The Cambridge Centre for Alternative Finance has published an 11-year study suggesting that Bitcoin’s peer-to-peer network is far more resilient to random submarine cable outages than many market participants may have assumed. The research concludes that the network can withstand a very high level of simultaneous undersea cable disruption before large-scale node disconnections begin to emerge.
Conducted by Wenbin Wu and Alexander Neumueller, the analysis reviewed network snapshots from 2014 through 2025 and incorporated 68 verified submarine cable fault events. Their simulations indicate that between 72% and 92% of transnational undersea cables would need to fail at the same time to cause substantial node loss, a threshold that materially reframes how physical connectivity risk should be understood.
Random Cable Failures Look Less Dangerous Than Concentration Risk
The study found that more than 87% of real-world cable failures in the dataset affected fewer than 5% of visible nodes. That pattern suggests that ordinary, random disruptions in undersea infrastructure have historically had limited impact on Bitcoin’s visible network footprint.
The researchers also tested whether those outages had any meaningful relationship with market pricing and found little evidence that they did. The reported correlation between submarine cable failures and Bitcoin price movements was near zero, at roughly -0.02, pointing to minimal direct price sensitivity from random cable incidents alone.
At the same time, the paper makes clear that resilience has not remained constant across the years. Geographic concentration in mining and hosting between 2018 and 2021 reduced the network’s robustness, pushing the critical failure threshold down to about 0.72 in 2021 before redistribution lifted it to 0.88 in 2022 and then to 0.78 by 2025.
The More Serious Risk Sits in Hosting and Routing Concentration
Where the findings become more operationally relevant is in the contrast between random failures and targeted disruption. The study argues that a focused attack on a small number of major hosting providers could create far more serious effects than broad but random damage to submarine cables.
According to the simulations, disrupting the five providers with the largest node counts — Hetzner, OVH, Comcast, Amazon, and Google Cloud — could produce network paralysis by removing as little as 5% of their combined routing capacity. That asymmetry shifts the risk conversation away from physical cable redundancy alone and toward dependency on concentrated internet and cloud infrastructure.
The paper also examined Tor adoption and found that it modestly improves resilience. With roughly 64% of sampled nodes using Tor by 2025, the study estimates that Tor raised the critical failure threshold by around 0.02 to 0.10, adding some protection against cable-cut scenarios because relay bandwidth remains concentrated in well-connected jurisdictions.
The bigger vulnerability is not random physical cable damage, but concentrated dependence on a small number of hosting and routing providers, which should now be reflected in operational risk frameworks, continuity testing, and board-level reporting.