Glossary term
Total cost of ownership
The full lifecycle cost of an AI system: model or API usage, compute, data, tooling, integration, security, operations, and human oversight.
Token pricing alone systematically understates this. The FinOps Foundation's own guidance puts the surrounding infrastructure — vector databases, embeddings, rerankers, orchestration, observability — at 40 to 60% of total feature spend in RAG and agentic architectures, and notes that most of that harness cost does not flow through the same proxy layer that handles token attribution.
Related research