R&D credits and the computer rental rules time forgot

Editor: Mary Van Leuven, J.D., LL.M.

Every year companies across the United States invest billions on software, computing power, and other technology tools to stay at the forefront of innovation. Certain investments may qualify for the Sec. 41 credit for increasing research activities (the R&D credit), but as technology has advanced, the regulatory framework that applies in determining credit eligibility has not.

The R&D credit was first introduced as part of the Economic Recovery Tax Act of 1981, P.L. 97–34. Congress feared that the U.S. economy was falling behind those of countries such as Japan and (at the time) West Germany and hoped a meaningful research credit would encourage U.S. businesses to invest more heavily in domestic R&D. Though the R&D credit has seen a number of changes over the last 4½ decades, the core framework remains largely intact: A taxpayer may claim a credit for a percentage of certain incremental qualified research expenditures (QREs) paid or incurred, including employees’ wages, supplies, and certain contract services.

A fourth category of potentially qualified expenses has for many years seen a relatively limited application. Originally codified in Sec. 44F(b)(2)(A)(iii), QREs potentially included amounts paid to unrelated parties for the right to use personal property in the conduct of research. The Tax Reform Act of 1986, P.L. 99–514, narrowed eligibility for these costs, however. Instead of amounts paid to use any personal property for qualified research, only amounts paid or incurred for the right to use computers in the conduct of qualified research remain eligible for the credit (Sec. 41(b)(2)(A)(iii)). At the time of the Tax Reform Act of 1986, computers were large, expensive pieces of equipment, and Congress believed that continued eligibility for these costs would benefit small businesses that could not afford to purchase or lease computers for research purposes (Joint Committee on Taxation, General Explanation of the Tax Reform Act of 1986 (JCS–10–87) (May 15, 1987), p. 130).

In Sec. 41(b)(2)(A)(iii), Congress granted explicit rulemaking authority to Treasury regarding whether amounts paid or incurred to another person for the right to use computers in the conduct of qualified research constitute QREs. In 1989, as part of T.D. 8251, Treasury issued regulations regarding, among other things, computer rental research expenses. Regs. Sec. 1.41–2(b)(4) established three requirements to qualify amounts paid or incurred for the right to use computers in the conduct of qualified research for the R&D credit: (1) The computer must be owned and operated by someone other than the taxpayer; (2) the computer must be located off the taxpayer’s premises; and (3) the taxpayer must not be the computer’s primary user.

Since their enactment and promulgation, neither Sec. 41(b)(2)(A)(iii) nor Regs. Sec. 1.41-2(b)(4) has been revised, and no proposed regulations have been issued. Moreover, the only mention of computer rental costs in the IRS Audit Techniques Guide: Credit for Increasing Research Activities (i.e., Research Tax Credit) IRC Section 41 (June 2025) is text from the statute itself. But while the statute and regulatory guidance have remained static, computers, technology, and their intersection with American business and industry have advanced by staggering amounts.

In 2025 Treasury issued final regulations regarding sourcing of digital and cloud transactions (T.D. 10022 and Regs. Secs. 1.861–18 and –19). In those regulations, Treasury took a fresh look at how taxpayers engage with technology and took a substance–over–form approach to crafting administrable guidance that carries out the spirit of the rules. However, as the gap between technology and the R&D credit regulations widens, taxpayers have struggled to apply the archaic computer rental rules of the 1980s to the modern ways in which they use computers for research purposes and claim the R&D credit, and a similar approach by Treasury is long overdue to update those regulations.

What is a ‘computer’?

The term “computer” has never been formally defined for the purposes of Sec. 41, though the modified accelerated cost recovery system rules in Sec. 168(i)(2)(B)(ii) describe it as follows:

The term “computer” means a programmable electronically activated device which —

(I) is capable of accepting information, applying prescribed processes to the information, and supplying the results of the processes with or without human intervention, and

(II) consists of a central processing unit containing extensive storage, logic, arithmetic, and control capabilities. [See alsoasset class 00.12, Information Systems, of Rev. Proc. 87–56.]

In the years leading up to the enactment of the R&D credit, computers used for research purposes primarily consisted of large, expensive, mainframe computers. “Time–sharing,” a term explicitly used in Regs. Sec. 1.41–2(b)(4), refers to a popular form of computing throughout the 1960s and 1970s where individual terminals were connected to a large central processing unit (as described in Sec. 168(i)(2)(B)(ii), quoted above). Users interfaced with the individual terminals, and their requests were processed by the central processing unit.

Fast forward to 2026, and computers are unrecognizable from those the drafters imagined while crafting the computer rental rules. Instead of renting access to a central processing unit that users access via a stationary terminal, users can access the cloud from anywhere in the world, and the cloud relies on a vast and complex array of servers and software to function. One could even argue that modern software has assumed the role of computers from the 1980s. Though “computer software” is defined separately from “computers” in the Internal Revenue Code (see Sec. 197(e)(3)(B) — “any program designed to cause a computer to perform a desired function”), that definition was added to the Code in 1993 and is similarly outdated for the research credit. Also, Sec. 41(b)(2)(A)(iii) uses the phrase “the right to use computers,” which naturally by extension would include the use of software.

So, in 2026, what is the “computer”? Is it a user’s local device that often is portable? Is it the physical server being accessed? Or is it the complex architecture of applications and logic that processes requests?

In the preamble to the proposed regulations regarding digital and cloud transactions (REG–130700–14, subsequently finalized by T.D. 10022), Treasury defined “cloud computing” as typically being “characterized by on–demand network access to computing resources, such as networks, servers, storage, and software.” In the abstract, cloud computing sounds strikingly similar to the format of time–sharing. Users’ personal devices (formerly, terminals) access servers (e.g., central processing units, graphics processing units, and tensor processing units) and connect wirelessly via the internet (rather than by being hardwired or by network connections). That said, how cloud computing exists practically does not align with the three qualification requirements described in Regs. Sec. 1.41–2(b)(4), as discussed further below.

1. The computer must be owned and operated by someone other than the taxpayer: At first glance, this first test seems relatively simple. However, although users typically access the cloud via their personal devices, defining the computer at the personal–device level seems to make little sense. Much like the terminals of old, users’ personal devices are a means to access infrastructure, containers, platforms, functions, and software hosted on third–party servers. That said, the ways in which taxpayers interact with cloud resources are rapidly evolving and do not always involve the full spectrum of computing services. Those ways include the following (see also preamble, REG–130700–14):

• Infrastructure as a Service (IaaS): Delivers on-demand infrastructure resources such as those for computing, storage, networking, and visualization;
• Containers as a Service (CaaS): Delivers and manages hardware and software resources to develop and deploy applications using containers (i.e., packages of necessary elements of an application);
• Platform as a Service (PaaS): Delivers and manages hardware and software resources to develop applications through the cloud;
• Function as a Service (FaaS): Allows users to build and deploy a small piece of code, or a function, that performs a task; and
• Software as a Service (SaaS): Provides the entire application stack, delivering an entire cloud-based application that customers can access and use.

Personal computers are much more powerful than terminals were, and when taxpayers employ cloud resources for only a portion of the computing process, who owns and operates the computer?

Alternatively, consider the situation in which a taxpayer leases specialized software used for research but due to information security concerns hosts the software locally on its own servers. From a legislative–intent perspective, this would seem to qualify — the taxpayer could not afford to develop the specialized research software and must lease it. In this case, the user has no right to the actual software beyond its use and has no access to the source code. However, due to technological advancements in how hardware and software are deployed, it is unclear whether the taxpayer would be deemed the computer’s owner and operator.

2. The computer must be located off the taxpayer’s premises: Assuming a taxpayer can clear the first hurdle regarding computer ownership, how they engage with development tools accessed via the cloud can vary. In a SaaS model, where the user accesses software and servers hosted by a third party, the computer seems clearly off the taxpayer’s premises. However, in the previous example where the taxpayer leases software and hosts it locally, does that place the computer on the taxpayer’s premises?

When the regulations were drafted, terminals lacked the computing power necessary to execute research tasks. Whether communicating via hardwired or network connections, computing was performed by the central processing unit, and results were returned to the terminals. Now that the hardware is more available and largely interchangeable, it is often the specialized software that taxpayers must rent or lease rather than purchase or produce.

3. The taxpayer must not be the primary user of the computer: The third consideration one must make when determining the qualification of computer rental costs is challenging for the same reasons cited above. Additionally, taxpayers must decide how to interpret “primary user.” Similar to the term “computer,” the term “primary” has not been defined for the purposes of Sec. 41. That said, analogies can be drawn. Generally, in the context of depreciation, “primary use” of property can be established by any reasonable method that is consistently applied (Regs. Sec. 1.168(i)-4(d)(2)(i)).

Assuming that it is reasonable to consider a taxpayer is the primary user of the computer if the taxpayer uses it more than 50% of the time, how exactly can taxpayers verify this? When a taxpayer accesses resources on demand, must that taxpayer also confirm with the service provider whether it was the primary user of those resources? Alternatively, must the taxpayer determine the specific servers its information was stored in to confirm whether those servers were in private or multitenant environments?

Suffice it to say, modernized regulations are needed to clarify the permissibility of computer rental costs in an era of cloud computing and remote access. When the statute was originally drafted in 1986, it was done in a very flexible manner; QREs included any amount paid or incurred to another person for the right to use computers in the conduct of qualified research activities. The statute and legislative history make no mention of the computer’s location or the ownership of the hardware.

Congress foresaw the outsized role computers would play in R&D and understood that it may not always be economically feasible to purchase such capital–intensive and specialized tools. It is time the regulations adopted a similar flexibility by moving away from the physical framework of mainframes and terminals to more modern technologies.

Editor

Mary Van Leuven, J.D., LL.M., is a director, Washington National Tax, at KPMG LLP in Washington, D.C.

For additional information about these items, contact Van Leuven at mvanleuven@kpmg.com.

Contributors are members of or associated with KPMG LLP.

The information in these articles is not intended to be “written advice concerning one or more federal tax matters” subject to the requirements of Section 10.37(a)(2) of Treasury Department Circular 230. The information contained in these articles is of a general nature and based on authorities that are subject to change. Applicability of the information to specific situations should be determined through consultation with your tax adviser. These articles represent the views of the authors only and do not necessarily represent the views or professional advice of KPMG LLP.