Electronic Arts, Inc. v. Commissioner, 118 T.C. 226 (2002), addressed the former Section 936 possessions tax credit and contract manufacturing in Puerto Rico. It did not decide eligibility for the Section 41 research credit or establish a software research qualification test. Its relevance to R&D planning is therefore an analogy about operational participation, not a direct holding on qualified research. Section 41 eligibility must be established independently under its statutory requirements and applicable regulations.

For software developers, business structure, the character of research activities, eligible expenditures, and substantiation are separate issues. Phoenix Design Group, Inc. v. Commissioner, T.C. Memo. 2024-113, and Little Sandy Coal Co. v. Commissioner, 62 F.4th 287 (7th Cir. 2023), illustrate the importance of proving qualifying activities. Deduction rules must also be evaluated by tax year: legislation enacted in July 2025 restored immediate deduction of domestic research or experimental expenditures under Section 174A for taxable years beginning after December 31, 2024. Historical comparisons between company profits and tax expense do not, without consistent accounting and geographic measures, establish the amount of tax saved through research credits.

Technical Foundation of the Section 41 Research Credit

Section 41 provides a credit for qualifying research expenditures, subject to eligibility, calculation, and utilization limitations. The regular credit generally depends on expenditures exceeding a base amount. The alternative simplified credit (ASC) uses a different historical comparison and includes a special rule when a taxpayer has no qualified research expenses in one or more of the preceding three taxable years. A usable credit reduces tax liability dollar for dollar; a deduction reduces taxable income. The overall benefit also depends on Section 280C and applicable tax limitations.

The Quantitative Mechanics of the Credit

The regular research credit generally equals 20 percent of current-year qualified research expenses (QREs) exceeding the base amount. The base is generally the fixed-base percentage multiplied by average annual gross receipts for the four preceding taxable years, subject to a minimum of 50 percent of current-year QREs. Startup fixed-base rules require separate analysis. The ASC generally equals 14 percent of current-year QREs exceeding 50 percent of average QREs for the three preceding taxable years; if any of those years has no QREs, the statutory alternative is 6 percent of current-year QREs. These figures precede any reduced-credit election under Section 280C.

For the regular research credit, the following simplified relationships apply. YBA below denotes the fixed-base percentage multiplied by the relevant average gross receipts; it is an explanatory abbreviation rather than a statutory term.

Regular research credit = 0. × max(0, current-year QREs − base amount)

Base amount = max(YBA, minimum base amount)

Minimum base amount = 0. × current-year QREs

Assume a taxpayer has current-year revenue of $40,000, QREs of $30,000, and a correctly determined historical base calculation of $900. The 50 percent minimum base is $15,000, so the regular research credit is 20 percent of $15,000, or $3,000, before Section 280C adjustments and other limitations. Current-year revenue alone does not establish the historical base or compliance with the startup rules.

The Electronic Arts Precedent: Contract Manufacturing and Its Limits

The Electronic Arts dispute concerned Electronic Arts Puerto Rico, Inc. (EAPR), which used facilities and workers supplied by Power Parts, Inc. to manufacture video game cartridges. EAPR provided manufacturing materials and equipment, owned inventory, and employed a manager at the leased facility. The Tax Court considered whether EAPR actively conducted a trade or business in Puerto Rico and whether it met the separate significant-business-presence requirement for the Section 936 profit-split method.

On the taxpayers’ motion for partial summary judgment, Judge Chabot held that EAPR satisfied the active-conduct requirement. The court distinguished Medchem’s limited operational involvement. However, it denied summary judgment on the disputed manufacturing element of the significant-business-presence requirement because further development was needed. The opinion therefore should not be described as an unconditional approval of every tax consequence of outsourcing.

For Section 41 purposes, the practical inference is limited: outsourced work does not by itself resolve eligibility. A payer’s qualifying contract research expenses generally include 65 percent of eligible payments, subject to statutory exceptions and the requirements of Treasury Regulation Section 1.41-2(e). A research provider must separately analyze the funded-research exclusion. Neither financial exposure nor management oversight substitutes for proof of qualified research.

The Modern Barrier: The Four-Part Test and Documentation Rigor

Section 41’s four-part test applies separately to each business component. Under Treasury Regulation Section 1.41-4(a)(6), at least 80 percent of the relevant research activities, measured by cost or another consistently applied reasonable basis, must constitute elements of a process of experimentation for a qualified purpose. This is not a requirement that 80 percent of every project’s total commercial activity, or every employee’s working time, be experimentation. Electronic Arts did not establish or relax this test.

The Research Expenditure Test: Elimination of Technical Uncertainty

Research must satisfy the expenditure test in Section 41(d)(1)(A), which now refers to Section 174A; earlier taxable years require the law applicable to those years. Technical uncertainty exists when available information does not establish capability, method, or appropriate design for developing or improving a business component. Business uncertainty about market demand or profitability is insufficient.

In Phoenix Design Group, the Tax Court found no qualified research in the three trial projects used by the parties to resolve the disputed claims. The record did not sufficiently establish the requisite technical uncertainty and evaluative experimentation. Complex engineering, compliance calculations, and design revisions do not alone establish qualification. For a software project, useful evidence might address uncertainty about multiplayer concurrency or cross-platform physics performance and explain the alternatives actually evaluated; those examples are illustrative, not findings about Electronic Arts.

The Process of Experimentation (POE)

The process-of-experimentation test requires an evaluative process directed at resolving technical uncertainty. Modeling, simulation, and systematic trial and error can satisfy it. A formal laboratory protocol or an industry-wide scientific breakthrough is not required, but simply following standard design phases does not establish that experimentation occurred. Phoenix Design Group illustrates this distinction.

A technical narrative should connect the initial uncertainty, the alternatives considered, the evaluation performed, and the resulting decisions. Revised drawings or code versions can support that explanation, but their existence alone does not prove the statutory test. Contemporaneous records are particularly helpful when they explain why changes were made rather than merely showing that work occurred.

The “Substantially All” Threshold and the Shrinking-Back Rule

The 80 percent threshold concerns research activities for the relevant business component, using a reasonable and consistently applied measurement basis. Where the overall component fails the qualification requirements, Treasury Regulation Section 1.41-4(b)(2) provides for testing progressively smaller significant subsets. In software development, this may mean examining an identifiable module or interface rather than the entire application.

Shrinking back does not cure a lack of evidence. A taxpayer must still prove qualification and substantiate the expenses associated with the smaller component. Little Sandy Coal also cautions against replacing an analysis of research activities with the proportion of a finished product considered new. The rule is a structured eligibility analysis, not a guarantee that selected technical tasks will qualify.

Funded Research and the Financial Risk Profile

Section 41(d)(4)(H) excludes research to the extent funded by another person or government. For a research provider, Treasury Regulation Section 1.41-4A(d) requires analysis of payment rights and substantial rights in the results. Whether payment depends on successful research is important; ordinary commercial risk is not necessarily sufficient. The payer’s potential contract-research credit is a separate inquiry.

The Impact of State Law on Risk Evaluation

In System Technologies, Inc. v. Commissioner, Docket No. 12211-21, the Tax Court’s January 3, 2025 order denied the IRS’s motion for partial summary judgment on funded research. The analysis considered Indiana commercial law and whether contractual limits on repair or replacement actually eliminated the customer’s remedies for failed delivery. This was a procedural ruling rejecting the IRS’s proposed summary disposition, not a final award of all claimed credits.

The December 18, 2024 order in Smith v. Commissioner likewise addressed the IRS’s funded-research arguments at the summary-judgment stage. It should not be characterized as a 2025 final decision establishing that fixed prices or milestone payments always prove eligibility. Contract language, enforceable state-law rights, payment conditions, and the particular research must be examined together. These earlier orders do not establish the ultimate disposition of every credit claimed in those proceedings.

Retention of Substantial Rights

Substantial rights need not be exclusive, but incidental knowledge gained from performing services is not enough. Contracts transferring deliverables do not automatically transfer every substantial right, and silence does not automatically establish retained rights. Agreements, licenses, applicable law, and any required payment for further use must be reviewed. Co-development or exclusive-platform arrangements require their own factual analysis; no particular Electronic Arts contract is assessed here.

Internal Use Software (IUS) and the High Threshold of Innovation

Software developed primarily for general and administrative functions, including financial management, human resources, and support services, may be internal use software under Treasury Regulation Section 1.41-4(c)(6). In addition to the ordinary qualification requirements and exclusions, such software generally must satisfy a high threshold of innovation. Regulatory exceptions and dual-function rules can affect this analysis.

Where the high-threshold test applies, credit eligibility requires all three of the following:

Innovation: The software must produce a substantial and economically significant reduction in cost, improvement in speed, or other measurable improvement.

Significant economic risk: Development must involve a significant commitment of resources and substantial uncertainty, because of technical risk, about recovering those resources within a reasonable period.

Commercial availability: The software cannot be purchased, leased, or licensed and used for the intended purpose without modifications satisfying the innovation and significant-economic-risk requirements.

A game-development engine is not internal use software merely because the developer uses it internally. Classification depends on its intended functions and the regulatory definitions; exceptions may apply to software used in qualified research or qualifying production activities. It is therefore inappropriate to classify EA’s Frostbite engine automatically as administrative internal use software or to assert that EA replaced every game engine with one platform. Any engine-development claim requires analysis of the particular functions, uncertainties, activities, and costs.

Documentation Strategies: The Shift to Contemporaneous Records

Phoenix Design Group and Little Sandy Coal demonstrate the risks of weak evidence about qualifying activities and their extent. They do not establish a universal legal prohibition on retrospective interviews, reasonable estimates, or consultant-prepared studies. The record must substantiate qualification and the amounts claimed; broad departmental percentages without a reliable factual foundation are vulnerable.

The Limits of Retrospective Interviews

The source draft refers to the government’s summary-judgment arguments in the Kyocera AVX refund litigation. A litigant’s contention is not a judicial holding, and it does not establish a categorical ban on retrospective evidence. Treasury Regulation Section 1.41-4(d) requires records in sufficiently usable form and detail to substantiate eligibility and amount. Contemporaneous time records can be persuasive, but no universal rule requires a specific time-tracking system. Employee wage qualification, including direct supervision and direct support, must also be distinguished from the business-component experimentation threshold.

Using Engineering Systems for Tax Substantiation

Engineering systems such as Jira, GitHub, Asana, and Bitbucket can support a research credit study when their records are connected to eligible activities and reconciled to financial data. Useful records may capture:

Technical uncertainty: Issues identifying unresolved capability, method, or design questions. An ordinary bug ticket is not automatically evidence of qualified research.

Evaluation of alternatives: Test plans, benchmark results, simulations, code reviews, and commit histories explaining what was evaluated and why. Iteration alone is insufficient.

Activity and cost mapping: Records connecting personnel, qualifying work