# How Should Standard-Essential Patent FRAND Royalties Be Valued in 2026?

iprs.cloud · September 27, 2026

> Direct Answer to the Question A defensible FRAND royalty valuation is a triangulated estimate, not a single formula and not a published universal rate...

## Direct Answer to the Question

A defensible FRAND royalty valuation is a triangulated estimate, not a single formula and not a published universal rate. “FRAND” generally means fair, reasonable, and non-discriminatory, and a standard-essential patent must be necessary to implement a standard before its owner can demand licensing on FRAND terms. The valuation should ask what royalty an efficient negotiator would have agreed before or after the patent’s essentiality became contested, using the same patent, standard, technology generation, and license scope relevant to comparable agreements.

**Also worth reading:** [What are the essential features of patent portfolio analytics software for IP counsel and product teams in 2026?](https://iprs.cloud/knowledge/what_are_the_essential_features_of_patent_portfolio_analytics_software_for_ip_counsel_and_product_teams_in_2026.php) · [How Should Companies Value SEP Royalties Before Accepting, Auditing, or Challenging a FRAND License?](https://iprs.cloud/knowledge/how_should_companies_value_sep_royalties_before_accepting_auditing_or_challenging_a_frand_license.php) · [What are the industry-standard best practices for AI patent pruning in 2026?](https://iprs.cloud/knowledge/what_are_the_industry-standard_best_practices_for_ai_patent_pruning_in_2026.php)

The strongest analysis usually combines a top-down approach based on the implementer’s total relevant licensing cost, a bottom-up replacement-cost approach based on the value of the patented technology, and real comparable licenses. A fourth check may examine the patent holder’s own portfolio, other declared essential patents, and the allocation of value within the relevant standard. These methods can produce materially different figures because they measure different economic questions: what the implementer can afford, what the patented alternative would cost, and what market parties actually paid for similar rights.

No reliable fixed threshold determines a FRAND royalty. A percentage of an implementer’s net revenue, a per-unit charge, or a component-based price can be useful, but each requires adjustment for geographic coverage, the relevant standard version, product features, sales channels, patent families, and whether the benchmark price already contains rights outside the asserted patent. The practical objective is not to produce the most favorable number for one side; it is to build a reproducible range, state assumptions, test sensitivities, and show why selected comparables deserve more or less weight than rejected ones.

## What FRAND Royalty Valuation Actually Measures

FRAND valuation measures the economic value of the licensed right under a specified set of conditions. The rate itself is only one input: the complete royalty base matters just as much. A 3% royalty on handset net sales, a 3% royalty on only the value added by a covered feature, and a fixed charge per covered device can create very different outcomes. Analysts should therefore define the licensor, licensee, products, territories, term, standard generation, number of patents, sublicensing rights, and any grant-back or cross-license conditions before doing any arithmetic.

FRAND does not mean the cheapest possible royalty. It also does not mean that each essential patent receives the same share of the implementer’s profit or a value equal to the implementer’s avoided costs. A technology may have low individual replacement cost while still being unavoidable, and an implementer may have a high total relevant licensing burden while the contribution of one patent family is small. The evidence must connect the selected rate to the benchmark and then show that the benchmark makes commercial sense in light of alternatives and bargaining positions.

The relevant counterfactual is also contested. Courts and economists may ask whether the technology was available before the patented innovation, whether standard-setting influenced adoption, whether the patent was capable of being declared essential, and whether comparing later portfolio agreements risks circular reasoning. WIPO’s work on bottom-up methods and mutually beneficial FRAND terms emphasizes that the commitment concerns access to standards while also requiring economically sustainable licensing. Thus, the valuation should be sensitive to both implementer cost and patent-holder incentives, without pretending that those goals always yield the same number.

## Top-Down, Bottom-Up, and Comparable-License Methods

A top-down method begins with a measurable aggregate, often the implementer’s net sales in covered products or a broader category such as smartphones. Analysts then estimate how much of that value can reasonably be assigned to the relevant patent families or denominator. If the benchmark product carries a US$200 average selling price and the royalty base is defined as net sales, a hypothetical 1% rate is US$2 per unit. The same 1% on a US$35 royalty base is only US$0.35, illustrating why the base must be stated precisely.

A bottom-up method asks what an efficient alternative technology would cost. The analysis estimates how much the patented feature improves system performance, cost, reliability, compatibility, or user experience, and then identifies the commercially and technically available substitute. Replacement cost is not necessarily the royalty rate, because a rational negotiator normally does not pay the entire value of an alternative innovation. It may instead pay a share that reflects bargaining power, remaining patent risks, and how much of the benefit can be captured without the technology.

A comparable-license method examines actual agreements, preferably involving similar standards, generations, licensors, and product markets. A 2015 smartphone license involving one family in one generation is more probative for that generation than a 2021 agreement covering several generations, though later data may still illuminate trends. Every comparable needs normalization for territory, term, exclusivity, affiliates, sublicensing, past sales, and payment conventions. A quoted royalty cannot be used responsibly without knowing whether it is a headline percentage, a litigation settlement, an aggregate portfolio rate, or an anonymized contractual figure.

| Feature | Top-Down Valuation | Bottom-Up Valuation | Comparable-License Valuation |
| --- | --- | --- | --- |
| Starting point | Implementer sales or total relevant licensing cost | Cost or benefit of a substitute technology | Observed arm’s-length licenses |
| Main economic question | How much value should the relevant rights receive? | What is the patented technology worth relative to alternatives? | What have similar rights traded for? |
| Common data need | Net sales, product mix, benchmark rate | Technical alternatives, performance data, engineering costs | Original license terms and sufficient comparability |
| Principal weakness | May assume the desired aggregate rate | Can mistakenly equate replacement cost with negotiated royalty | Data can be sparse, aggregated, or unlike the asserted patent |
| Best use | Cross-check a benchmark and royalty base | Test technological contribution and alternatives | Establish a market-centered range |

## The Royalty Base, Denominator, and Aggregation Questions
The royalty base is often more important than the numerical rate. A proper base should represent the economic unit in which the patented technology creates value, exclude products or transactions outside the license, and follow a clearly stated accounting convention. “Net sales” may mean invoice value after stated deductions, while “covered revenue” may include a broader set of channel transactions. If an implementer reports US$1 billion in net sales and the FRAND rate is 1%, the first-pass royalty is US$10 million, but the result changes materially if the royalty base is US$300 million or US$1.5 billion.

Aggregation raises a different problem. A portfolio agreement covering 3,000 patent families does not show that every family has equal essentiality or value. The FRAND commitment is generally portfolio-wide, but an adjudicator assessing a specific patent may still require evidence linking that family to the implementer’s technical contribution. Analysts should avoid both extremes: dividing one agreed portfolio rate mechanically by every patent, or including thousands of declarations without testing whether they are necessary, current, and technically enabled by the product.

A common analytical simplification is to model a standard product with identifiable layers, such as connectivity, processing, memory, software, display, enclosure, and distribution. Patent-relevant layers are then assigned shares of a suitable benchmark, but the allocation is not automatically scientifically objective. Technology layers can overlap, and some standards rely on several interacting innovations. The assumptions should therefore be visible in a sensitivity table: if the relevant layer moves from 15% to 10% of a US$100 base and the attributable FRAND rate is 1%, annual royalties move from US$150,000 to US$100,000.

## Practical Procedure for Building a Defensible Estimate

First, define the exact licensing proposition. Record the asserted family, jurisdictions, relevant ETSI or other standard version, implementation path, product population, and whether the estimate is for one component or an aggregate portfolio. Then establish the negotiating date and the information available to the parties, since rates can be informed by later license announcements without allowing hindsight to distort the benchmark. A 2026 analysis of a 2018 generation should distinguish information available in 2018 from evidence used only to test consistency.

Second, assemble reliable evidence. This may include an implementer’s US$100 net-sales denominator, an aggregate licensing budget of US$5 million, license agreements disclosed publicly, technical evidence about alternatives, and portfolio counts as of a fixed date. The analyst should convert each source into comparable units, document adjustments, and explain whether a number is observed, estimated, or hypothetical. Evidence from a 2020 license should not be silently presented as the market rate for a 2026 product merely because it is more recent.

Third, calculate several scenarios rather than one false point estimate. A useful range might use royalty bases of US$80 million, US$100 million, and US$120 million and candidate rates of 0.5%, 0.75%, and 1%. The resulting matrix tests whether the conclusion depends on a small change in the denominator. Fourth, select a central estimate only after weighing technical contribution, essentiality, commercial adoption, and comparable-license evidence. Legal and technical experts should review the assumptions, while the commercial model should show currency, escalation, payment timing, and the effect of volume.

Before filing or negotiating, perform an error and consistency check. Confirm that no rate is divided by a product population already embedded in a benchmark, that aggregate and component analyses are not mixed, and that tax, distributor deductions, and currency treatment are clear. Where a valuation is submitted in litigation, disclosures and confidentiality duties vary by forum. In a negotiation, a sensitivity-based presentation is usually more persuasive than a single unsupported percentage, because it demonstrates that the core conclusion survives reasonable disagreement about inputs.

## Common Valuation Mistakes and Weak Heuristics

The most common mistake is treating “FRAND” as a statutory royalty rate with one recognized number. It is not. A second error is selecting the smallest royalty found in any agreement and calling it the benchmark without considering whether the license covered an older standard, a narrow territory, a larger licensor portfolio, or different payment timing. A third error is using the implementer’s total licensing cost as though it were entirely attributable to the patents in suit. Total relevant licensing cost is an upper-bound discipline, not an automatic proof of each family’s value.

Other weak methods rely on fixed product-category assumptions, such as applying one smartphone percentage to every generation, or assuming that the declared-essential patent count determines the licensor’s share. The Unwired Planet decisions in the United Kingdom reinforced the importance of the patent’s contribution to the value of the relevant standard and the implementer’s own value beyond the patented technology. That does not create a universal formula, but it undermines simplistic methods that identify the royalty solely through portfolio size or total net sales.

An especially fragile shortcut treats 25% of the patented technology’s value as the royalty. That figure derives from a judicial method used in U.S. apportionment analysis, particularly the approach discussed in Ericsson v. D-Link, and is not a worldwide FRAND safe harbor. Similarly, the European Commission’s former 6.5% royalty cap under its Antitrust SEP framework should not be treated as a current universal ceiling. Courts may consider whether a rate exceeds or falls below a benchmark, but comparison with such thresholds is not the same as presuming that the threshold is legally dispositive in every jurisdiction.

## Timing, Costs, and Operational Decisions

The right time to calculate FRAND royalties is before major product design decisions, portfolio negotiations, acquisitions, or litigation positions become locked into an undifferentiated narrative. Early analysis is especially useful when choosing among technically similar implementations, because alternative designs can alter the value of one or more patent families. For counsel, a first-pass model can often be assembled from known products, public agreements, and a clearly limited set of assumptions, but credible expert work may require technical mapping, comparable-license review, and economic analysis.

There is no responsible single worldwide price for a FRAND valuation because scope drives cost. A narrowly scoped calculation for one family, one product, and one territory may require less work than a portfolio model covering multiple countries and three standard generations. Expert fees should be quoted for defined deliverables, data rooms, number of experts, technical complexity, and number of scenarios rather than advertised as a guaranteed universal rate. Costs can rise sharply where confidential sales, internal alternatives, or complex royalty accounting must be reconciled.

For product teams, the practical trigger is a contemplated design change that could affect declared essentiality or patent coverage. For licensing teams, it is a new license proposal or major renewal. For investors and IP registries, the trigger may be portfolio diligence before an acquisition or an update of declared patent records. A reasonable initial deadline is 30 to 60 days for a preliminary issue map and scenario model, followed by a more detailed 60- to 180-day analysis if the matter proceeds toward negotiation or proceedings. Those periods are planning ranges, not legal deadlines, and should be shortened where an injunction, technical design freeze, settlement conference, or limitation issue is imminent.

Timing must also account for legal deadlines. In the United States, 35 U.S.C. § 286 generally provides a six-year limitations period for patent damages, and reasonable royalty damages for certain activities predating patent issuance may be considered under 35 U.S.C. § 154(d). In the United Kingdom, the Patents Act 1977 as amended now permits recovery in respect of infringement up to ten years before the action, subject to the statutory framework. These rules affect when historical revenue is relevant; they do not tell an analyst what a FRAND rate should be. Separate technical validity, infringement, essentiality, jurisdiction, and damages questions should not be collapsed into one unexplained number.

## What the Final Valuation Should Deliver

A final valuation should contain a defined scope, a source register, a transparent data set, normalized comparable-license analysis, and at least one independent cross-check. It should show a low case, central case, and high case, explain the uncertainty, and distinguish aggregate portfolio economics from the value of a particular family. A central estimate of 0.8% on a US$200 million base produces US$1.6 million in annual royalties before any separately agreed escalator, minimum, or maximum, while 0.6% produces US$1.2 million and 1.0% produces US$2 million. Those calculations are useful only if the base and rate are economically connected.

The strongest conclusion will often be a range such as 0.6% to 1.0%, accompanied by a central estimate of 0.8% and sensitivity analysis, rather than a claim that 0.8% is objectively “the” FRAND rate. The report should state that the estimate depends on the product, standard generation, territory, patent scope, comparable set, and negotiation date. It should also identify what evidence could change the range, such as a closer comparable, a reliable alternative-cost study, revised sales data, or a finding that a feature is optional rather than necessary to implement the standard.

Registry and workflow software can help maintain patent-family records, standard-version metadata, license comparisons, assumptions, and version history. Software does not replace technical or economic judgment, and automating an unsupported royalty assumption can make a weak conclusion look more precise. The correct use of a registry SaaS platform is to improve provenance, consistency, and reviewability. For legal and product teams, that operational discipline is often more valuable than forcing every FRAND matter into a false universal rate.

## Quick answers

### Is there a single standard FRAND royalty percentage?

No. FRAND is a negotiation and adjudication framework, not a fixed worldwide percentage. The appropriate result depends on the relevant standard, technology generation, patent scope, territory, product economics, alternatives, and comparable licenses.

### What is the best primary method for valuing a FRAND royalty?

Most defensible analyses combine top-down, bottom-up, and comparable-license methods. A single method may provide a useful cross-check, but agreement is more credible when the methods are reconciled and their assumptions are disclosed.

### Does a 25% royalty contribution cap determine a FRAND rate?

No. The 25% figure is associated with a judicial apportionment approach and is not a universal FRAND safe harbor. It may be relevant in some disputes, but the ultimate result still depends on the evidence, jurisdiction, and applicable law.

### How should comparable SEP licenses be selected?

The closest comparables normally involve similar products, standard generations, technical contributions, patent portfolios, territories, and license terms. Analysts should normalize differences and avoid treating an aggregate portfolio rate as the value of one family without analysis.

### How long does a FRAND royalty analysis take?

A preliminary scenario model may be prepared in 30 to 60 days when core data are available. A defensible multi-territory, multi-generation, or litigation-grade analysis can require 60 to 180 days or longer, depending on technical complexity and evidence.

Canonical: https://iprs.cloud/knowledge/how_should_standard-essential_patent_frand_royalties_be_valued_in_2026.php
Markdown: https://iprs.cloud/knowledge/how_should_standard-essential_patent_frand_royalties_be_valued_in_2026.php/index.md