Patent royalty rates are quoted as percentages, which makes them look more precise than they are.
There is no schedule and no standard. A royalty rate is whatever two parties agree, anchored loosely on what similar patents have commanded before.
And the percentage is the smaller half of the question. A rate means nothing without the base it applies to — the revenue figure it is a percentage of. Negotiating the base is routinely worth more than negotiating the rate.
Typical rates by field
Reference points, not standards. Every one of these moves substantially on exclusivity, remaining life and how essential the patent is.
| Field | Typical running royalty | Why |
|---|---|---|
| Software and computing | 3–5% | Crowded prior art, design-around often feasible |
| Telecommunications | 3–6% | Standards exposure; large portfolios cross-licensed |
| Semiconductors | 2–5% | High volume, thin per-unit margins |
| Medical devices | 5–10% | High margins, regulatory barriers to design-around |
| Pharmaceuticals | 5–15%+ | Single patent can cover the entire product |
| Industrial and mechanical | 3–6% | Established comparables, mature markets |
| Consumer products | 2–5% | Low margins, price-sensitive |
| University / early-stage tech | 1–3% plus milestones | Unproven, requires further development |
The spread within a field is wider than the spread between fields. A foundational patent in consumer products can command more than a marginal improvement in pharmaceuticals.
Higher margins support higher rates. A licensee cannot pay 10% of net sales on a product with an 8% margin. Field averages largely reflect the margin structure of those industries.
The base matters more than the rate
This is the part most first-time licensors get wrong.
| Structure | Rate | Base | Annual royalty |
|---|---|---|---|
| Royalty on the finished device | 5% | $500 × 100,000 units | $2,500,000 |
| Royalty on the covered component | 5% | $20 × 100,000 units | $100,000 |
Same rate. Twenty-five times the money.
Which base is correct is a genuine legal question, not just a negotiating position. Where a patent covers one component of a multi-component product, the royalty base is generally the smallest saleable patent-practising unit rather than the whole product — unless the patented feature drives consumer demand for the entire thing.
Define the base precisely in the agreement.
| Base term | What to nail down |
|---|---|
| Net sales | Which deductions are allowed — returns, freight, taxes, discounts |
| Unit | What counts as one licensed unit |
| Covered product | Which SKUs, and what happens to future versions |
| Combination products | How to apportion when the licensed part is one of several |
| Sublicences | Whether sublicence income is included and at what rate |
Undefined deductions are where royalty income disappears. A "net sales" base with unlimited permitted deductions can be reduced substantially by a licensee acting entirely within the contract.
How rates actually get set
Comparable licences are the primary anchor. What have similar patents in similar fields commanded recently? This is the first question in any negotiation and the first question a court asks.
The 25% rule is dead as evidence. The old rule of thumb — the licensee pays roughly a quarter of expected profit — was rejected by the Federal Circuit in Uniloc v. Microsoft (2011) because it takes no account of the specific facts. It survives as a rough sanity check between negotiators, nothing more.
The Georgia-Pacific factors govern litigation. Fifteen factors from Georgia-Pacific Corp. v. United States Plywood Corp. (1970), which courts use to reconstruct a reasonable royalty. The ones that move the number most:
| Factor | Effect |
|---|---|
| Established royalty for the patent | Strongest evidence if one exists |
| Rates paid for comparable patents | The main anchor |
| Exclusivity and restrictions | Exclusive commands more |
| Commercial relationship of the parties | Competitors pay more than non-competitors |
| Utility over old modes | How much better than the alternative |
| Commercial success of the product | Larger pie to divide |
| Remaining patent term | Shorter window, lower rate |
| Portion of profit attributable to the invention | Apportionment |
| The hypothetical negotiation | What willing parties would have agreed when infringement began |
The hypothetical negotiation is the frame everything else feeds. Courts ask what a willing licensor and willing licensee would have agreed at the moment infringement started, assuming both knew the patent was valid and infringed.
Structure, not just rate
| Structure | How it works | Suits |
|---|---|---|
| Running royalty | Percentage of sales, paid periodically | Licensor confident in growth; requires audit rights |
| Lump sum | One payment, no reporting | Licensor wanting certainty or lacking monitoring resources |
| Hybrid | Upfront payment plus reduced running rate | Most negotiated deals |
| Per-unit | Fixed dollars per unit | Where sale price varies widely |
| Tiered | Rate steps down as volume rises | High-volume products |
| Minimum annual | Floor regardless of sales | Prevents a licensee shelving the technology |
Minimum annual royalties are the most important term in an exclusive licence. Without them, a licensee can take exclusive rights and do nothing, locking the patent away for the rest of its term while paying nothing.
Audit rights matter for running royalties. A percentage of a number the licensee calculates and reports is only as reliable as your ability to check it.
Worked example: negotiating a licence
A patent covering a fluid-control mechanism used in an industrial pump.
The opening positions
| Licensor asks | Licensee offers | |
|---|---|---|
| Rate | 6% | 2% |
| Base | Whole pump ($4,000) | The valve assembly ($240) |
| Structure | Running royalty | Lump sum $150,000 |
| Term | Life of patent | 3 years |
The base gap is worth more than the rate gap. At 10,000 units a year:
| Scenario | Calculation | Annual royalty |
|---|---|---|
| Licensor's position | 6% × $4,000 × 10,000 | $2,400,000 |
| Licensee's position | 2% × $240 × 10,000 | $48,000 |
| Split the rate only, licensee's base | 4% × $240 × 10,000 | $96,000 |
| Split the base only, licensee's rate | 2% × $4,000 × 10,000 | $800,000 |
Conceding the rate and winning the base beats the reverse by eight times. A negotiator focused on the percentage is arguing about the wrong number.
Where it lands
| Term | Agreed | Reasoning |
|---|---|---|
| Rate | 4.5% | Mid-range for industrial mechanical |
| Base | Valve assembly plus control module, $600 | Smallest saleable unit that practises the claims |
| Upfront | $100,000 | Credited against first-year royalties |
| Minimum annual | $75,000 | Licensor's floor |
| Term | Life of patent, rate drops to 2% in final 3 years | Reflects shrinking enforcement window |
| Audit | Annual, licensor pays unless variance exceeds 5% | Standard |
Annual royalty at volume: 4.5% × $600 × 10,000 = $270,000.
The step-down in the final three years is worth including. It reflects something real — a patent with a short remaining window is worth less — and conceding it early often buys a better base.
Royalties end when the patent does
Charging royalties beyond expiry is patent misuse. Brulotte v. Thys (1964), reaffirmed in Kimble v. Marvel (2015), makes a royalty tied to a single expired patent unenforceable.
Which makes remaining life a direct input to the rate. A licence signed on a patent with fifteen years left is worth several times one signed with four years left, on identical technology.
Structures that legitimately extend beyond expiry include agreements covering multiple patents with staggered expiry, and deals bundling know-how or trade secrets that have independent value. These need drafting care.
Most patents never reach a royalty at all. Of US utility patents granted in 2014, only 41.4% were maintained to full term — the other 58.6% were abandoned because owners judged the maintenance fee not worth paying. A patent nobody practises generates no royalty at any rate, which is why the licensing question and the renewal question are the same question. See the patent survival curve.
Rate versus selling outright
| Licence | Sale | |
|---|---|---|
| Payment | Over time, tied to performance | Once, at closing |
| Upside if product succeeds | Yours | The buyer's |
| Maintenance fees | Still yours | Buyer's problem |
| Enforcement burden | Still yours against third parties | Transfers |
| Administrative load | Reporting, audits, disputes | None after closing |
| Certainty | Low | High |
The maintenance fee point catches people out. A licensor keeps the patent and keeps paying $2,150, $4,040 and $8,280 to maintain it. Those fees have to be covered by the royalty stream before anything is profit.
A licence with modest running royalties and a third maintenance fee due can lose money. Running the arithmetic before signing is worth the hour. See how to sell my patent for the alternative route and what it pays.
Standard-essential patents and FRAND
A different regime applies where a patent is essential to an industry standard. If a product cannot comply with the standard without infringing, the patent holder has leverage no ordinary licensor has.
Standards bodies require a commitment in exchange. Members declaring essential patents typically undertake to license on fair, reasonable and non-discriminatory terms — FRAND, or RAND in US usage.
| Ordinary licence | FRAND commitment | |
|---|---|---|
| Duty to license | None | Obligation to offer |
| Rate freedom | Whatever is agreed | Must be fair and reasonable |
| Different terms per licensee | Permitted | Non-discriminatory |
| Injunction | Generally available | Restricted in most jurisdictions |
| Typical rate level | Field range | Often well below, given volume |
Royalty stacking is the underlying problem. A smartphone may read on thousands of declared-essential patents. If every holder charged 5%, the total would exceed the device price, so FRAND rates are set against the aggregate burden rather than each patent in isolation.
Rates are often quoted per unit rather than as percentages for this reason — a few cents to a few dollars per device, rather than a share of revenue.
If you hold a declared-essential patent, your rate is constrained by the commitment your standards body membership created, whether or not you were the one who made it. That commitment travels with the patent when it is sold.
Finding comparable rates
Comparables are the anchor for everything, so knowing where to look is the practical skill.
| Source | What it gives | Access |
|---|---|---|
| SEC filings (10-K, 8-K, exhibits) | Actual licence agreements filed by public companies | Free, EDGAR full-text search |
| Litigation records | Rates disclosed in damages rulings and expert reports | Free via PACER, some paywalled |
| University technology transfer offices | Published rate ranges and standard terms | Often free on institution sites |
| Industry surveys and licensing associations | Aggregated ranges by field | Usually paid |
| Court opinions applying Georgia-Pacific | Reasoned analysis of what is reasonable | Free |
SEC exhibits are the most undervalued source. Public companies file material licence agreements as exhibits, frequently with rates and bases intact. Searching EDGAR full text for licence terms in your technology area produces real agreements rather than survey averages.
Litigation damages rulings are the second. A court that has worked through the Georgia-Pacific factors on a comparable patent has effectively published a reasoned valuation, including why it rejected the figures each side proposed.
Adjust every comparable for the differences. Exclusivity, remaining term, field of use, the relationship between the parties, and whether the deal settled litigation. A rate agreed to end a lawsuit is not a market rate.
What caps any royalty rate
A licensee will pay up to roughly what not licensing would cost them, and no more. Every comparable in the world does not change that ceiling.
| The licensee's alternative | Effect on the rate you can command |
|---|---|
| Design around cheaply | Caps the royalty at the design-around cost |
| Design around expensively | Supports a high rate |
| Challenge validity | Caps the rate near the cost of an IPR plus risk |
| Switch to a substitute technology | Caps it at the switching cost |
| Ignore the patent and risk suit | Caps it at expected litigation cost × probability |
| No alternative at all | Rate limited only by product margin |
Design-around cost is the ceiling that matters most. If a competent engineer can circumvent the claims for $200,000, a royalty stream worth more than that in present value will not be paid — it will be engineered away instead.
Which makes claim scope the real determinant of rate. One broad claim covering the only viable approach supports a high royalty. Forty narrow claims describing one specific implementation support very little, because the alternatives are right there.
Validity risk discounts every rate. A licensee who believes there is a meaningful chance of invalidation will price that in, and PTAB institution rates have been volatile — roughly 65% in October 2024, falling to around 37% by February 2026.
Assess the alternatives before setting your number. A rate justified entirely by comparables, with no view of what the licensee could do instead, is a number without a floor under it.
Setting a patent royalty rate: the checklist
- Establish the base before the rate. Identify the smallest saleable unit that practises the claims, and decide whether the patented feature drives demand for anything larger.
- Find comparable licences in your field. This is the anchor for any negotiation and the primary evidence in any dispute.
- Check remaining patent life. It caps the total value of any running royalty and justifies a step-down in the final years.
- Decide running versus lump sum on whether you can monitor and audit. A percentage you cannot verify is an estimate.
- Define net sales precisely, listing every permitted deduction. Open-ended deductions are where royalty income quietly disappears.
- Insist on minimum annual royalties for any exclusive licence. Without them a licensee can shelve the technology and pay nothing.
- Include audit rights with a variance threshold that shifts costs to the licensee when they have under-reported.
- Price exclusivity properly. An exclusive licence excludes you too, and should carry a rate at the top of the field's range.
- Confirm the royalty stream covers the maintenance fees — $14,470 across a large entity patent's life — before treating any of it as profit.
- Set royalties to end at expiry, and structure multi-patent or know-how elements separately if payments need to continue.