The Statutory Architecture of the Research Credit

The research credit was introduced in 1981 and made permanent in 2015. Its qualification requirements operate at the business-component level. A business component is a product, process, computer software, technique, formula, or invention held for sale, lease, or license, or used in the taxpayeru2019s trade or business. All applicable elements of the four-part test must be satisfied, together with the expense rules and exclusions in u00a7 41. The cases examined here concern historical tax years, for which the applicable version of u00a7 174 informed the research-expense test.

For those historical years, the first part of the test required expenditures eligible for treatment under u00a7 174. Research must address uncertainty about the capability or method of developing or improving a product, or its appropriate design. The inquiry concerns information available to the taxpayer when the relevant activities begin. For tax years beginning after December 31, 2024, Congress amended u00a7 41(d)(1)(A) to refer to domestic research or experimental expenditures under new u00a7 174A. Historical case terminology should therefore not be mistaken for a complete statement of the current expense-treatment rules.

The research must be technological in nature, relying on principles of the physical or biological sciences, engineering, or computer science. The information must be intended to develop or improve a business component for a permitted purpose: function, performance, reliability, or quality. Substantially all of the relevant research activities must constitute elements of a process of experimentation. That process evaluates alternatives to resolve uncertainty and may involve modeling, simulation, or systematic trial and error. Technical complexity, novelty, or a successful result alone does not demonstrate that these requirements were met.

Case Identification and the Smith Proceedings

Smith concerned research credits flowing through Adrian Smith + Gordon Gill Architecture, LLP (AS+GG), to partners Adrian D. Smith, Carlisle G. Gill, and Robert J. Forest and their spouses. The consolidated docket numbers were 13382-17, 13385-17, and 13387-17. These are not Funkhouser proceedings. Phoenix Design Group was a separate corporate taxpayer; the verified decisions do not establish the ownership or litigation connection alleged in the source document.

AS+GGu2019s disputed credits related to architectural projects for tax years 2008 through 2010, with six projects selected for analysis. The projects included developments in the United Arab Emirates and Saudi Arabia. A foreign project location does not by itself determine where the research was performed; u00a7 41 separately excludes research conducted outside the United States, Puerto Rico, and U.S. possessions. The funded research dispute concerned payment risk and retained rights under the relevant contracts.

The Tax Court denied the IRSu2019s summary-judgment motion in an order dated December 18, 2024. That procedural ruling did not establish final entitlement to the credits. In its subsequent opinion, Smith v. Commissioner, T.C. Memo. 2026-50, filed June 16, 2026, the court found that none of the six contracts made payment contingent on research success. AS+GG nevertheless retained substantial rights under four contracts, permitting potential partial credits for research expenses exceeding payments, subject to Treasury Regulation u00a7 1.41-4A(d)(3). The court could not determine the precise credit amounts, if any. It also upheld the reasonableness of the partnersu2019 total 2008 compensation. The later opinion therefore supersedes any description of Smith as merely awaiting trial or as an unconditional taxpayer victory.

Forensic Analysis of Phoenix Design Group, Inc. v. Commissioner

While the Smith case centered on the “funded research” question, the decision in Phoenix Design Group, Inc. v. Commissioner (T.C. Memo 2024-113) addressed the substantive merits of the engineering activities themselves. Phoenix Design Group (PDG) was a multidisciplinary engineering consulting firm specialized in mechanical, electrical, plumbing, and fire protection (MEPF) systems, primarily for complex facilities like hospitals and laboratories. PDG claimed R&D credits for the tax years 2013 through 2016, identifying 238 projects as qualified research.

The parties agreed to a nonbinding three-project sample for trial. The court held that none of those three projects involved qualified research, either as a whole or under the shrinking-back analysis. It also held PDG liable for accuracy-related penalties under the partiesu2019 express stipulation. The opinion should not be described as a merits examination of every one of the 238 projects or as a free-standing determination that every engineering credit claim warrants penalties.

The Breakdown of the Trial Projects

The three trial projects involved hospital and university systems. The court examined the work and the evidence for each project rather than treating technical sophistication as proof of qualification.

Table 1: Factual Deficiencies in PDG Trial Projects.

For Gerald Champion, the record described mechanical, electrical, and plumbing design decisions, including selection of a variable air volume system. PDGu2019s descriptions of design changes and ultimate solutions did not adequately establish the investigative or experimental activities that produced those solutions. Drawings can support a research claim, but their evidentiary value depends on explaining what they demonstrate about uncertainty, alternatives, and the work performed.

The Rejection of AIA Design Phases as Process of Experimentation

PDG argued that its six-stage design processu2014Pre-Design, Schematic Design, Design Development, Construction Documents, Bidding, and Construction Administrationu2014demonstrated experimentation. The argument relied on engineers revisiting earlier stages when later design issues arose. The court considered whether the evidence established actual experimentation within that process.

The court rejected reliance on the six-stage description alone. Time-entry narratives did not consistently support PDGu2019s account of the process or connect employeesu2019 work to the uncertainties identified at trial. This does not mean that work performed within AIA design phases is necessarily ineligible or that a separate laboratory protocol is always required. It means that a project-management framework must be supported by evidence showing how the actual activities satisfy the statutory test.

The Evidentiary Wall: Substantiation and Retrospective Narratives

PDG engaged alliantgroup in 2017 to conduct a research credit study relating to earlier years. The evidentiary problem was not simply that the study was retrospective: PDG had time sheets and design materials, but the court could not reliably connect those materials to qualifying activities. Treasury Regulation u00a7 1.41-4(d) requires records in sufficiently usable form and detail to substantiate eligibility. Contemporaneous documentation is valuable, while credible testimony and retrospective analysis may supplement the record; neither a consultantu2019s study nor an unsupported percentage allocation establishes eligibility by itself.

The source also refers to the Kyocera AVX refund litigation. In July 2024 the government challenged the substantiation of an additional research-credit refund claim of approximately $1. million. That challenge was a litigation position, not itself a judicial holding that retrospective interviews are categorically insufficient. The supplied material does not establish a verified final merits ruling on that research-credit claim, and this study does not treat Kyocera as settled precedent. Its discussion illustrates disputed substantiation issues rather than a universal prohibition on interview-based evidence.

The Accuracy-Related Penalty Regime

Phoenix Design Group involved accuracy-related penalties under IRC u00a7 6662. The parties stipulated that PDG would be liable for penalties for deficiency years if none of the trial projects qualified, and the court applied that stipulation. The opinion therefore does not establish that using a retrospective study automatically defeats reasonable cause or triggers a penalty. More generally, the statutory penalty is applied to the relevant portion of an underpayment, subject to applicable requirements and defenses; it is not automatically 20% of a credit claimed.

Table 2: Deficiencies and Accuracy-Related Penalties Determined in the PDG Notice of Deficiency. The amounts below are the notice figures reproduced in the opinion, not a schedule of credits utilized or independently recalculated final liabilities. The 2016 penalty is reproduced as $9,021; it should not be changed merely to equal 20% of the displayed deficiency. The courtu2019s penalty holding operated under the partiesu2019 stipulation.

Disallowance and penalties can create material financial exposure. Taxpayers should assess whether the evidence supports each claimed expense and whether any applicable penalty defense is established on its own facts. The PDG stipulation should not be generalized into an automatic penalty rule for other taxpayers.

Technical Uncertainty and the Section 174 Threshold

The PDG opinion distinguished actual technical uncertainty from the continuing possibility of design changes. PDG contended that uncertainty persisted until construction ended because designs might be revised. The court rejected that proposition as a sufficient explanation of uncertainty in an entire system. The relevant inquiry is whether available information established the capability, method, or appropriate design of the particular component being developed or improved.

The court also distinguished investigation from routine calculations using information already available. Merely identifying an engineering problem and its eventual solution does not establish the intervening research activities. Applying established principles does not automatically disqualify research, but taxpayers must explain the uncertainty and the evaluative work undertaken to resolve it. The analysis must remain specific to the facts and the component under examination.

The Mathematical Rigor of the “Substantially All” Test

Treasury Regulation u00a7 1.41-4(a)(6) generally requires at least 80% of the relevant research activities, measured by cost or another consistently applied reasonable basis, to constitute elements of a process of experimentation for a qualified purpose. This is an activities test, not an automatic percentage of every hour charged to a commercial project. The regulation also addresses the remaining activities when the 80% threshold is met. Expense eligibility must still be evaluated separately. If the component fails the qualification requirements, the shrinking-back rule considers its most significant subset and then successively smaller subsets.

For a simplified illustration, assume that 100 hours properly belong in the relevant research-activity denominator and that only 70 hours constitute elements of a qualifying process of experimentation. The ratio is 70%, below the threshold. This example depends on those classifications: drafting or compliance work must not automatically be included in the denominator or excluded from experimentation solely because of its label. Shrinking back requires a genuine component or subcomponent that satisfies the tests; it does not permit selecting only favorable hours and calling them a component.

The illustrative activity ratio is:

POE ratio = activities constituting elements of qualifying experimentation u00f7 relevant research activities u2265 0.80. Use the same measurement basis in the numerator and denominator.

If the component does not qualify, evaluate the most significant subset under the shrinking-back rule and repeat as necessary. For an eligible subset:

Subset POE ratio = activities constituting elements of qualifying experimentation within the subset u00f7 relevant research activities within the subset u2265 0.80. All other qualification requirements must also be met.

PDG illustrates the practical importance of this evidence. The court examined mechanical, electrical, and plumbing systems and considered smaller subsets, but generic time descriptions and unexplained drawings did not establish the activities or permit reliable allocation. Shrinking back could not cure missing proof of investigation and experimentation. Individual tasks are evidence about a component; they do not become business components merely through relabeling.

Contractual Risk and Funded Research: Lessons from System Technologies

System Technologies, Inc. v. Commissioner, Docket No. 12211-21, involved industrial finishing systems and a January 3, 2025 order denying the IRSu2019s motion for partial summary judgment on funded research. The IRS argued that purchase orders did not expressly condition payment on successful research and that warranty terms displaced buyersu2019 ordinary remedies.

The court considered Indiana law and concluded that the warranty terms did not eliminate remedies for total breach, including recovery of payments where the promised system was not delivered. That supported the taxpayeru2019s position on research risk. The order denied the IRSu2019s motion; it was not a final determination allowing every research expense or the entire credit.

Contracts must be assessed together with the law that governs their interpretation. This does not establish a rule that inserting any refund clause or choosing Indiana law guarantees eligibility. Nor is a separately stated financial penalty universally required. Payment contingency and substantial rights require a fact-specific analysis of the agreements and their legal effect.

Comparison of Regulatory and Judicial Outcomes

These separate proceedings address different aspects of research-credit eligibility. Activity qualification, substantiation, and funding should be evaluated independently, with careful attention to whether a source describes a partyu2019s argument, an interlocutory order, or an opinion after trial.

Table 3: Comparative Analysis of Recent R&D Tax Credit Jurisprudence.

A taxpayeru2019s success in defeating summary judgment does not establish entitlement to a credit. Smithu2019s later opinion particularly illustrates why retained rights, payment terms, and the amount of funding must be assessed together. Conversely, the PDG trial findings should not be presented as a blanket exclusion of engineering or architectural services.

Implications for Future R&D Tax Credit Applications

The practical implications concern evidence quality, component identification, and contracts. These are compliance recommendations drawn from the issues discussed above, rather than newly imposed universal documentation rules.

The Shift Toward Activity-Level Documentation

Taxpayers should retain project records that explain the relevant activities in enough detail to connect expenses with qualifying research. Time records, design revisions, technical emails, test results, and descriptions of alternatives can work together. Contemporaneous activity descriptions can make that connection easier to establish. No particular software, automated system, or minute-by-minute time sheet is universally mandated by u00a7 41. Project-level records may be adequate where their content supplies the necessary detail.

Decoupling from Industry Standa