Inside the Methods That Price Patents, Brands, and Software

In 1975, intangible assets made up about 17% of the S&P 500's market value. By the end of 2025, Ocean Tomo puts that figure at roughly 92%. The accounting machinery for pricing what sits behind that number has not caught up with how much of it there now is.

How intangible value overtook the balance sheet

Ocean Tomo's Intangible Asset Market Value study calculates the figure simply: market capitalization minus net tangible asset value, tallied at each calendar year-end. Tracked for the S&P 500 since 2017, the study now spans 50 years of US data. According to study co-author James E. Malackowski, the shift accelerated sharply between 1985 and 2005, when the intangible share rose from 32% to 79%, a 47-percentage-point move in two decades. It kept climbing after that: 84% by 2015, past 90% during the 2020 COVID-era update, and to roughly 92% by year-end 2025, an inversion Ocean Tomo compares in scale to the Industrial Revolution, compressed into a single working lifetime.

That 92% figure is a residual, not an appraisal. It tells a reader how much of the market's pricing decisions can't be explained by what's on the balance sheet as physical property, plant, and inventory. It says nothing about which specific patent, brand, or piece of software is worth what. That question is where three named accounting approaches take over.

Three ways to put a number on an idea

Under IAS 38, an intangible asset is a non-monetary asset without physical substance that is either separable or arises from a contractual or legal right. Patents, trademarks, licences, and computer software all qualify. Recognition still requires two conditions: it must be probable that future economic benefit will flow to the entity, and the cost must be reliably measurable. For assets a company builds itself, IAS 38 draws a hard line between research spending, which must be expensed, and development spending, which can only be capitalized once technical feasibility, intent to complete, and several other conditions are all demonstrated at once.

IFRS 13 then gives three routes to a fair-value number: the market approach (pricing from comparable transactions), the cost approach (replacement cost less obsolescence), and the income approach (discounted future cash flow, including relief-from-royalty and option-pricing variants). In practice, the market approach is the hardest to use for intangibles: genuinely comparable, arm's-length sales of a similar patent or trademark are rare. That scarcity is why the income approach dominates for brands and much of patent valuation, while the cost approach tends to govern internally built software instead.

Inside the relief-from-royalty calculation

The relief-from-royalty method values an asset by asking what the owner avoids paying by not having to license it from someone else. Applying it involves forecasting the revenue the trademark, patent, or technology is expected to generate, then attaching a royalty rate drawn from real, comparable licensing agreements, often sourced through databases such as RoyaltySource, RoyaltyRange, or ktMINE. That rate is applied to the forecast revenue stream, and the resulting after-tax royalty savings are discounted to present value. In the United States, a trademark valued this way is generally amortized straight-line over 15 years for tax purposes under Internal Revenue Code Section 197.

Brand Finance's annual Global 500 ranking is this same logic run at portfolio scale: forecast brand-attributable earnings, a royalty rate, and a brand-strength score combine into a single dollar figure for each company's name. The 2026 edition puts the top five brands at Apple ($607.6 billion, up 6%), Microsoft ($565.2 billion, up 23% and above $550 billion for the first time), Google ($433.1 billion, up 5%), Amazon ($369.9 billion, up 4%), and Nvidia ($184.3 billion, up 110% on AI-infrastructure demand, overtaking Facebook and Walmart).

What software's accounting rules leave off the books

Software follows a different path than brands, because most of it is built in-house rather than licensed. Under ASC 350-40, costs during the preliminary project stage, exploratory research, technology evaluation, vendor demos, are expensed as incurred. Only once management has authorized funding and completion is probable does the application-development stage begin, and only costs incurred there get capitalized. Training and maintenance costs are expensed even after that. FASB's ASU 2025-06 replaces the old three-stage language with two triggering criteria, management commitment and probable completion, but keeps the same underlying split: research and exploration don't become an asset, only the build itself does.

IAS 38 draws an equivalent line for any internally generated intangible, patent, trademark, or software alike: research is always expensed, and development is capitalized only if a company can demonstrate technical feasibility and several other conditions simultaneously. The practical effect is that a meaningful share of what a technology company actually spends money discovering never shows up as an intangible asset at all. That gap sits underneath even the raw filing numbers: WIPO recorded 3.7 million patent applications filed worldwide in 2024, a record and the fifth straight year of growth, led by China (1.8 million), the United States (501,831), and Japan (419,132). A rising filing count measures activity, not worth. What ends up on a balance sheet, or in a licensing negotiation, still depends on which of the three approaches, cost, market, or income, gets applied, and on the royalty rate, discount rate, and revenue forecast that go into it.