The surprising resilience of Bitcoin to damage to physical infrastructure
I read this study from Cambridge and honestly it is more interesting than I expected. Researchers compared 11 years of Bitcoin network data to 68 real-world submarine cable faults. What they discovered was quite surprising: Bitcoin can apparently handle 72% to 92% of the world’s undersea cable outages at one time before things get really bad.
That’s a lot of cables. I mean, think about it: we’re talking about most physical connections between continents collapsing simultaneously. The network is not just collapsing. It gradually degrades, which is perhaps what one hopes for in a decentralized system.
They ran thousands of simulations and random cable failures had virtually no impact. More than 87% of the real cable faults studied affected less than 5% of the nodes. Even this large event off the coast of the Ivory Coast in March 2024, which damaged 7-8 cables at a time, only took down about 0.03% of the world’s Bitcoin nodes.
The vulnerability of targeted attacks
This is where it gets worrying. While random failures require massive scale to matter, targeted attacks are a different story. If someone knows what they’re doing and goes after the right cables – the ones that serve as choke points between continents – the threshold drops to just 20%.
But the real vulnerability isn’t even in the cables. They are hosts. The study found that targeting just five hosting providers – Hetzner, OVH, Comcast, Amazon and Google Cloud – requires removing just 5% of routing capacity to achieve significant impact.
It’s a completely different threat model. Random cable outages are comparable to natural disasters or accidents. But going to specific hosting providers? This is something a state actor could do, or a coordinated regulatory action. It makes you think about how centralized some aspects of the network could be.
How Bitcoin’s Resilience Has Changed Over Time
The study tracked the evolution of this resilience, and it was not a straight line. Bitcoin was actually most resilient during its early years, from 2014 to 2017. Then things got worse between 2018 and 2021, as the network grew but became more geographically concentrated.
The lowest point was reached in 2021, during the peak of mining concentration in East Asia. China’s mining ban forced redistribution, which contributed to the recovery. But it is interesting to see how network geography affects resilience.
The TOR surprise
This is perhaps the most counterintuitive finding. Around 64% of Bitcoin nodes use TOR in 2025, making their physical location difficult to determine. One might think that this hides a fragility: if all these TOR nodes were grouped in the same place, the network could be more vulnerable than it seems.
But the opposite seems true. TOR relay infrastructure is concentrated in Germany, France and the Netherlands – countries with extensive connectivity via submarine cables and land borders. An attacker attempting to disrupt TOR by cutting cables faces a complex problem, as these countries are among the most difficult to disconnect.
Researchers have found that TOR actually adds to network resilience. It’s as if the Bitcoin community has moved toward censorship-resistant infrastructure without central coordination, and this shift has also made the network physically more difficult to disrupt.
With current tensions in places like the Strait of Hormuz, these questions are not just academic. The study suggests that random damage to the cables is unlikely to break Bitcoin. But targeted attacks on specific infrastructures? That’s a whole different story. This raises questions about the balance between decentralization and practical operation of the network.
