Qualifying AI for the R&D Tax Credit | KBKG
Last Updated on 9/21/2026

By Jonathan Tucker | Principal, R&D Tax Credit

Qualifying AI for the R&D Tax Credit | KBKG

Artificial intelligence (AI) has recently been a mainstay of news headlines worldwide, and for good reason. Publicly released AI tools now allow individuals and companies to generate images and text nearly indistinguishable from those created by humans. 

Companies developing or materially improving artificial intelligence technology may qualify for federal and state R&D tax credits under IRC Section 41. Potentially qualifying work includes developing machine-learning models, natural-language processing systems, retrieval architectures, recommendation engines, image-processing tools, and other AI-enabled business components. 

Simply purchasing, configuring, or using an existing AI platform does not create an R&D tax credit on its own. The underlying development work must satisfy the applicable Section 41 requirements, including technological uncertainty and a process of experimentation. 

What Are R&D Tax Credits?

The R&D Tax Credit, also known as the Research and Development Tax Credit or Research and Experimentation Tax Credit, is a federal tax incentive under Internal Revenue Code (IRC) Section 41 for companies that design, develop, or improve products, processes, software, formulas, techniques, and other business components. (IRC Section 41). 

For AI companies and software developers, qualified research activities may include technical work performed by software engineers, machine learning engineers, data scientists, business analysts, and other professionals involved in AI development, testing, and experimentation  (Treas. Reg. §1.41-4). 

Benefits of the R&D Tax Credit

The R&D tax credit may provide several benefits for companies investing in AI development and other qualifying research activities.  

  • Generate federal R&D tax credits and evaluate separate state research-credit opportunities, where available. 
  • Reduce federal and state income tax liability on a dollar-for-dollar basis. 
  • Lower the company’s effective tax rate. 
  • Improve cash flow and earnings-per-share. 
  • Carry unused federal credits back one year and forward for up to 20 years, subject to the general business credit rules. 
  • Use look-back studies to identify unclaimed credits from open prior tax years. 

Qualifying AI Activities

Many AI development activities may qualify for R&D tax credits when they involve technical uncertainty and a process of experimentation. 

To qualify under IRC Section 41, research activities generally must relate to a business component, involve technological uncertainty, rely on a process of experimentation, and be technological in nature (IRC Section 41; Treas. Reg. §1.41-4). 

The activity label does not determine eligibility. Fine-tuning a model, building a retrieval-augmented generation system, or developing an AI automation tool qualifies only when the company is resolving technical uncertainty through a documented process of evaluating alternatives. Routine prompt configuration, ordinary use of a third-party AI platform, content generation, data entry, and basic implementation generally do not qualify on their own. 

  • Developing software that recognizes behavior patterns and generates recommendations. 
  • Improving robotics or AI systems used in manufacturing or assembly-line environments. 
  • Developing image processing, speech recognition, or natural language processing software. 
  • Fine-tuning large language models or developing retrieval-augmented generation systems when the work involves resolving uncertainty around accuracy, latency, or model performance. 
  • Creating and testing AI-powered tax software or automation tools. 
  • Developing algorithms, architectures, or application software involving AI functionality. 
  • Training or improving AI tools used to support patients, caregivers, or healthcare providers. 

According to KBKG’s R&D tax credit experts, the strongest AI claims identify the specific technical problem being solved, the alternatives evaluated, the tests performed, and the measurable results used to choose or reject an approach. 

Does Internal-Use AI Software Face Additional Rules?

AI software developed primarily for a company’s internal financial, human-resources, or support functions may need to satisfy the high-threshold-of-innovation test in addition to the standard Section 41 requirements. The company must generally demonstrate innovation, significant economic risk, and the absence of a commercially available alternative that could meet its needs without qualifying modifications. 

AI software developed for sale, lease, license, or third-party interaction may receive different treatment. Dual-function systems require a separate review of their internal and third-party uses.

Qualifying Research Expenses for AI Development

Qualified research expenses generally include wages, certain nondepreciable supplies, qualifying computer rental or lease costs, and eligible contract research expenses connected to activities that satisfy the Section 41 requirements. 

Potential QREs for AI development include: 

  • Wages: The portion of compensation paid to employees who directly perform, directly supervise, or directly support qualified AI research. Job titles such as data scientist, engineer, or business analyst do not qualify automatically.  
  • Computer costs: Certain rental or lease costs for off-premises computers used in qualified research. Not every cloud-hosting, storage, software-subscription, or production-computing cost qualifies.  
  • Contract research: Generally 65% of eligible payments for qualified research performed by third parties, provided the taxpayer retains the required rights and bears the applicable economic risk.  
  • Supplies: Nondepreciable materials consumed during qualifying experimentation. This category is often limited in software-focused AI projects.  

Companies should allocate cloud and computing expenses between qualified development or testing and nonqualifying production, hosting, maintenance, or general business use. 

KBKG’s R&D team recommends connecting each claimed expense to a defined AI business component, employee role, development activity, and supporting technical record. This creates a clearer trail from project documentation to the amounts reported on Form 6765. 

How Sections 174 and 174A Affect AI Development Costs

Sections 174 and 174A govern when research or experimental expenditures are deducted or amortized. Section 41 separately determines whether activities and expenses qualify for the R&D tax credit. 

For tax years beginning after December 31, 2024, Section 174A generally allows companies to deduct domestic research or experimental expenditures, including domestic software-development costs, in the year paid or incurred. A business can instead elect to capitalize and amortize those domestic costs over at least 60 months. 

Foreign research expenditures remain subject to 15-year amortization under Section 174. Transition rules also apply to certain domestic costs capitalized during tax years beginning after 2021 and before 2025. 

AI companies should coordinate Section 41 credit eligibility, Sections 174 and 174A expense treatment, and the Section 280C election rather than evaluating each provision separately. 

Is Your Company Eligible for a Tax Credit?

AI development is taking place across many industries. Companies that develop or improve AI-powered systems may be eligible for the R&D tax credit if their activities meet the applicable requirements. 

Companies in these industries do not qualify merely because they use AI. The potential credit arises when they develop or improve the underlying technology through activities that satisfy the Section 41 requirements. 

Health Care

The AI-associated healthcare market is growing at a rapid rate. Healthcare organizations and software companies may use AI to analyze medical data, improve patient outcomes, support clinical decisions, enhance imaging tools, and improve patient experiences. Development of tools such as AI-powered diagnostic imaging systems may involve qualifying technical research when teams are resolving uncertainty around detection accuracy, processing speed, or model performance. 

Pharmaceutical

Pharmaceutical companies may use AI to analyze genetic, proteomic, clinical, and drug development data. AI tools used to improve target identification, compound screening, research, discovery, and development processes may support qualifying R&D activities. 

Aerospace

The aerospace industry has been developing AI since the invention of autopilot over a century ago. Aerospace companies may use AI to improve control systems, automation, navigation, operations, and other technical functions. Development and testing of systems such as autonomous navigation tools or predictive maintenance models may involve qualifying experimentation. 

Game Development

The gaming industry is worth more than $300 billion and serves as one of the largest markets directing the development of AI technology. Developers in the gaming industry create innovative technologies that improve mechanics and create a more adaptive and immersive gaming experience. Additionally, game developers who research and develop their AI-powered applications have breached other industries, selling their technology to Google Translate and Google Photos.

Accounting and Finance

The development of AI software that can identify patterns within data to predict outcomes and identify risks allows financial institutions to detect fraud quickly and efficiently. Eligible companies include those that develop generative AI for businesses or have a department dedicated to creating innovative AI for improving the finance industry. 

R&D tax credit software development can help you determine your company’s eligibility. 

Next Steps: Evaluate Your AI Development Activities

Companies should begin by identifying the AI products, models, algorithms, software modules, or internal systems developed or materially improved during the year. For each business component, document the technical uncertainty, alternatives considered, experiments performed, employees involved, and related development costs. 

KBKG’s R&D Tax Credit team helps AI companies and their CPAs apply the four-part test, evaluate internal-use and dual-function software rules, identify qualified wages and computing costs, calculate federal and state credits, and coordinate the claim with Sections 174, 174A, and 280C. The team also helps organize the technical and financial documentation supporting Form 6765. 

Companies evaluating current or open prior tax years can contact KBKG for a complimentary qualification assessment. 

Frequently Asked Questions

Does Using ChatGPT or Another Third-Party AI Platform Qualify for the R&D Tax Credit?

No. Using an existing AI platform for content, analysis, customer service, or workflow automation does not qualify on its own. Technical development performed to create or materially improve an AI-enabled business component qualifies only when it satisfies the Section 41 requirements. 

Yes, when the project addresses technical uncertainty and evaluates alternatives related to matters such as model accuracy, retrieval quality, architecture, latency, scalability, hallucination reduction, or system performance. Routine configuration or prompt changes without a qualifying experimental process generally do not qualify. 

Certain off-premises computer rental or lease costs used directly in qualified research can qualify. General hosting, production usage, data storage, software subscriptions, and routine cloud operations do not automatically qualify and should be separately allocated and documented.