Abstract

This paper presents a semantic ontology for high-performance computing (HPC) centers powered by artificial intelligence (AI) and embedded in HPC infrastructures. These systems are complex, with knowledge representation to ensure coherence in their design, efficiency in their operation, and scalability. The ontological architecture has five domains: computational infrastructure, cognitive systems, data intelligence services, research-driven applications, and the innovation ecosystem, with the potential to scale through the incorporation of different subclasses. The ontology was engineered using established methodologies and validated through its deployment in real-world AI-HPC environments, including LUMI (Finland), Summit (USA), except one in Valparaíso, Chile. The model, developed in OWL 2.0 and run with HermiT, enabled semantic inferences about cognitive components, network hierarchies, and traceability of scientific data. The results show that this semantic framework improves cross-system interoperability, supports architectural modeling, and lays the groundwork for standardised AI-HPC integration across scientific domains.