How are AI models classified as intangible assets?

Full Explanation

Under IFRS 3 (for acquisitions), AI models are classified within the technology-based intangible asset class, alongside software, databases, and trade secrets. They meet the identifiability criteria because they can be separated from the entity (licensed, sold, or transferred) and arise from contractual or legal rights (copyright, trade secret protection). Under the Corrado-Hulten-Sichel (CHS) framework used for strategic analysis, AI models fall within the computerised information category — specifically as software and databases. However, a well-trained AI model often creates value across multiple CHS categories simultaneously: the model itself is computerised information, the training data may constitute a separate database asset, the model's outputs may strengthen customer relationships (economic competencies), and the research process generates innovative property. This multi-category nature makes AI assets particularly interesting for valuation purposes. When performing a purchase price allocation, each separately identifiable AI-related asset should be valued independently: the model architecture, training data, fine-tuning datasets, and any proprietary inference infrastructure.

Related Glossary Terms

Related Questions