Accounting for Cloud AI GPU Capacity Contracts Under IFRS 16: A Proposed Recognition and Measurement Framework
Main Article Content
Abstract
The advent of generative Artificial Intelligence (AI) and Large Language Models (LLMs) has led to a substantial demand for specialised high-performance computing resources, particularly Graphics Processing Unit (GPU) clusters. Enterprises are moving away from buying physical hardware outright and instead are entering into multi-year cloud-service contracts for computing capacity with hyperscale providers. In practice, there are significant challenges in applying existing financial reporting requirerments, and preparers often classify these arrangements as executory service contracts (operating expenses), based on the assumption that software-virtualisation and standardized contractual substitute clauses preclude lease accounting. This paper sets out a thorough Recognition and Measurement Framework for cloud AI GPU capacity contracts under International Financial Reporting Standard 16 (Leases). Specifically, it formalises (1) a proposed decision-tree model that prescribes assessment of lease existence prior to accounting policy election consideration; (2) an objective approach to separating bare-metal-hardware lease components from non-lease services (orchestration, cooling and network); (3) an initial measurement model for duration-matched Incremental Borrowing Rates (IBR) and take-or-pay capacity floors; and (4) subsequent measurement mechanics to address rapid technological obsolescence and IAS 36 impairment testing. Methodologically, the study adopts a descriptive-analytical approach, supported by a normative accounting framework and scenario-based numerical analysis to evaluate the recognition and measurement of cloud AI GPU capacity contracts under IFRS 16. The framework is tested with the help of an illustrative numerical application in three representative scenarios (Dedicated Cluster, Shared Substitutable Capacity and Hybrid Minimum Commitment) together with full journal-entry postings, reconciled financial statements extracts, and financial-ratio analyses. The findings show that requiring qualifying GPU contracts to be capitalised makes previously off-balance-sheet lease obligation visible and improves reporting transparency. The study recommends that entities assess cloud AI GPU capacity contracts based on the substantive lease identification criteria of IFRS 16, particularly asset identification, supplier substitution rights, and control over asset use, rather than relying solely on contractual classifications.
Downloads
Article Details

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.