Content reviewed and verified by Graham Chee, with FCPA-led practice at Local Knowledge, Mascot NSW. Continuous CPA Australia member since 1986. Prior career at Goldman Sachs, BNP Investment Management and Merrill Lynch.. Last reviewed August 2026. Next review scheduled for November 2026.
TL;DR
Unlock the balance sheet potential of your proprietary AI: expert guidance on AASB 138 for Australian businesses.
Introduction: Accounting for AI as a Self-Created Asset
The rapid advancement of Artificial Intelligence (AI), particularly Large Language Models (LLMs), presents Australian businesses with unprecedented opportunities. While many organisations leverage AI through Software-as-a-Service (SaaS) subscriptions, a growing number are investing in developing proprietary AI models. This strategic shift from 'buying' to 'building' AI creates a critical accounting question: how should the significant costs associated with internal AI development be treated on the balance sheet? This article, guided by Principal Advisor Graham Chee (FCPA, GRCP), delves into the technical accounting treatment under Australian Accounting Standards Board (AASB) 138 'Intangible Assets' for businesses capitalising internal LLM development costs. We move beyond the operational efficiency narrative of AI to dissect its balance sheet implications as a self-created intangible asset. You will learn to distinguish between research and development phases, understand the stringent capitalisation criteria, explore valuation methodologies, and navigate the practical implications and common pitfalls in accounting for your custom AI intellectual property.
Beyond SaaS: Why Proprietary AI Triggers AASB 138 Considerations
For many Australian businesses, AI adoption begins with readily available SaaS solutions. These typically involve subscription fees treated as operating expenses, with no complex balance sheet implications. However, when a business undertakes the internal development of a custom AI model, such as a proprietary LLM tailored to its specific operations or customer base, the accounting landscape shifts dramatically. These custom models, unlike off-the-shelf software, are designed to provide future economic benefits, often through enhanced efficiency, new revenue streams, or protected intellectual property. As such, they are not merely operational tools but potential intangible assets. AASB 138 'Intangible Assets' provides the framework for recognising and measuring internally generated intangible assets, which includes certain software development costs. The decision to capitalise these costs, rather than expensing them, can significantly impact a company's financial statements, affecting profitability, asset base, and ultimately, investor perception. Understanding this distinction is crucial for businesses aiming to accurately reflect the value of their innovation.
The AASB 138 Framework: Distinguishing Research from Development in AI
Meeting the Capitalisation Criteria for Internally Generated AI Assets
For development costs of an internally generated intangible asset like a custom LLM to be capitalised, AASB 138 sets out six stringent criteria that must all be met. These criteria are designed to ensure that only assets with a high probability of generating future economic benefits are recognised on the balance sheet. For AI development, demonstrating these can be complex:
- Technical Feasibility: The entity must demonstrate the technical feasibility of completing the intangible asset so that it will be available for use or sale. For an LLM, this means proving the model can function as intended (e.g., generate coherent text, perform specific tasks) within realistic constraints.
- Intention to Complete and Use/Sell: Management must have the intention to complete the intangible asset and use or sell it. This requires clear strategic alignment and business plans for the AI model.
- Ability to Use or Sell: The entity must demonstrate its ability to use or sell the intangible asset. This involves considering market demand, internal infrastructure, and necessary operational changes.
- Future Economic Benefits: The intangible asset must be able to generate probable future economic benefits. For an LLM, this could be cost savings, new revenue streams, or competitive advantage. This often requires a robust business case and financial projections.
- Availability of Resources: Adequate technical, financial, and other resources to complete the development and to use or sell the intangible asset must be available. This includes access to talent, computing power, and funding.
- Reliable Measurement of Cost: The entity must be able to reliably measure the expenditure attributable to the intangible asset during its development. This necessitates meticulous cost tracking, including direct labour, materials, and allocated overheads directly related to the AI development project.
Failure to meet even one of these criteria means the expenditure must be expensed as incurred. The complexity of AI development often makes the 'technical feasibility' and 'future economic benefits' criteria particularly challenging to satisfy early in the development cycle. [AASB 138.57]
Valuing Your Custom AI: Challenges and Methodologies Under AASB 138
Once the capitalisation criteria are met, the next challenge is to reliably measure the cost of the internally generated AI asset. AASB 138 specifies that the cost of an internally generated intangible asset is the sum of expenditure incurred from the date when the asset first meets the recognition criteria. This includes all directly attributable costs necessary to create, produce, and prepare the asset to be capable of operating in the manner intended by management. For an LLM, this can encompass:
- Direct Labour Costs: Salaries and wages of data scientists, AI engineers, software developers, and project managers directly working on the LLM development.
- Computing Resources: Costs associated with cloud computing, GPU clusters, and other infrastructure specifically used for model training, fine-tuning, and testing.
- Data Acquisition and Preparation: Costs of obtaining, licensing, cleaning, and labelling data sets essential for training the LLM.
- External Consultancy: Fees paid to external experts for specialised AI development, algorithm design, or technical reviews.
- Directly Attributable Overheads: A reasonable and consistent allocation of overheads (e.g., depreciation of equipment used, facility costs) that are directly related to the development activity.
Costs that cannot be directly attributed or reliably measured, or those incurred during the research phase, must be expensed. Subsequent expenditure on an internally generated intangible asset is added to its carrying amount only if it enhances the economic benefits embodied in the asset. Otherwise, it is expensed. The initial measurement of the asset is at cost. Post-recognition, the entity must choose either the cost model or the revaluation model for the entire class of intangible assets. Given the nascent nature of AI asset markets, the cost model is typically more appropriate for internally generated AI. [AASB 138.65-68]
Practical Implications: Balance Sheet Impact and Disclosure Requirements
Capitalising AI development costs has significant implications for an Australian business's financial statements. On the balance sheet, it increases the reported asset base, potentially improving financial ratios like return on assets. However, these assets are subject to amortisation over their useful life, which will impact future profit and loss statements. Determining the useful life of an AI model can be challenging, given the rapid pace of technological change. Factors to consider include obsolescence, market demand, and the expected period over which the model will generate economic benefits. Amortisation should reflect the pattern in which the asset's future economic benefits are consumed. If this pattern cannot be reliably determined, the straight-line method is used. [AASB 138.97-98]
Furthermore, AASB 138 mandates extensive disclosure requirements for intangible assets. Entities must disclose for each class of intangible assets, distinguishing between internally generated and other intangible assets: the useful lives or the amortisation rates used; the amortisation methods used; the gross carrying amount and accumulated amortisation (aggregated with accumulated impairment losses) at the beginning and end of the period; and a reconciliation of the carrying amount at the beginning and end of the period showing additions, disposals, amortisation, and impairment losses. For internally generated intangible assets, specific disclosures are required regarding the nature of the asset and the amount of expenditure recognised as an expense during the period in relation to research and development activities. Transparency in these disclosures is crucial for stakeholders to understand the true investment in AI innovation. [AASB 138.118-126]
Navigating the Nuances: Common Pitfalls in AI Intangible Asset Accounting
Accounting for internally generated AI assets under AASB 138 is fraught with potential pitfalls that can lead to misstatements or non-compliance. Australian businesses must be acutely aware of these to ensure accurate financial reporting:
- Misclassifying Research and Development: This is arguably the most common error. Expensing development costs that meet capitalisation criteria understates assets and overstates expenses, while capitalising research costs overstates assets and understates expenses. Clear internal policies and project stage gates are essential. [ATO: TR 2015/3, 'Income tax: research and development tax incentives - core technology activities']
- Inadequate Documentation: Without robust documentation supporting the six capitalisation criteria, auditors will likely challenge capitalised amounts. This includes detailed project plans, technical feasibility assessments, market analyses, and precise cost tracking.
- Over-optimistic Future Economic Benefits: Projecting future economic benefits for novel AI can be subjective. Management must ensure these projections are realistic, supported by verifiable data, and regularly reviewed for impairment indicators.
- Incorrect Determination of Useful Life: Assigning an excessively long useful life to an AI asset can lead to understated amortisation and overstated asset values. The rapid evolution of AI technology means useful lives may be shorter than for traditional software.
- Failure to Test for Impairment: Intangible assets, including AI, must be tested for impairment annually, or more frequently if there are indicators of impairment. A significant technological breakthrough by a competitor, or a failure of the AI model to perform as expected, could trigger an impairment loss. [AASB 136, 'Impairment of Assets']
- Non-Compliance with Disclosure Requirements: Omitting or inadequately detailing the required disclosures for internally generated intangible assets can lead to audit qualifications and reduced transparency.
Proactive engagement with experienced accounting professionals is critical to navigate these complexities and ensure compliance with Australian accounting standards.
Strategic Financial Planning for AI Innovation: An FCPA Perspective
From an FCPA perspective, the decision to invest in and account for proprietary AI development goes beyond mere compliance; it's a strategic financial planning imperative. Businesses developing custom AI models are not just building technology; they are creating valuable intellectual property that can drive long-term growth and competitive advantage. Proper accounting treatment under AASB 138 ensures that the true investment in this innovation is reflected on the balance sheet, providing a more accurate picture of the company's financial health and future potential. This can be particularly important for attracting investors, securing financing, or even in future M&A activities, where the value of such intangible assets can be a significant determinant of enterprise value.
An FCPA-led practice brings institutional-grade experience to help owner-operated SMEs and founder-led businesses navigate these complex accounting standards. This includes assisting with the establishment of robust internal controls for project cost tracking, developing defensible methodologies for distinguishing research from development, and preparing comprehensive financial disclosures. Strategic financial planning for AI innovation also involves considering the interplay with other incentives, such as the Australian R&D Tax Incentive, which can provide significant cash flow benefits for eligible development activities. A holistic approach ensures that the business not only complies with accounting standards but also maximises the financial and strategic value of its AI investments. [business.gov.au: R&D Tax Incentive]
About the Author
Graham Chee, FCPA, CPA, GRCP, GRCA
Principal and Founder, Local Knowledge
Graham Chee is the principal and founder of Local Knowledge, an FCPA-led Australian practice that brings institutional-grade compliance, investment-structure and intellectual-property experience directly to owner-managed businesses. Graham is a Fellow of CPA Australia (FCPA since November 2005, continuous CPA member since 1986) and holds the OCEG Governance, Risk & Compliance Professional (GRCP) and Governance, Risk & Compliance Auditor (GRCA) designations. His prior career includes senior roles at Goldman Sachs, BNP Investment Management and Merrill Lynch. Graham was previously portfolio manager of the Asian Masters Fund (IPO December 2007 – 31 December 2009), which returned +29% in AUD terms versus the MSCI Asia Pacific (ex Japan) benchmark. He signs off on 100% of client files personally.
Areas of Expertise:
Industry-specific insights
This article is especially relevant to these industries. See how we tailor our services for each.
Contact Us Today
This article provides general information only and does not constitute financial or accounting advice. It is essential to speak with us for advice specific to your situation. Every file is signed off by our principal under the CPA Code of Ethics to ensure the highest standards of professional conduct and integrity.
Graham Chee FCPA, CPA, GRCP, GRCA · Principal, Local Knowledge · Mascot NSW · CPA-signed files