Historical Context: Garrison v. Commissioner and Capitalization

Garrison v. Commissioner, 86 T.C. 764 (1986), concerned an author’s expenses associated with the manuscript Still a Distant Drum and the application of former Section 280. Chief Judge Sterrett concluded that the statute required the costs to be recovered over the income stream from the production activity.

The Second Circuit reversed that decision in Hadley v. Commissioner, 819 F.2d 359 (2d Cir. 1987). It concluded that former Section 280 did not apply to the authors’ manuscript expenses and allowed current deductions for the years before it. The court distinguished an author’s creation of a manuscript from a publisher’s production of books and considered the legislation’s tax-shelter context.

That reversal makes it misleading to describe the original Tax Court opinion as the foundation of present research capitalization rules. The case offers historical context for disputes about deduction timing, but the TCJA and OBBBA changes arise from legislation governing R&E expenditures. Neither a general duty to capitalize every new intangible asset nor a heightened Section 41 documentation standard follows from Garrison.

The Modern Regulatory Framework: Sections 174, 174A, and 41

Sections 174 and 174A address when R&E expenditures are recovered for federal income tax purposes. Section 41 provides a separate credit and imposes additional conditions on the activities and expense categories that qualify. Deductibility does not itself establish credit eligibility, and a business can incur research expenditures that do not generate a research credit.

Section 41’s four-part framework addresses qualifying research expenditures, technological information, a new or improved business component, and a process of experimentation for a permitted purpose. The permitted improvements concern function, performance, reliability, or quality. Each business component must be evaluated separately, and statutory exclusions remain relevant. For years beginning after 2024, the expenditure requirement reflects the statutory reference to Section 174A.

The Three-Phase Evolution of R&E Expenditure Treatment

The OBBBA changed cost recovery; it did not establish a new universal documentation format for Section 41. Phoenix Design Group and other cases illustrate the application of existing credit requirements to particular evidentiary records.

Judicial Scrutiny in 2024: Phoenix Design Group and Engineering Design

In Phoenix Design Group, Inc. v. Commissioner, T.C. Memo. 2024-113, issued December 23, 2024, the Tax Court considered research credits claimed by a mechanical, electrical, plumbing, and fire-protection engineering firm. Three sampled projects formed the basis for determining the disputed credits for 2015–2019. The court found the claimed research did not satisfy the relevant qualification requirements.

The Failure to Establish a Process of Experimentation

The firm’s six-stage design process and project complexity did not, by themselves, establish qualified research. The decision addressed both the research uncertainty requirement and the process of experimentation. Routine engineering calculations, design revisions, and compliance work must be distinguished from activities that actually evaluate alternatives to resolve technological uncertainty.

Complexity alone is insufficient, but use of established engineering principles is not disqualifying: qualified experimentation ordinarily relies on those principles. The relevant inquiry concerns uncertainty about capability, method, or appropriate design and the process used to resolve it. Neither industry-wide novelty nor discovery beyond the existing state of scientific knowledge is a universal requirement.

Documentation Standards and Contemporaneous Evidence

Phoenix Design Group illustrates the weakness of general design descriptions when the underlying evidence does not establish qualifying uncertainty and experimentation. It does not prohibit all engineering credits or create a universal requirement for one type of timekeeping software.

Taxpayers should retain records sufficient to substantiate the activities and expenses claimed. Contemporaneous technical records are useful because they show what was uncertain, which alternatives were considered, and what work occurred. Credible testimony and supported estimates may contribute to the evidence; unsupported percentages or retrospective generalizations can fail to establish the necessary facts.

The table identifies useful evidence categories. Its heading does not mean that every taxpayer must create each listed document. The governing requirement is sufficient substantiation, assessed in light of the activities and records involved.

The Funded Research Exclusion: Meyer, Borgman & Johnson and Risk Allocation

Section 41(d)(4)(H) excludes funded research. Under the applicable regulations, a contractor’s payment rights and retention of substantial rights in the research results are central to determining whether its research is funded. All relevant agreements must be considered; describing an engagement as “fixed-price” does not resolve the issue.

In Meyer, Borgman & Johnson, Inc. v. Commissioner, No. 23-1523, decided May 6, 2024, the Eighth Circuit affirmed the denial of the engineering firm’s research credits on funded-research grounds. The payment provisions did not make compensation contingent on research success. Ordinary professional obligations and the potential for cost overruns did not establish the required contingency.

The decision does not prescribe mandatory wording that automatically makes a contract unfunded. Contract interpretation examines the parties’ substantive rights and obligations. A standard-of-care provision, acceptance requirement, or fixed fee must be assessed in context. Retention of substantial research rights also requires separate consideration; financial risk alone is insufficient.

Technological Uncertainty in Software Development

Software development requires a distinction between expense recovery and research credit eligibility. Section 174A expressly treats amounts paid or incurred in connection with software development as R&E expenditures for its purposes. The resulting deduction treatment does not make all software development activities qualified research under Section 41.

Software credit eligibility depends on the applicable Section 41 requirements and exclusions. Technical documentation should identify the business component, the uncertainty faced during development, and the alternatives evaluated. The label “software development” does not replace that analysis.

For example, work on scalability or latency may qualify when the facts establish technological uncertainty and a qualifying evaluation of alternatives. Routine configuration does not qualify merely because it concerns a complex cloud system. Conversely, using existing libraries or APIs does not automatically defeat a claim. Internal-use software may also be subject to additional regulatory requirements, depending on its function and the applicable exceptions.

The Substantially All Rule and the Shrink-Back Rule

In Little Sandy Coal Co., Inc. v. Commissioner, 62 F.4th 287 (7th Cir. 2023), the Seventh Circuit affirmed the denial of research credits where the taxpayer failed to substantiate the required experimentation threshold. The substantially-all test concerns research activities, rather than the percentage of a finished product that is new.

Under Treasury Regulation Section 1.41-4(a)(6), at least 80% of the relevant research activities must constitute elements of a process of experimentation, measured by cost or another consistently applied reasonable basis. The remaining activities must satisfy the other applicable requirements. This threshold is distinct from the employee wage rule and the business-component disclosure thresholds on Form 6765.

When the overall business component fails the requirements, the shrink-back rule applies the analysis to the most significant subset of its elements, continuing as appropriate. An engineering example might involve a defined system or subsystem, provided that the taxpayer’s business component and the evidence support that boundary. An entire client building should not automatically be treated as the engineering firm’s business component.

The Shrink-Back Logic Hierarchy

• Full Business Component: Identify the taxpayer’s product, process, software, technique, formula, or invention and assess the qualification requirements.
• Major System: If necessary, consider the most significant subset of the component’s elements.
• Subcomponent: Continue to an appropriate smaller subset where the evidence permits application of the tests.
• Activity Evidence: Connect employee work and expenses to the identified subset; isolated employee activities do not replace identification of a qualifying component or subset.

Shrink-back cannot cure an evidentiary gap merely by giving a project a smaller label. Taxpayers must establish the activities and costs associated with the subset they claim qualifies.

The Legislative Response: The One Big Beautiful Bill Act of 2025

The OBBBA restored current domestic R&E deductions for amounts paid or incurred in tax years beginning after December 31, 2024. This is a change in deduction timing, distinct from the research credit itself. Section 280C coordination rules must also be considered when a taxpayer claims both research deductions and a credit.

Key Provisions of the OBBBA of 2025

The small-business retroactive election uses the Section 448(c) gross-receipts test for the first tax year beginning after 2024, with a $31 million threshold for 2025. It is not simply a test of revenue being below $31 million in one year. Aggregation rules, the applicable three-year averaging period, and the exclusion for tax shelters matter.

Revenue Procedure 2025-28 generally required the retroactive election by July 6, 2026, and ordinary refund limitation periods could produce an earlier deadline. As of September 13, 2026, that general election deadline has passed. Taxpayers should not assume they can now make the election merely by amending a return.

Separate transition relief allows taxpayers to elect recovery of remaining unamortized domestic R&E expenditures from tax years beginning in 2022–2024 in the first tax year beginning after 2024, or ratably over two years beginning with that year. This election is distinct from the small-business retroactive election and requires attention to its own procedures and deadlines.

Strategic Implications and Future Outlook

The practical lesson is to assess expense recovery, research qualification, contractual funding, and evidence together without treating them as interchangeable. A successful new product is not sufficient proof of qualified research. A failed development effort may nevertheless qualify when the statutory requirements are met.

For Engineering and Architecture Firms

Firms should identify technical uncertainty within their actual business components, distinguish routine services from qualifying investigation, and retain evidence showing how alternatives were evaluated. Design files, calculations, meeting records, test results, and reliable labor allocations can support that analysis. Documentation should reflect the work performed rather than impose a research narrative on every design revision.

Contract reviews should address payment contingencies and substantial rights before a claim is made. Any prospective change should reflect the commercial arrangement and legal obligations. Altering contract terminology alone does not establish qualification or retroactively change the economic substance of completed work.

For Software Developers

The December 2025 Instructions for Form 6765 make Section G optional for tax years beginning before 2026 and generally required for years beginning after 2025, subject to exceptions. Exceptions include qualifying small businesses making the payroll tax credit election and certain original-return filers with no more than $1.5 million in controlled-group QREs and no more than $50 million in average annual gross receipts.

Where Section G applies, it separates wages for conducting, directly supervising, and directly supporting research. Its current instructions apply the description of information sought in column 49(f) to specified amended claims, not timely original returns. Software firms should follow the instructions for the relevant year and return type, while retaining underlying qualification evidence regardless of a disclosure exception.

The Convergence of State and Federal Incentives

State incentives require their own eligibility analysis. Georgia’s Job Tax Credit can include qualifying businesses in R&D industries, and designated military zones have location-specific rules. Oklahoma’s clean-burning fuel property incentive concerns qualifying fuel-related investments rather than constituting a general R&D credit.

Neither program establishes federal Section 41 eligibility. State conformity, geographic boundaries, job or investment conditions, coordination restrictions, and application deadlines must be