A component whose failure takes down the whole system, the architectural smell HA design exists to eliminate.
A single point of failure (SPOF) is any component, server, database, network link, person, whose failure causes the whole system to fail. SPOF analysis is a foundational architecture review technique: walk every layer, ask 'what happens if this dies', and either remove the SPOF (replication, redundancy) or accept it (with documentation and recovery plan). Common hidden SPOFs include a single load balancer, a single Kafka broker, a single deploy automation account, a single on-call engineer who knows the system.
Outages caused by SPOFs are the most embarrassing kind because they're the most predictable. The component was always single, the team always knew, and one day it failed exactly as expected. SPOF analysis before launch (and after every architectural change) is the cheapest reliability investment because the SPOFs are usually obvious once someone is asked to find them.
See the part of the platform that handles spof (single point of failure) in production.