IAS 38 R&D Costs, Software Capitalisation and Indian IT Sector
Author
Sai Manikanta Pedamallu
Published
Reading Time
19 min read
Table of Contents
IAS 38 R&D Costs, Software Capitalisation and Indian IT Sector
By Sai Manikanta Pedamallu (ACCA, CMA US, CSCA US, CGMA, ACMA, Dip IFRS, M.Com, MBA, MA)
Lead Instructor, Global Fin X | www.globalfinx.in/manikanta
Post 36 covered the IAS 38 recognition framework and the internally generated problem. This post applies those principles to the two areas where they create the most operational complexity in Indian practice: research and development expenditure in the pharmaceutical sector, and software capitalisation in the IT services and product sectors. I also cover the AI development cost question, which is becoming practically relevant faster than standard-setters expected.
R&D in Indian Pharmaceuticals: Where the Line Gets Drawn
India's pharmaceutical industry invests significantly in R&D, spanning generic formulation development, novel drug synthesis, biosimilar development, and increasingly, new chemical entity (NCE) research. How this spend maps to the IAS 38 research-versus-development distinction is not always obvious, and the answer has direct earnings consequences.
Mapping Drug Development Stages to IAS 38
The drug development timeline in pharma does not come with a clean label saying "research" or "development." The IAS 38 assessment requires judgment at each stage.
Discovery and early preclinical (always research): Identifying candidate molecules, screening for biological activity, synthesising variants, early animal studies. At this stage, no entity can demonstrate technical feasibility or probable future economic benefits. All expenditure is expensed as incurred.
Late preclinical through Phase I clinical trials (almost always research): Toxicology studies, pharmacokinetic profiling, first-in-human dose-escalation studies. The failure rate across all drugs entering clinical development exceeds 90%. Probability of future economic benefits cannot be demonstrated. Expensed.
Phase II clinical trials (judgment required): Proof-of-concept studies testing efficacy in a small patient population. For innovative drug companies, Phase II still carries very high failure rates, and most auditors and standard-setters take the position that the six IAS 38 development criteria cannot typically be met until Phase II completion demonstrates sufficient efficacy signals. For generic formulations (where technical feasibility is demonstrable much earlier), the switchover may occur earlier.
Phase III and regulatory submission (development phase likely begins for generics; still judgment for NCEs): For generic formulations, where technical feasibility is more predictable and the regulatory pathway is established, the switchover from research to development typically occurs when Phase III bioequivalence studies commence and all six criteria can be demonstrated. For NCEs, Phase III failure rates remain significant, and many companies continue to expense on grounds that probability of future economic benefits cannot be reliably established.
Post-approval development: Line extensions, new dosage forms, new indications. These are typically development costs that meet all six criteria, capitalised from the point the new development programme begins with demonstrably probable commercial success.
Indian Pharma Practice: A Sector-Level Observation
Most large Indian pharmaceutical companies, including Sun Pharma, Dr. Reddy's, Cipla, and Lupin, have historically expensed the majority of their R&D costs even through what might technically qualify as development phases under IAS 38. The reasons are practical: the six-criteria test is demanding to document rigorously, the failure rate for even Phase III studies means "probable future economic benefits" requires careful judgement, and capitalising development costs creates amortisation charges that persist even when the drug fails.
Some companies do capitalise a portion of development expenditure, primarily post-Phase III for generic formulations. The capitalised amounts are then amortised over the estimated commercial life of the product, beginning when the product is approved and launched.
The income statement effect of this choice is significant. Two pharmaceutical companies with identical R&D programmes can report materially different operating profits depending on whether they capitalise development costs or expense them. Analysts reading Indian pharma financials must check the accounting policy note carefully.
The Generic vs NCE Distinction in Practice
The IAS 38 application is genuinely different between generic formulations and new chemical entities.
For generics, the science is largely established. You are proving that your formulation is bioequivalent to the reference listed drug. Technical feasibility is demonstrable relatively early. Market existence is established (the reference drug already sells). The main uncertainties are regulatory approval timing and competitive intensity.
For NCEs, you are proving that a previously untested compound works safely and effectively in humans. Technical feasibility cannot be demonstrated until Phase III completion at the earliest, and even then the probability of commercial benefits depends on comparative efficacy data against existing treatments. The IAS 38 bar is much harder to clear for NCE development.
Software Capitalisation: The Three-Zone Framework
IAS 38 has no specific guidance for software. It applies the general internally generated intangible asset framework. In practice, accounting literature and standard industry interpretation have converged on a three-zone approach for internally developed software.
Zone 1: Preliminary Project Phase (Research Equivalent)
Activities in the preliminary phase include:
Conceptual formulation of alternatives. Evaluation of alternatives. Determination of the existence of needed technology. Final selection among alternatives.
These activities are research-equivalent. All costs are expensed. Technical feasibility has not been established; the entity has not yet committed to a specific approach.
Zone 2: Application Development Phase (Development Equivalent)
Once technical feasibility is established and the entity commits to a specific design and approach, the development phase begins. Costs capitalisable from this point include:
Design of the chosen path. Coding. Installation of software. Testing, including parallel processing.
The six IAS 38 criteria must be demonstrably met at the start of this phase. For most software projects, they will be, provided:
Technical feasibility: the chosen technical approach is proven to work (perhaps a prototype exists, or the technology stack is established). Intention and ability: the entity is committed to completing the project and will use the resulting software internally or sell it. Future economic benefits: the software will generate revenue (for software products) or cost savings (for internal-use software). Resources: developers are assigned, budget is approved. Measurement: project accounting tracks development hours by project.
Zone 3: Post-Implementation (Maintenance Equivalent)
After the software is available for use, costs incurred to maintain it, fix bugs, make minor adjustments, and keep it running are expensed. These are the equivalent of maintenance and repair costs on PPE.
The boundary between development and post-implementation is where judgment gets difficult. Adding a new module with distinct functionality to an existing system is development (capitalise if criteria met). Patching a security vulnerability is maintenance (expense). Migrating the system to a new server is maintenance. Rebuilding the system on a new architecture is development.
Indian IT Services: Why Most Software Is Not Capitalised
Indian IT services companies, Infosys, TCS, Wipro, HCL Technologies, build enormous amounts of software. Most of it does not appear on their balance sheets as an intangible asset. The reason is not a failure to apply IAS 38 correctly. It reflects the nature of what they build.
Client-specific software: When TCS builds a custom banking system for HDFC Bank, the software is built to HDFC Bank's specification and ownership transfers to HDFC Bank on completion. TCS does not own the software; HDFC Bank does. TCS recognises revenue on the contract under IFRS 15. No intangible asset arises on TCS's books.
Software tools used in service delivery: Internal tools, productivity utilities, and automation platforms developed for use in client engagements are internally generated software used as part of service delivery. These can qualify for capitalisation under IAS 38's development phase criteria, and some Indian IT companies do capitalise a portion of such tools. The amounts are typically modest relative to total assets.
Intellectual property licences: When an Indian IT company licenses proprietary frameworks, platforms, or methodologies, the development costs of those frameworks may qualify for capitalisation during their development phase. Post-capitalisation, they are amortised over their expected commercial life and tested for impairment when revenue from licensing declines.
The contrast with Indian IT product companies, such as Zoho, Freshworks, and NASSCOM-member companies with software product lines, is significant. A software product company developing a cloud HR platform for commercial licensing has a clear development asset: technical feasibility is established, commercial intent is clear, and future economic benefits (subscription revenue) are probable. Development costs qualify for capitalisation.
Wipro: Real Numbers from the Annual Report
Wipro's FY2025 financial statements (filed with the SEC) provide a useful illustration of how IAS 38 intangibles actually look in practice for a major Indian IT company.
Wipro's intangible assets on the balance sheet are almost entirely from acquisitions, not internally generated. The intangibles schedule shows:
Customer-related intangibles: gross carrying value of Rs. 43,672 million as at 31 March 2024, arising from business combinations. These are customer relationships of the companies Wipro acquired, recognised at fair value at the acquisition date under IFRS 3. Wipro itself never capitalised these relationships; the targets generated them internally and expensed the cost. Only through acquisition were they brought onto a balance sheet.
Marketing-related intangibles: Rs. 11,972 million gross carrying value as at 31 March 2024, comprising trade names and similar items acquired through business combinations.
Amortisation and impairment on intangibles: Rs. 11,756 million for the year ended 31 March 2024, including impairment charges of Rs. 1,701 million related to customer-relationship and marketing-related intangibles from prior acquisitions where revenue and earnings estimates declined.
For FY2025, Wipro recognised further impairment of Rs. 1,155 million on customer-relationship and marketing-related intangibles due to declining revenue estimates for certain acquired businesses. The impairment charge flows through the income statement as an operating expense, affecting segment results.
What is absent from Wipro's intangibles schedule: any internally generated software or technology assets of material scale. The balance sheet is transparent about this: Wipro's most valuable assets are its people, its processes, its client relationships, and its brand. None of these meet IAS 38's recognition criteria when internally generated.
Infosys: Impairment of Acquired Intangibles
Infosys's FY2025 SEC filing discloses a similar pattern. Customer-related intangible assets recognised through business combinations, amortised and tested for impairment. During the year ended 31 March 2025, Infosys recognised Rs. 188 crore as impairment of customer-related intangible assets where carrying value exceeded estimated recoverable amount due to declining revenue estimates.
This impairment was tested under IAS 36: the recoverable amount of the customer relationship asset (value in use or fair value less costs to sell) was compared to its carrying amount. When future revenues from those customer relationships were revised downward, the carrying amount exceeded recoverable amount, and the excess was expensed.
The lesson for readers of Indian IT annual reports: the intangible assets line for major Indian IT companies reflects acquisitions, not internally built capabilities. The actual source of competitive advantage, the talent, the methodologies, the client trust, sits off the balance sheet entirely.
AI Development Costs: The Emerging Question
The accounting treatment of AI development costs is one of the most actively debated areas in financial reporting globally in 2025 and 2026. IAS 38's existing framework applies, but its application to AI development is not always straightforward.
The fundamental question is the same six-criterion test. The complication is that AI model development often blurs the research-development boundary more severely than traditional software.
Training a foundation model from scratch (likely research): An entity building a large language model through pre-training on massive datasets cannot reliably demonstrate, at the training phase, that the resulting model will meet technical feasibility for a specific commercial use, that it will generate probable future economic benefits of a determinable nature, or that the cost can be reliably attributed to a specific asset (training costs are highly interdependent). Most commentators conclude that pre-training a foundation model is research-equivalent: expensed as incurred.
Fine-tuning a foundation model for a specific commercial application (potentially development): If an entity takes an existing foundation model (either owned or licensed) and fine-tunes it for a specific, well-defined commercial product (say, a medical diagnosis assistant with a defined regulatory pathway), the six criteria may be demonstrable. Technical feasibility is more assessable. Commercial intent is defined. Resources are quantifiable. These costs may qualify for capitalisation from the point all six criteria are met.
Building an AI-powered feature into an existing software product (development, subject to criteria): An IT services company adding AI-generated code completion to its developer tools product, with established commercial sales and a clear technical roadmap, is likely in development phase for the AI feature costs. The framework does not change; only the nature of the costs being assessed changes.
Indian IT companies investing in AI capabilities, whether building proprietary models or integrating AI into client solutions, face these questions in practice from FY2025 onwards. The IASB is monitoring the issue but has not issued specific guidance on AI cost accounting as of mid-2026. IAS 38's existing framework applies by analogy.
Cloud Computing: Configuration Costs and SaaS
The IFRIC agenda decision from 2021 clarified that configuration and customisation costs for cloud-based SaaS arrangements are generally expensed, not capitalised. The entity does not control the underlying software (the cloud provider hosts and operates it), so the costs do not give rise to an intangible asset under IAS 38.
This has significant implications for Indian companies. A bank implementing a cloud-based core banking system (such as Infosys Finacle hosted on a private cloud or a SaaS solution from a global provider) cannot capitalise the configuration and customisation costs under IFRS/Ind AS. Those costs are expensed, even though they may represent years of work and tens of crores of spend.
The costs that can be capitalised in a cloud arrangement are those that create a distinct, identifiable asset controlled by the entity: code written specifically for the entity's use and residing in a separate application layer the entity controls, data migration costs that create a usable asset, and similar items.
The practical consequence: many Indian IT and fintech companies transitioning to cloud platforms have had to expense implementation costs that they previously might have capitalised under old Indian GAAP. This front-loads the cost in the income statement compared to the old treatment.
Amortisation of Capitalised Software and R&D
Where development costs are capitalised, they are amortised from the date the asset is available for use.
For pharmaceutical products, the amortisation period equals the product's estimated commercial life, which considers the patent protection remaining, the competitive landscape, and the regulatory exclusivity period. A generic drug approved for the US market may have a commercial life of 3-5 years before intense price competition erodes margins. A proprietary biologics product may have a longer commercial life protected by biosimilar development timelines.
For software products, the amortisation period reflects the technology's expected useful life. Software in fast-moving technology stacks (mobile applications, AI-driven features) may become technologically obsolete in 2-3 years. Enterprise software with embedded client workflows may have useful lives of 5-10 years.
Revenue-based amortisation is prohibited under IAS 38, the same prohibition that applies to PPE under IAS 16. The method must reflect economic benefit consumption, not revenue generation.
Ind AS 38 vs IAS 38: R&D and Software
| Area | IAS 38 | Ind AS 38 |
|---|---|---|
| Research: always expensed | Same | Same |
| Development: capitalise when six criteria met | Same | Same; mandatory when criteria are met, not optional |
| Pharma: typical switchover point | Judgment; typically Phase III for NCEs, earlier for generics | Same |
| Software: three-zone framework | By analogy from IAS 38 principles | Same |
| AI development costs | IAS 38 applies by analogy; no specific guidance yet | Same |
| SaaS configuration costs | IFRIC 2021: generally expensed | Ind AS follows IFRIC; generally expensed |
| Old Indian GAAP treatment | Not applicable | Under old IGAAP, development cost capitalisation was optional; under Ind AS 38, capitalisation is mandatory when criteria are met; many Indian companies had to retrospectively reassess on Ind AS adoption |
| Tax treatment of R&D (Section 35 of Income Tax Act) | Not applicable | Section 35 allows 100-150% deduction for R&D expenditure, creating temporary differences that must be tracked under Ind AS 12 |
The Indian income tax dimension is practically relevant. Section 35 of the Income Tax Act allows enhanced deductions for approved R&D expenditure. When development costs are capitalised under Ind AS 38 and deducted for tax purposes in the year of expenditure (under Section 35), a taxable temporary difference arises. Deferred tax liability must be recognised. The interaction between Ind AS 38 capitalisation, Ind AS 12 deferred tax, and the Section 35 enhanced deduction requires careful tracking in Indian pharma and technology companies.
What Big 4 Auditors Focus On
Research-to-development switchover documentation. Auditors test whether the switchover date is supported by contemporaneous documentation: management papers, technical feasibility assessments, business cases, and resource allocation records. Switchover determined retrospectively, at year-end, to align with a desired capitalisation outcome, is a significant audit risk.
Six-criteria substantiation for software. For software projects where costs are capitalised, auditors test each of the six criteria at the commencement of capitalisation: specifically, the technical feasibility evidence and the probability of future economic benefit. For a new software product entering a competitive market, probability requires market research, customer engagement evidence, or a proof of concept with revenue commitment.
Pharma: continuity of capitalisation across reporting periods. If a drug candidate fails Phase III, all previously capitalised development costs must be assessed for impairment immediately. Auditors test whether management has identified trial failures during the year and whether impairment has been recognised promptly.
SaaS configuration cost treatment. Auditors review software implementation projects to confirm that configuration costs for SaaS arrangements have not been capitalised. In large ERP and core banking implementation projects, this line between capitalised and expensed costs is routinely challenged.
AI development cost classification. From FY2025-26 onwards, auditors at large Indian IT and technology-enabled companies are beginning to test whether AI capability development costs have been appropriately classified between research (expense) and development (capitalise if criteria met). This is an emerging area without settled practice, and auditor judgment is developing alongside company practice.
Dip IFRS Exam Angle
R&D and software capitalisation questions in Dip IFRS combine the research-development distinction with multi-period calculation of capitalised amounts and amortisation.
Most tested areas:
Research vs development: given a timeline of pharma or software activities, identify the point at which all six criteria are first demonstrably met, and calculate the amount capitalised from that point. Know that no costs before that point are capitalised, even if the project ultimately succeeds.
Capitalised amount calculation: include only directly attributable development costs: employee salaries for time spent on development, materials consumed in testing, and allocated overheads that are directly attributable. Exclude general overheads, administration costs, and training.
Amortisation: begin when the asset is available for use. Straight-line unless another method better reflects consumption. Revenue-based amortisation is prohibited.
Impairment of capitalised development costs: if a project is abandoned or commercial viability is lost, the capitalised amount is written off immediately in profit or loss.
Common traps:
Capitalising research costs if the project later succeeds. Never. Past expensed research stays expensed.
Using revenue as the amortisation basis for a software product. Prohibited under IAS 38, just as it is under IAS 16.
Capitalising SaaS configuration costs because the effort was substantial. Effort and cost do not drive capitalisation; control and the six-criteria test do.
Assuming all software development costs qualify from the start of coding. The six criteria must be demonstrably met. If technical feasibility is not established at the start of coding (because the entity has not yet committed to a specific technical approach), early-stage coding may still be research-equivalent.
FAQ
Can an entity capitalise R&D costs incurred by a third party on its behalf?
Yes. If the entity controls the resulting asset and the costs meet the six IAS 38 criteria, contract R&D costs can be capitalised. The control element is key: if the third party retains ownership of the resulting intellectual property and licenses it back to the entity, no asset arises on the entity's books.
What happens to capitalised development costs if the project is abandoned?
The carrying amount is written off immediately in profit or loss. The asset is derecognised. No partial recovery or phased write-off. If IAS 36 impairment testing identifies impairment before formal abandonment, the write-down should occur at that point.
How does an entity account for in-process R&D acquired in a business combination?
Acquired in-process R&D is recognised as an intangible asset at fair value at the acquisition date, regardless of whether the acquiree itself had capitalised it. Post-acquisition, subsequent expenditure on that R&D project is assessed under the six IAS 38 development criteria. Expenditure that qualifies is added to the asset; expenditure that does not is expensed.
Is employee time spent on development capitalised?
Yes, if the employees' work is directly attributable to the development of the asset during the development phase. The cost is the salary and directly associated employment costs (benefits, payroll taxes) for the portion of time attributable to development. Time and attendance tracking at the project level is necessary to support this.
Can a pharmaceutical company capitalise Phase II clinical trial costs?
It depends on whether all six IAS 38 development criteria can be demonstrated at the start of Phase II. For most innovative drugs, they cannot (particularly probability of future benefits and technical feasibility). For generic drug bioequivalence studies, it may be possible earlier. Judgment is required and must be documented.
How does the AI accounting treatment differ from traditional software?
The IAS 38 six-criterion framework applies to both. The practical difference is that AI model training blurs the research-development boundary more severely: technical feasibility, the specific commercial application, and probable future economic benefits are harder to establish during foundational model development than during the development of a traditional software application with defined functionality.
Enroll with Global Fin X
R&D capitalisation questions combine accounting judgment with calculation mechanics across multiple periods. Getting the switchover point wrong fails the entire question. Our programme covers IAS 38 across two posts with in-depth lectures on pharmaceutical and software applications, exam-style MCQs, and a dedicated LMS for working professionals.
Enroll Now: Dip IFRS Programme
Faculty profile: www.globalfinx.in/manikanta
This is Post 37 of the Global Fin X IFRS Series. Previous: IAS 38: Recognition Criteria and the Internally Generated Problem. Next: Post 38: IFRS 5 Non-Current Assets Held for Sale and Discontinued Operations.




