
Full redundancy is expensive and rarely complete. What is achievable is deciding, in advance, what runs on a fraction of the usual capacity. That means identifying which applications are essential, whether they can operate on a cached or offline mode, whether staff can be productive without the large collaboration platforms, and how customers will be told. A satellite or wireless standby that carries payments and core communications is a modest cost for a scenario measured in weeks.
The cable will be repaired. The question a plan answers is what the organisation does during the three weeks the ship is still on its way.
Sources and further reading
This article summarizes publicly available research. Source findings retain their original geographic and sector scope.
- [01]Beneath the Waves: Addressing Vulnerabilities in Africa’s Undersea Digital InfrastructureCarnegie Endowment for International Peace · 2025
- [02]2024 West Africa Submarine Cable Outage ReportInternet Society · 2024
Put this thinking to work
Tell us about your estate and we’ll map what to build, secure, or fix first.
Keep reading
More insights

Cyber security
Business email compromise: verify the payment, not the email
11 August 2026 · 7 min read

Infrastructure
AI’s electricity bill, and why power now decides where compute goes
10 August 2026 · 8 min read
