Claroty analyzed 750,000 cyber-physical systems across large data centers, finding that approximately 18% of infrastructure assets are one network hop away from internet-exposed systems, potentially allowing attackers to reach them. Less than 0.4% of these assets are directly internet-exposed. The vulnerable assets include HVAC, power distribution units, and building management systems, many of which use insecure protocols or outdated firmware. Thousands of devices have known exploited vulnerabilities. The research identifies attack vectors such as insecure communication protocols, unmanaged remote access, and weak authentication. The report stresses that attacks on these operational technologies could disrupt critical data center functions, including cooling, power, and environmental controls. It advises adopting continuous exposure management, zero trust network segmentation, and protocol-aware threat detection to mitigate risks.

The exposure of nearly one in five data center infrastructure assets to potential attack paths increases the risk of disruption to critical data center operations such as cooling, power distribution, fire management, and backup systems. Vulnerabilities in operational technology and building management systems, including known exploited flaws, could be leveraged by attackers to degrade operational resilience and cause service interruptions. This risk affects the availability and reliability of data center services but does not indicate confirmed active exploitation in the wild.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Claroty recommends implementing continuous exposure management to monitor asset connectivity, adopting zero trust network segmentation to limit lateral movement, hardening building management systems by updating firmware and securing protocols, and deploying protocol-aware threat detection to identify anomalous activity. These measures focus on reducing exposure and strengthening defenses around cyber-physical systems in data centers.