Your new sensor ships with Wi-Fi and a cellular fallback. You designed the board, wrote the firmware, and bought every component from a reputable distributor — and you still may owe royalties to dozens of patent owners you have never heard of. The moment your product complies with a standard like Wi-Fi 6 or 5G, it necessarily uses inventions patented by the companies that built that standard. More than 100,000 patent families have been declared essential to 5G alone, and global royalties on these patents exceeded $35 billion in 2024. The system that keeps this from blocking your launch is called FRAND licensing, and if you ship anything connected, its rules decide what you pay, to whom, and on what terms.
This guide explains what a standard-essential patent is, what each word of "fair, reasonable, and non-discriminatory" actually promises, the two-sided standoff of hold-up versus hold-out, the royalty-base fight that moves the decimal point on what you owe, and the concrete steps to take before your standards-based product leaves the warehouse.
Standard-Essential Patents and the Bargain Behind Them
A standard-essential patent (SEP) protects an invention you cannot avoid while complying with a technical standard. Use 5G, Wi-Fi, Bluetooth, or HEVC video compression, and you infringe the relevant SEPs by definition — there is no way to implement the standard without practicing the patented technology. That is what makes SEPs different from ordinary patents, where you can usually design around the claims.
Standards come from standards development organizations (SDOs) such as ETSI (cellular, including 5G), IEEE (Wi-Fi), and ITU (telecom). These bodies ask participating companies to contribute their best technology to the standard — and to disclose patents that would be essential to it. The catch is obvious: once a patented invention is baked into a standard the whole industry must follow, the patent owner holds market power far beyond what the invention alone would command. It could refuse to license, or demand a ransom, effectively taxing every company that ships a compliant product. Economists call this the hold-up problem.
FRAND is the bargain that prevents it. As a condition of having its technology adopted into the standard, the SEP owner gives the SDO an undertaking — at ETSI it is irrevocable — to license its essential patents to any implementer on terms that are fair, reasonable, and non-discriminatory. In exchange, the owner gets its technology designed into an entire industry and collects royalties across it. You may also see the same commitment called RAND or F/RAND; the obligation is identical.
Two practical consequences follow. First, FRAND is a commitment to license, not a license itself: you still have to negotiate (or join a pool) and pay. Second, the commitment runs to all implementers, which is why a small IoT shop has the same theoretical right to a license as a smartphone giant — though, as the sections below show, exercising that right takes deliberate work.
What "Fair, Reasonable, and Non-Discriminatory" Actually Means
The three words sound like a pricing schedule. They are closer to a set of boundaries on a negotiation, and understanding each one tells you what you can and cannot demand.
Fair
Fairness is about the bargaining process more than the number. A FRAND offer is fair when it reflects good-faith dealing rather than exploitation of the implementer's lock-in: no tying the SEP license to unrelated patents you do not need, no conditioning a license on surrendering your own patent rights, and no royalty that captures value created by the standard itself rather than by the patented invention. Courts and regulators ask, in effect, whether the licensor is using standardization as leverage to extract more than the technology is worth.
Reasonable
Reasonableness is where the royalty math lives, and it is the most fought-over word of the three. A reasonable royalty is meant to approximate what willing parties would have agreed before the standard locked the industry in — the value of the patented contribution, not the value of being unavoidable. In practice, reasonableness is argued through comparable licenses (what similarly situated licensees actually paid), apportionment (isolating the patented feature's contribution from the rest of the product), and caps on royalty stacking (the aggregate burden when hundreds of SEPs read on one device). No SDO publishes a rate card; "reasonable" is established license by license, and sometimes court by court.
Non-Discriminatory
Non-discrimination means similarly situated implementers should get similar terms — the SEP owner cannot charge you ten times the rate it gave your direct competitor for the same portfolio. It does not mean everyone pays the identical price: volume differences, cross-licenses, geographic scope, and timing legitimately move rates. And it does not give you a most-favored-licensee clause by default; if you want protection against a better deal going to a rival later, you have to negotiate it into your agreement.
The honest summary: FRAND guarantees you a license on defensible terms, not a cheap one, and not without a negotiation. The rest of this guide is about how that negotiation actually plays out.
Hold-Up vs. Hold-Out: Pressure Runs Both Ways
The classic fear is hold-up: the SEP owner exploits your lock-in to demand supra-competitive royalties, or threatens an injunction to force your signature. FRAND commitments exist precisely to blunt this — the owner has already promised to license, which weakens any threat to shut you out, and in many jurisdictions an injunction against a willing licensee is difficult or unavailable.
But pressure runs the other way too, and licensors have a name for it: hold-out. An implementer can ship standard-compliant products today with no license at all, ignore or slow-walk every licensing letter, and force the SEP owner to sue in multiple countries to collect anything. The worst case for the implementer is eventually paying roughly what was asked years earlier — with the time value of money working in its favor — while the licensor bears all the enforcement cost. Implementers can manufacture standard-compliant devices without any upfront commitment to pay, which puts the burden of enforcing the SDO's licensing policy on patent owners.
This two-sided standoff shapes strategy on both sides. Licensors document their FRAND offers carefully and move toward enforcement when talks stall. Sophisticated implementers do the opposite of stonewalling: they respond to licensing letters promptly, make counter-offers, and keep a paper trail showing they are a willing licensee. Willingness is the single most protective posture an implementer can hold — it is what keeps injunctions off the table and keeps a later rate-setting proceeding focused on the number rather than on your conduct. Delay without engagement is the posture that converts a royalty negotiation into litigation.
The Royalty-Base Fight: Chip Price or Device Price?
The most expensive ambiguity in FRAND is the base to which the royalty rate applies. Two methodologies dominate, and they can differ by a factor of forty or more on the same product.
The smallest saleable patent-practicing unit (SSPPU) approach applies the royalty to the cheapest component that actually embodies the invention — for cellular SEPs, typically the baseband or modem chip, priced around $20. The logic is apportionment: the patented contribution is the connectivity, not the camera, the display, or the brand, so the royalty should scale with the connectivity component.
The end-device (entire market value) approach applies the royalty to the finished product — a smartphone with a net selling price of $800 to $1,200. Licensors argue that connectivity drives the device's value and that component prices understate the invention's contribution; some courts have agreed that where the patented feature drives demand for the whole product, the device price is the proper base.
Courts have split. Some U.S. decisions have required apportioning down to the smallest saleable unit closely tied to the claims; others have upheld end-device royalties as FRAND, with one court holding that ETSI's intellectual property policy neither requires nor precludes either base — what counts as FRAND depends on the facts. Outside the U.S., end-device licensing with a percentage royalty is the common commercial practice, giving implementers certainty without litigating every patent's validity.
For your budgeting, the takeaway is concrete: always ask which base a quoted rate assumes before comparing offers. A 1% rate on a $20 chip is $0.20 per unit; a 0.5% rate on a $600 device is $3.00 — fifteen times more despite the lower percentage. And when licensors quote per-unit dollar figures instead of percentages, the base question is already answered; compare those figures directly against pool rates.
How Implementers Actually Get Licensed
Few small companies negotiate with every SEP owner individually. Three paths cover nearly all real-world licensing.
Bilateral Licenses
Large implementers negotiate directly with major SEP holders — the cellular portfolio owners whose names appear on flagship phones' licensing disclosures. A bilateral deal can reflect your volume, your own patents available for cross-licensing, and your geographic footprint. For a small shipper, bilateral talks usually begin when a licensor's letter arrives; the correct response is engagement, not silence, ideally with counsel experienced in SEP negotiations.
Patent Pools
A patent pool bundles many owners' SEPs into a single license at a published rate, trading per-owner negotiation for one-stop shopping. The best-known example is Avanci, which licenses the major cellular SEP portfolios to automakers and IoT device makers at a flat per-unit rate for connected vehicles and devices. Video codecs have their own pools, such as Access Advance for HEVC. Pools deserve a hard look whenever your product fits their program: compare the pool's all-in per-unit cost against the sum of bilateral demands you would otherwise face, and check exactly which holders and standards the pool covers before assuming you are fully licensed.
Cross-Licenses
If your company holds patents of its own — even a modest portfolio in adjacent technology — those patents are negotiating currency. Cross-licensing, where each side grants rights to the other, can reduce or eliminate cash royalties between two portfolio holders. This path matters less for most small implementers but becomes relevant quickly if you acquire patents or join an SDO working group yourself.
Whichever path you take, verify essentiality before paying. Independent technical analysis finds that only about 35–45% of declared SEPs actually withstand scrutiny as truly essential — declarations are made by the patent owners themselves, often early and broadly. A declaration database entry is a claim, not a verdict. Reviewing which asserted patents genuinely read on your implementation, with technical and legal help, is routinely the highest-return step in the process.
What to Do Before You Ship a Standards-Based Product
Work through this checklist during development, not after the licensing letters arrive. Each step is cheaper now than under a deadline.
-
Inventory the standards in your bill of materials. List every connectivity and media standard your product implements — cellular generation, Wi-Fi version, Bluetooth profiles, video and audio codecs. Each one maps to a different set of SEP owners and pools. Your module and chipset vendors can often tell you which standards their components implement, but their license typically covers their component, not your finished device; confirm in writing where their coverage ends.
-
Search the SDO declaration databases. ETSI, IEEE, and other SDOs publish searchable databases of declared SEPs. Regular review gives early warning of which portfolios are likely to assert against your product and lets you budget before you have committed to a retail price that leaves no room for royalties.
-
Budget royalties per unit, early. Build a line item for SEP royalties in your unit economics alongside components and assembly. For connected vehicles and cellular IoT, published pool rates give a starting benchmark; for codecs, check the relevant pool's rate card. A product whose margin cannot absorb single-digit dollars per unit in licensing needs a different price or a different feature set — a decision you want in the design phase.
-
Compare pool versus bilateral coverage. If a pool covers your product category, model its all-in per-unit cost against expected bilateral demands. Note the gaps: pools rarely cover every SEP owner, so budget a reserve for holders outside the pool rather than assuming one license ends the matter.
-
Respond to licensing letters as a willing licensee. Answer promptly, ask for claim charts mapping asserted patents to your product, make timely counter-offers, and keep records of every exchange. Good-faith negotiation conduct is closely examined if a dispute ever reaches a rate-setting proceeding, and in the EU a mandatory FRAND determination procedure — up to nine months of structured negotiation before either side goes to court — now provides a formal safe harbor for parties that engage.
-
Get a freedom-to-operate review for your highest-risk standard. A focused review of the largest asserted portfolio against your implementation — testing essentiality, validity, and whether your supplier's license already covers you — costs far less than one year of unnecessary royalties or one month of litigation.
Keeping the Royalty Books Honest
SEP royalties are a per-unit cost that accrues as you ship but gets invoiced quarterly or annually, often a year or more after the units left the warehouse. That timing gap is where books go wrong: revenue is recognized now while the matching royalty expense sits unrecorded until a pool invoice or a settlement lands, overstating margin in the meantime. Accrue the expense as units ship, not when the bill arrives.
A per-unit accrual needs two inputs: units shipped per period and the blended royalty per unit across your pool licenses and bilateral deals. Revisit the blended rate whenever you add a standard, enter a new pool, or settle with a holder outside the pools — and keep a reserve line for asserted-but-unsettled portfolios, since an implementer that ships first and negotiates later owes for every unit already sold.
In a Beancount ledger, the month-end accrual for shipped units looks like this:
2026-10-31 * "Accrue SEP royalties — October shipments, 4,000 units x $3.85 blended"
Expenses:COGS:Royalties:SEP-Licensing 15,400.00 USD
Liabilities:Accrued-Royalties:SEP -15,400.00 USDWhen the pool invoice arrives and you pay it, clear the accrual rather than booking a second expense:
2026-11-20 * "Pay Q3 Avanci IoT pool invoice INV-2026-1184"
Liabilities:Accrued-Royalties:SEP 11,200.00 USD
Expenses:COGS:Royalties:SEP-Licensing 350.00 USD
Assets:Checking:Operating -11,550.00 USDThe $350 top-up is the true-up between your accrued estimate and the actual invoice — small when your blended rate is current, and a signal to update it when it is not. Tracking royalties in their own sub-accounts, separate from general license fees, also keeps the per-unit cost visible when you price the next hardware revision.
The Last Step Before Launch
FRAND licensing is ultimately a system for turning a legal minefield into a line item: the SEP owners get paid across the industry, and you get the right to ship without negotiating with a hundred patent holders from scratch. Your leverage as an implementer comes from preparation — knowing which standards you ship, what the pools charge, which asserted patents are genuinely essential, and having a paper trail that shows you negotiated in good faith. Build the royalty budget into your unit economics, accrue it as you ship, and answer the licensing letters when they come. The companies that get hurt by SEPs are rarely the ones that paid too much; they are the ones that priced a product as if the royalties did not exist.
