The accounting rules for intellectual property turn on one distinction: if your company developed the IP internally, you almost always expense the costs as you incur them; if you bought the IP from someone else, you capitalize it on the balance sheet at fair value and then either amortize it over its useful life or test it for impairment each year. Everything else in intellectual property accounting rules — the valuation methods, the amortization schedules, the tax treatment, the disclosures — flows from that single call.
The Capitalize-or-Expense Rule
Under U.S. GAAP, costs tied to internal research and development are expensed as incurred. The FASB’s Accounting Standards Codification Topic 730 sets the rule, and the IRS uses the same framework to identify what counts as R&D.1Internal Revenue Service. Appendix E – Directive Definitions The reasoning is conservatism: most R&D carries too much uncertainty about commercial viability to justify treating it as an asset.
The practical consequence is striking. A company can pour hundreds of millions into research that eventually produces a blockbuster product, and none of those internal costs will sit on the balance sheet as an asset. The economic value is real. GAAP keeps it off the books until somebody else buys it.
IP acquired from a third party or picked up in a business combination goes on the balance sheet at fair value on the acquisition date. The recorded cost includes the purchase price plus directly attributable costs needed to put the asset into service, such as legal fees for due diligence and registration fees to transfer title. In a business combination, the acquirer allocates the total purchase price among all identifiable assets based on their respective fair values; any portion that cannot be assigned to a specific identifiable asset becomes goodwill.
Software: The Big Exception
Internal-use software is the main carve-out from the expense-everything rule. ASC 350-40 splits software development into three stages and treats each differently:
- Preliminary project stage. All costs are expensed. This covers research, vendor evaluations, and feasibility work.
- Application development stage. Costs tied directly to building the software are capitalized once management commits to funding the project and the preliminary stage is complete. Capitalizable costs include coding, testing, and installation. Training and data conversion costs stay in expenses.
- Post-implementation stage. Costs revert to being expensed. Routine maintenance and training after go-live fall here.
For software built for sale or licensing to external customers, ASC 985-20 applies instead. Capitalization cannot begin until the product reaches technological feasibility, typically shown by completing a detailed program design or producing a working model.2U.S. Securities and Exchange Commission. SEC EDGAR Filing – Software Development Costs Accounting Policy Everything before that point is expensed as R&D.
Cloud computing and SaaS arrangements add another layer. When a company pays for a hosted service rather than buying or building software it controls, the arrangement is usually a service contract rather than a software license. Under ASC 350-40, implementation costs such as custom configuration and integration work during setup can still be capitalized, but ongoing subscription fees and training are expensed as incurred.
The Patent Filing Fee Wrinkle
One detail catches people off guard. The legal fees and filing costs to register a patent can be capitalized, even when the underlying invention was developed internally and all the R&D costs behind it were expensed. The rationale is that these costs secure the legal right itself, not the research that produced the invention.3U.S. Securities and Exchange Commission. SEC EDGAR Filing – Intangible Assets Accounting Policy
How the Four Types of IP Are Classified
Before anything hits the balance sheet, the IP needs to be classified correctly. Classification drives amortization schedules, impairment testing frequency, and disclosure requirements. The dividing line that matters most is finite useful life versus indefinite useful life.
Patents. A utility or plant patent generally lasts 20 years from the date the application was filed in the United States.4Office of the Law Revision Counsel. 35 USC 154 – Contents and Term of Patent The built-in expiration makes patents a finite-life asset.
Trademarks and trade names. Protection is renewable indefinitely as long as the mark stays in active use, so these assets are typically classified as having an indefinite useful life. No amortization. Annual impairment testing instead.
Copyrights. For works created after January 1, 1978, copyright protection generally lasts for the life of the author plus 70 years.5Office of the Law Revision Counsel. 17 USC 302 – Duration of Copyright That lengthy but finite term makes copyrights a finite-life asset, though the economic useful life is often much shorter than the legal life.
Trade secrets. Protection lasts only as long as the owner maintains confidentiality. There is no statutory expiration, which makes the accounting useful life harder to pin down.
When a company acquires another business, additional identifiable intangibles often emerge, including customer relationships, non-compete agreements, and order backlogs. Each one needs the same finite-or-indefinite call before you can account for it.
Valuing IP You Acquire
Establishing fair value for acquired IP is not optional. ASC 820 defines fair value as the price that would be received to sell the asset in an orderly transaction between market participants at the measurement date. The work usually goes to a valuation specialist, who chooses among three established approaches.
Market Approach
This method estimates value by looking at what similar IP has sold for in recent arm’s-length transactions. It is the most intuitive approach, but it rarely serves as the primary method for unique IP because most proprietary technology and distinctive brands lack a deep market of comparable sales. When good transaction data does exist, market-based values tend to carry the most credibility with auditors because they rely on observable inputs rather than projections.
Income Approach
The income approach is the workhorse for unique assets. It focuses on the future cash flows the IP is expected to generate and converts them to a present value. Two techniques dominate.
The discounted cash flow method projects the incremental cash flows attributable solely to owning the IP and discounts them using a rate that reflects the asset’s risk profile. The relief-from-royalty method estimates what the company would have to pay in royalties if it licensed the IP from someone else instead of owning it; a market royalty rate is applied to projected revenues and the hypothetical royalty stream is discounted to present value. The second approach is especially common for trademarks and established technology.
Every income-based method shares the same weakness: it depends heavily on forward-looking projections and the choice of discount rate. Small changes in either input can produce meaningfully different valuations, which is why auditors scrutinize the assumptions closely.
Cost Approach
The cost approach estimates value based on what it would cost to recreate the IP from scratch or build something with equivalent functionality, reduced for physical, functional, and economic obsolescence. It works best for relatively new IP or assets whose income potential is hard to forecast, and it functions as a ceiling on value because a rational buyer would not pay more than the cost to replicate the asset. For established trademarks or proven technology where economic value far exceeds historical creation cost, the cost approach typically understates fair value and is used only as a supporting data point.
Amortization and Impairment After It’s on the Books
Once IP sits on the balance sheet, two mechanisms keep its carrying value honest: amortization for finite-life assets, and impairment testing for everything capitalized.
Amortizing Finite-Life Assets
Patents, copyrights, and other finite-life IP are amortized over the period the asset is expected to contribute to cash flows. The useful life equals the shorter of the legal life and the economic life. A patent with 18 years of legal protection remaining but only 8 years of commercial relevance gets amortized over 8 years.
The default method under ASC 350 is straight-line amortization. If the company can reliably determine that the economic benefits are consumed in a different pattern, the amortization method should reflect that pattern. In practice most companies use straight-line, because demonstrating an alternative consumption pattern is hard to support with evidence.
Impairment for Finite-Life Assets
Finite-life IP does not need annual impairment testing. Testing is triggered when events or circumstances suggest the carrying amount may not be recoverable. Triggering events include a major adverse shift in the business climate, a competitor’s technological breakthrough, or a significant drop in the asset’s market value.
When a trigger exists, the test runs in three steps. First, confirm whether impairment indicators are actually present. Second, run a recoverability test by comparing the asset’s carrying amount to the total undiscounted future cash flows expected from using and eventually disposing of it. If undiscounted cash flows exceed the carrying amount, no impairment is recognized, even if fair value is lower than book value. Third, if the asset fails the recoverability test, the loss equals the amount by which carrying value exceeds fair value, and it hits the income statement immediately. Once recognized, an impairment loss cannot be reversed under U.S. GAAP.
Impairment for Indefinite-Life Assets
Indefinite-life IP, such as certain trademarks, is never amortized but must be tested for impairment at least once a year, and more frequently if circumstances suggest the asset may be impaired.6Financial Accounting Standards Board. Goodwill Impairment Testing The annual test compares fair value directly to the carrying amount. Any shortfall is recognized immediately.
Companies have the option to perform a qualitative assessment first, sometimes called Step Zero. This evaluates factors like macroeconomic conditions, industry trends, and company-specific events to determine whether it is more likely than not that fair value has dropped below carrying amount. If the qualitative assessment indicates no impairment is likely, the company can skip the full quantitative test for that period, which saves significant time and cost in years when the asset is clearly not impaired.
How the Tax Rules Differ
Tax rules for IP often diverge sharply from GAAP, creating book-tax differences that show up as deferred tax assets or liabilities on the balance sheet.
Acquired Intangibles: Section 197
When a company acquires intangible assets as part of buying a business, most of those assets fall under IRC Section 197, which requires amortization over a flat 15-year period beginning in the month of acquisition. The rule covers goodwill, going concern value, patents, copyrights, trademarks, customer-based intangibles, covenants not to compete, and government-granted licenses.7Office of the Law Revision Counsel. 26 USC 197 – Amortization of Goodwill and Certain Other Intangibles
The 15-year period is mandatory regardless of the asset’s actual useful life. A patent with 5 years of remaining legal protection still gets amortized over 15 years for tax, while GAAP would amortize it over 5. A trademark that GAAP treats as indefinite-life and never amortizes will generate annual tax deductions over 15 years. These mismatches are among the most common sources of deferred tax items for companies with significant acquired IP.
Internal R&D: Section 174 and 174A
The tax treatment of internally developed IP has shifted more than once in recent years. Before 2022, companies could generally deduct domestic R&D spending immediately. A provision in the 2017 tax legislation changed that, requiring companies to capitalize and amortize domestic R&D costs over 5 years (15 years for foreign research) for tax years beginning after December 31, 2021.8Office of the Law Revision Counsel. 26 USC 174 – Amortization of Research and Experimental Expenditures
In 2025, Congress enacted Section 174A through the One Big Beautiful Bill Act, permanently restoring immediate expensing for domestic research and experimental expenditures for tax years beginning after December 31, 2024. Foreign R&D costs still must be capitalized and amortized over 15 years. Software development costs are treated as R&D expenditures eligible for immediate expensing under the domestic rule.
For GAAP, internal R&D is expensed immediately under ASC 730. For tax, domestic R&D is also now immediately deductible again. The brief window from 2022 through 2024, when tax required capitalization while GAAP required expensing, created deferred tax assets on many company balance sheets that are still unwinding.
Where IP Shows Up on the Financial Statements
On the balance sheet, intellectual property appears as a non-current asset under intangible assets, reported net of accumulated amortization and any recognized impairment losses. On the income statement, amortization expense typically falls within operating expenses, and impairment losses appear either there or as a separate line item depending on materiality. On the cash flow statement, acquiring IP shows up as a cash outflow under investing activities, while amortization expense is added back to net income in the operating section under the indirect method because it is a non-cash charge.
The footnote disclosures required under ASC 350-30-50 are where analysts get the detail they need. For finite-life intangible assets, companies must disclose the gross carrying amount and accumulated amortization for each major class, the total amortization expense for the period, and estimated aggregate amortization expense for each of the next five fiscal years. For indefinite-life intangibles, the total carrying amount and a breakdown by major class are required. When an impairment loss is recognized, the notes must describe the impaired asset, the facts leading to the impairment, the amount of the loss, the method used to determine fair value, and the income statement line where the loss appears.
The IFRS Difference for Internally Developed IP
If you report under or benchmark against IFRS, one difference changes the comparison. Under U.S. GAAP, nearly all internal R&D is expensed immediately. IFRS takes a different approach through IAS 38, which separates the R&D process into a research phase and a development phase.9IFRS Foundation. IAS 38 Intangible Assets
Research expenditure is always expensed under IFRS, as it is under U.S. GAAP. Development expenditure that meets specified criteria, including demonstrated technical feasibility, intent to complete the asset, and the ability to measure costs reliably, must be capitalized as an intangible asset. Two companies doing identical R&D work can report very different balance sheets and earnings profiles depending solely on which framework they follow. A U.S. GAAP reporter with heavy internal R&D will show lower assets and higher current-period expenses than an IFRS reporter in the same situation, all else being equal.