ATLAS GRIDWATCH

The subsea cable network faces significant physical resilience challenges due to geographic chokepoints and potential for sabotage. The 'Global Undersea Cable Chokepoints and Landing Clusters' report (2026-01-03) highlights critical vulnerabilities in areas like the Bab-el-Mandeb Strait, Strait of Malacca, and the English Channel. These narrow passages concentrate numerous cables, creating single points of failure. Historical precedents, such as the 2008 Mediterranean cuts and the 2023 Red Sea attacks, demonstrate the real-world impact of cable disruptions. The recent Finnish investigation into suspected cable sabotage in the Baltic Sea (2026-01-02 news) underscores the ongoing threat of deliberate attacks. Cables being only centimeters thick makes them vulnerable. High shipping traffic and shallow channels pose physical risk, especially anchor drags and fishing. The concentration of cables in specific landing stations further compounds the risk, as any disruption at these sites can have cascading effects across the network.

The strategic placement of new AI/Hyperscale facilities is deeply intertwined with great power competition, particularly between the U.S. and China. The 'Global Data Center Density and AI Compute Concentration' report (2025) reveals a significant asymmetry, with the U.S. hosting over half of the world's hyperscale data centers and a dominant share of AI compute capacity. China, while second, lags considerably behind. The planned buildout of AI-focused infrastructure, such as xAI's Colossus 2 in the U.S., and the expansion of cloud regions by major players like Amazon, Google, and Microsoft, reflects the intensifying race to control AI capabilities. The report notes that only 32 countries host specialized AI data centers, implying that 80% of nations globally have zero infrastructure for large-scale AI compute. Recent news highlights Japan's increased defense budget and responses to Chinese naval activity, suggesting an escalating geopolitical landscape in which control over AI infrastructure could be a critical factor. Considering also the mention of Equinix building AI-ready data centers in South Africa, it shows the continued expansion of 'the West' into new digital territories. This expansion could be viewed in the context of the Digital Silk Road, with China seeking to establish alternative digital infrastructure and influence in developing countries to rival Western dominance.

Power and cooling constraints represent significant vulnerabilities for data centers and AI infrastructure. The 'Global Data Center Density and AI Compute Concentration' report (2025) emphasizes the increasing power density of new facilities driven by AI workloads. This creates a dependency on reliable and high-capacity power grids. Kinetic vulnerabilities, such as substations and water supply infrastructure, could be exploited to disrupt data center operations. A coordinated attack on these physical assets could have devastating consequences. The 'Compute Divide' mentioned in the report highlights that only a few countries hold the lion's share of AI-oriented computing power, leading to asymmetric capabilities and dependencies on specific regions for compute resources. This increases risks, as disruptions in these key regions can have global ramifications. Additionally, the concentration of AI compute in the hands of a few companies (e.g., Amazon, Microsoft, Google) creates a choke point in the supply chain for AI services. Any disruption to their operations could have cascading effects across various sectors. Vatajankoski and E-Heat's move to plug containerized data centers into local heating networks shows the industry push to use waste energy. However, the dependence of substations of these data centers makes them prime targets.

Alternative View: What if the perceived threat of sabotage is a deception, designed to mask other activities or divert attention from other vulnerabilities? For example, a sophisticated cyber-attack could be underway while all eyes are on physical security. What if the 'safe' jurisdictions, like the U.S. and Europe, are actually compromised by insider threats or sophisticated long-term espionage campaigns? The concentration of data in these regions could make them attractive targets for adversaries seeking to steal intellectual property or disrupt critical infrastructure. The reliance on a small number of hyperscalers also creates a single point of failure. A successful attack on one of these companies could have widespread consequences, regardless of the geographic location of their data centers. The focus on great power competition may be overshadowing the risks posed by non-state actors, such as ransomware groups or hacktivists, who could exploit vulnerabilities in the AI supply chain or data center infrastructure.