Document type: Article Practice area: Intellectual Property — Patents Jurisdiction: United States Last reviewed: 5 September 2026

The question that licensing programmes get wrong

A patent gives its owner the right to exclude others from making, using, offering to sell, selling, and importing the claimed invention. That right is set out in 35 U.S.C. § 271, and for the term granted under 35 U.S.C. § 154 it is a genuinely powerful thing. What it is not is a right that attaches to a physical object and follows it forever.

The confusion arises because patent owners think in terms of the patent and buyers think in terms of the thing. The patent owner has a claim; the claim covers a category; every article in the category infringes unless authorized. So a patent owner naturally reasons: if I authorize a sale on condition that the buyer use the article only once, then a second use is unauthorized, and an unauthorized use of a patented article is infringement.

That reasoning is wrong, and it has been wrong for a very long time, but it took until 2017 for the Supreme Court to say so without qualification. In Impression Products, Inc. v. Lexmark International, Inc., 581 U.S. 360 (2017), the Court held that a patentee's decision to sell an article "exhausts all of its patent rights in that item, regardless of any restrictions the patentee purports to impose or the location of the sale." Not most rights. Not rights other than the ones expressly reserved. All of them.

This article is about what that means in practice: what still works, what never worked, and how a licensing programme should be built now that downstream control cannot be achieved through patent law.

Where the doctrine comes from

Exhaustion is older than the Federal Circuit, older than the 1952 Patent Act, and older than most of the industries that now argue about it.

In Adams v. Burke, 84 U.S. 453 (1873), an assignee held rights to coffin lids within a ten-mile radius of Boston. He sold lids to an undertaker who used them just outside the circle. The Court held there was no infringement: "when the patentee, or the person having his rights, sells a machine or instrument whose sole value is in its use, he receives the consideration for its use and he parts with the right to restrict that use." The patentee got paid. The buyer got the article. That transaction was complete.

The high-water mark of the opposite view came and went quickly. In Motion Picture Patents Co. v. Universal Film Manufacturing Co., 243 U.S. 502 (1917), a projector was sold with a notice restricting it to use with film supplied by the patentee. The Court refused to enforce the restriction as a matter of patent law, observing that the patent covered the projector, not the film, and that the patentee was attempting "to derive its profit, not from the invention on which the law gives it a monopoly, but from the unpatented supplies with which it is used." That is the tying instinct that runs through exhaustion cases even when nobody uses the word.

United States v. Univis Lens Co., 316 U.S. 241 (1942), added the piece that matters for component sales. Univis sold lens blanks — unfinished articles that had no use except to be ground into finished lenses, which were covered by Univis's patents. The Court held that the sale of the blanks exhausted the patents on the finished lenses, because the blanks embodied the essential features of the patented invention and had no reasonable non-infringing use. Sell the thing that only works one way, and you have sold the patent rights in the way it works.

Against that line stands General Talking Pictures Corp. v. Western Electric Co., 305 U.S. 124 (1938), which remains good law and is the single most important case for anyone drafting a licence. Western Electric licensed a manufacturer to make and sell amplifiers only for private, non-commercial use. The manufacturer sold amplifiers for commercial theatre use, knowing of the restriction; the buyer also knew. The Court held that the sale was outside the licence, was therefore not authorized, and did not exhaust — the patentee could sue the buyer.

The distinction that survives from these cases is the one that licensing lawyers must internalize:

A limitation on what a licensee is authorized to do defines whether a sale is authorized at all. A restriction imposed on a purchaser in an authorized sale is a contract term, not a patent right.

Quanta: the component problem

Quanta Computer, Inc. v. LG Electronics, Inc., 553 U.S. 617 (2008), applied Univis to the modern electronics supply chain and produced a result that a great many licensing programmes were not built for.

LGE licensed Intel to make and sell microprocessors and chipsets practising LGE's patents. The licence said, in terms, that no licence was granted to any third party to combine licensed products with non-Intel components. A separate agreement required Intel to notify its customers of that limitation, and Intel did. Quanta bought Intel chips, combined them with non-Intel memory and buses, and LGE sued.

The Court held the patents exhausted. Three points did the work.

First, method claims are not immune. LGE argued that method patents cannot be exhausted because methods are not sold. The Court rejected this, noting that the opposite rule would let any patentee "shield practically any patented item from exhaustion" by drafting method claims alongside apparatus claims.

Second, the Univis test governs incomplete articles: exhaustion applies where the article sold substantially embodies the patent — where its only reasonable and intended use is to practise the patent and where it embodies the essential features of the invention. The Intel chips had no use except to be incorporated into computers in a way that practised LGE's patents.

Third, and this is the part that catches drafters, the sale was authorized. The licence permitted Intel to sell the products. The notice requirement was a separate covenant; breaching it, or a customer disregarding it, did not convert an authorized sale into an unauthorized one. The Court was explicit that "the License Agreement authorized Intel to sell products that practised the LGE Patents. No conditions limited Intel's authority to sell products substantially embodying the patents."

The lesson is that a limitation must be on the grant to prevent exhaustion. A promise to notify customers is not a limitation on the grant. Neither is a covenant that the licensee will use best efforts to discourage a particular downstream use. If the licensee is authorized to sell, and it sells, the patent rights in that article are gone.

Impression Products: the end of the conditional sale

For twenty-five years, the Federal Circuit's decision in Mallinckrodt, Inc. v. Medipart, Inc., 976 F.2d 700 (Fed. Cir. 1992), had held that a patentee could sell an article subject to a lawful, clearly communicated restriction and enforce that restriction through an infringement action against a violator. Single-use-only medical devices, reduced-price cartridge programmes, and field-restricted equipment sales were all built on it.

Impression Products dismantled it.

Lexmark sold toner cartridges in two flavours: full price with no restriction, and a discounted "Return Program" cartridge sold on condition that the buyer use it once and return it to Lexmark. Impression Products acquired used Return Program cartridges, had them refilled, and resold them in the United States. It also imported cartridges Lexmark had sold abroad.

The Court held for Impression Products on both counts.

On domestic sales: "A patentee's decision to sell a product exhausts all of its patent rights in that item, regardless of any restrictions the patentee purports to impose or the location of the sale." The restriction may be a perfectly good contract term, enforceable against the buyer who agreed to it, but it is not enforceable as a patent right against anyone.

On foreign sales: the Court aligned patent law with the copyright rule of Kirtsaeng v. John Wiley & Sons and held that an authorized sale outside the United States exhausts U.S. patent rights just as a domestic sale does. This overruled the Federal Circuit's contrary rule from Jazz Photo Corp. v. International Trade Commission, 264 F.3d 1094 (Fed. Cir. 2001).

The Court's reasoning was rooted in the common law's antipathy to restraints on alienation, and its illustration is worth quoting because it explains the doctrine better than the doctrine explains itself. Consider, the Court said, a shop that restores and sells used cars: the business would grind to a halt if a patentee could sue over "the sale of a car" containing components sold subject to restrictions decades earlier. "Extending the patent rights beyond the first sale would clog the channels of commerce, with little benefit from the extra control that the patentees retain."

Critically, the Court preserved General Talking Pictures and did so explicitly. A patentee may still limit a licensee's authority. What it may not do is retain patent rights in an article once an authorized sale of that article has occurred.

What survives

The doctrine, stated as a working rule:

1. Limitations on a licensee's grant are enforceable and prevent exhaustion. If a licence authorizes manufacture and sale only for a defined field, a sale outside the field is unauthorized. The article is not exhausted, and the patentee can sue the licensee for infringement (not merely breach) and can sue a knowing downstream purchaser. This is General Talking Pictures, and Impression Products took care to leave it standing.

2. Restrictions on a purchaser do not prevent exhaustion. A single-use restriction, a no-resale covenant, a geographic limitation, or a field limitation imposed on a buyer in an authorized sale is a contract term. Breach yields a contract claim against the buyer. It yields nothing against the buyer's transferee, and no patent claim against anyone.

3. Exhaustion is worldwide. An authorized sale anywhere exhausts U.S. rights. Regional pricing programmes cannot be defended with U.S. patents; they must be defended with contract, distribution structure, trademark, regulatory approval, warranty, and service.

4. Component sales exhaust downstream claims where the component substantially embodies the invention. If the part has no reasonable non-infringing use and embodies the essential features, its sale exhausts method and system claims practised when it is used as intended.

5. Only sales exhaust. A true lease, or a licence that does not transfer title, does not trigger exhaustion — but courts look at substance. A "lease" with a nominal end-of-term purchase price and a term matching the article's economic life is a sale.

The covenant not to sue

Practitioners often reach for a covenant not to sue when a licence feels too formal, or when a party does not want to grant rights it may not have, or when the deal is a settlement rather than a licence. It is worth knowing what the instrument actually does.

In TransCore, LP v. Electronic Transaction Consultants Corp., 563 F.3d 1271 (Fed. Cir. 2009), TransCore settled litigation with a manufacturer by giving an unconditional covenant not to sue for infringement of listed patents. The manufacturer then sold systems to a customer, and TransCore sued the customer. The Federal Circuit held the patents exhausted. "The [covenant] authorized all acts that would otherwise be infringements," and an authorized sale exhausts.

The holding is worth stating plainly: an unconditional covenant not to sue functions as a non-exclusive licence for exhaustion purposes. Calling it something else changes nothing. If you do not want the counterparty's sales to exhaust your rights against its customers, you must limit the covenant the way you would limit a licence — by field, by product, by channel — and say so.

The mirror-image case is Helferich Patent Licensing, LLC v. New York Times Co., 778 F.3d 1293 (Fed. Cir. 2015). Helferich licensed handset manufacturers under patents covering handsets, expressly reserving claims directed to content providers. It then sued content providers whose services were used with the licensed handsets. The Federal Circuit held the reservation effective: exhaustion protects the authorized acquirer of the article sold in respect of that article, and does not immunize different parties practising distinct claims that were never embodied in the sold article. The case is a genuine narrowing of exhaustion's reach, and it depends entirely on the claims being distinct and the reservation being explicit.

Worked example one: the component supplier

Marisol Ferreira is general counsel at Kestrel Silicon, which designs a controller chip. Northlight Instruments holds patents on a calibration method that Kestrel's chip performs, and on a system claim covering an instrument containing such a chip. Northlight offers a licence.

Northlight's first draft grants Kestrel the right to "make, have made, use, offer for sale, sell, and import Licensed Products," defines Licensed Products as chips implementing the calibration method, and includes a covenant that Kestrel will inform its customers that no licence is granted under the system claims.

Marisol recognizes Quanta immediately. The chip has no reasonable use except to be placed in an instrument and run the method. The grant authorizes Kestrel to sell it. The notice covenant is not a limitation on the grant. Northlight's system claims will be exhausted against every one of Kestrel's customers.

She could say nothing, sign, and let Northlight discover the problem later. She does not, for two reasons: the resulting licence is worth more to Kestrel if Northlight understands what it is selling, and a licensor that later feels defrauded is a licensor that litigates.

She proposes the structure that actually works. Northlight grants Kestrel a field-limited licence: to make and sell chips for incorporation into instruments in the industrial process control field only, with an express statement that no licence is granted for chips destined for the clinical diagnostics field. Kestrel is authorized to make have-made arrangements with its foundry. The royalty is a per-chip amount, tiered by volume. Because the field limitation is on the grant, a Kestrel sale into clinical diagnostics is unauthorized, does not exhaust, and gives Northlight both a breach claim against Kestrel and an infringement claim against a knowing buyer.

For the industrial field, Northlight knows its system claims are exhausted against Kestrel's customers, and it prices accordingly — the royalty reflects the value of the whole authorized chain, not just the chip. Northlight also retains the right to license the clinical field separately, at a different rate.

Marisol adds two provisions. First, a compliance mechanism: Kestrel's standard terms of sale will state the field limitation, Kestrel will not knowingly sell to a buyer that has told it the chips are for clinical use, and Kestrel's obligation is defined by knowledge rather than by outcome — an important distinction, since Kestrel cannot police what a distributor's customer's customer does. Second, an audit and true-up: if chips sold as industrial are found in clinical instruments in commercially meaningful volume, the parties will negotiate in good faith, and absent agreement Kestrel will pay the clinical rate on the excess.

That is a workable deal. What made it workable was moving the limitation from a covenant to the grant.

Worked example two: the cartridge programme

Devon Achebe runs the aftermarket strategy at Halyard Imaging, which sells printers below cost and makes its margin on cartridges. Halyard has historically sold two cartridge lines: a full-price cartridge and a discounted cartridge sold under terms restricting the buyer to a single use and requiring return to Halyard.

After Impression Products, the patent enforcement piece of the programme is gone. Halyard cannot sue a refiller for patent infringement over a cartridge Halyard sold, and it cannot use its U.S. patents against cartridges it sold in Chile and somebody imported.

Devon's team maps what remains.

Contract. The single-use restriction is enforceable against a buyer who agreed to it, in contract, with contract remedies. Against an enterprise customer that signed a supply agreement, this is real. Against a consumer who clicked through a box-top notice, it is worth less than the cost of enforcing it, and pursuing individual consumers is commercially self-destructive.

Design. Cartridges can be engineered so that refurbishment is difficult, though the line between legitimate design and unlawful tying or unfair competition needs watching, and consumer-protection and right-to-repair regimes are moving against aggressive designs.

Repair versus reconstruction. Exhaustion permits an owner to repair a patented article but not to reconstruct it — to make a new one. The boundary is fact-intensive and the Supreme Court's cases place it generously toward repair. Replacing a consumable component is repair. Rebuilding the entire article from a spent shell may be reconstruction. Halyard's counsel concludes the boundary will not carry the programme.

Trademark. Refilled cartridges sold as Halyard cartridges without adequate disclosure can raise trademark issues; refilled cartridges sold as remanufactured, with clear disclosure, generally cannot.

Pricing. Halyard can offer the discount through a rebate conditioned on return, or through a subscription that never transfers title. A genuine service model — where Halyard retains ownership, supplies cartridges as part of a managed print service, and takes them back — does not involve a sale, and so does not exhaust. This is the structural answer, and it is what large parts of the industry moved to.

Devon's conclusion is the right one: Impression Products did not destroy the business model; it destroyed one enforcement tool and forced the model to be built on ownership structure and contract instead of on the patent.

Worked example three: single-use medical devices

Dr. Priya Raghunathan is chief medical officer at Anvil Surgical, which sells an electrosurgical instrument labelled for single use. A reprocessing company collects used instruments from hospitals, sterilizes and refurbishes them, and sells them back at a discount.

Before 2017, Anvil would have sued for patent infringement, relying on the single-use restriction. That route is closed.

What Anvil has instead is a combination that is, in this sector, quite strong. Reprocessing of single-use devices is regulated: a reprocessor is a manufacturer for regulatory purposes and must meet the applicable premarket requirements for the device type. Anvil's regulatory affairs team can and should engage with that framework. Anvil's clinical evidence about performance after reuse is relevant to hospital purchasing decisions and to informed consent. Anvil's warranty can, entirely lawfully, exclude devices that have been reprocessed by third parties. Anvil's contracts with hospital systems can address reuse directly.

What Anvil cannot do is treat the reprocessor as a patent infringer for restoring an article Anvil sold. And Anvil's counsel must be careful about a related risk: a campaign of communications to hospitals asserting patent infringement that the law does not support carries its own exposure, both to unfair competition claims and, where the assertions are objectively baseless and subjectively motivated by interference with a competitor's business, to antitrust exposure.

Drafting the grant

Everything above converges on the grant clause. Here is what to attend to.

Enumerate the rights. "Make, have made, use, sell, offer for sale, and import" are separate rights under § 271. A licence that says "use" does not permit manufacture. A licence that omits "have made" may not permit contract manufacturing — and in an industry that uses foundries and contract manufacturers, that omission can make the licence worthless.

Define the field on the grant, not in a covenant. "Licensor grants Licensee a licence to make and sell Licensed Products solely for use in the Field, and no licence is granted outside the Field" is a limitation. "Licensee shall not sell Licensed Products for use outside the Field" is a covenant. The first prevents exhaustion; the second does not.

Define the field precisely. Field definitions litigate constantly. "Automotive applications" is a lawsuit. Define by end product, by customer type, by technical specification, or by regulatory classification — something a court can apply.

Address sublicensing expressly. Silence is unpredictable. A non-exclusive licence generally does not carry a right to sublicense. State whether sublicensing is permitted, whether consent is required, whether sublicensees must accept the field limitation, whether the licensee must report and pay on sublicensee revenue, and what happens to sublicences when the head licence terminates. That last point is the one that gets forgotten and the one that matters most when the relationship fails.

Address assignment and change of control. Patent licences are presumptively non-assignable under federal common law, and that presumption is stronger than most parties expect. If assignment in a merger is intended, say so.

Address territory realistically. A territorial limitation on a licensee's authority works. A territorial restriction on a purchaser does not preserve U.S. patent rights after Impression Products.

Consider what "authorized sale" means for affiliates. If the licence extends to affiliates, an affiliate's sale is authorized and exhausts. Define affiliate, and define what happens when an entity ceases to be one.

Royalties, term, and the Brulotte problem

Brulotte v. Thys Co., 379 U.S. 29 (1964), held that a patentee may not collect royalties for use of an invention after the patent expires; such an agreement is unenforceable per se. The rule was widely criticized, and in Kimble v. Marvel Entertainment, LLC, 576 U.S. 446 (2015), the Court reaffirmed it on stare decisis grounds while going out of its way to describe the workarounds.

The workarounds are the practical content of the doctrine:

  • Deferred payment. Royalties accruing during the patent term may be paid in instalments after expiration. What is prohibited is charging for post-expiration use, not spreading payment for pre-expiration use.
  • Hybrid licences. A licence covering patents and trade secrets or know-how may provide for a reduced rate after patent expiration, provided the post-expiration payment is genuinely attributable to the non-patent rights. A single undifferentiated rate that continues unchanged is the trap.
  • Multiple patents. A royalty running until the last-to-expire licensed patent is fine.
  • Non-patent consideration. Payments tied to a trademark licence, a supply arrangement, or ongoing services survive expiration.
  • Step-downs. A rate that declines as patents expire is safe and is usually what the parties would have negotiated anyway.

Draft with an expiration schedule in the licence, and draft the rate to change on stated dates rather than by reference to an undefined event.

Validity challenges and the licensee

Lear, Inc. v. Adkins, 395 U.S. 653 (1969), abolished licensee estoppel: a licensee may challenge the validity of the licensed patent notwithstanding the licence. The public interest in eliminating invalid patents outweighs the contract-law instinct that a party should not attack the consideration it accepted.

MedImmune, Inc. v. Genentech, Inc., 549 U.S. 118 (2007), went further: a licensee in good standing, paying royalties under protest, has standing to bring a declaratory judgment action challenging validity without first breaching. Before MedImmune, a licensee had to stop paying and invite suit; after it, the licensee can litigate from safety.

Licensors responded with drafting. The tools that have held up:

  • Termination on challenge. The licensor may terminate if the licensee challenges validity. Generally enforceable.
  • Royalty escalation on challenge. A higher rate applies from the filing of a challenge. Generally enforceable; a rate high enough to be a penalty is not.
  • Fee-shifting. The challenger pays the licensor's costs if the challenge fails. Enforceable in a negotiated commercial agreement.
  • No-challenge covenants. Enforceable in a settlement of actual litigation where validity was genuinely contested, and generally not enforceable in a bare licence.

Lear also bears on royalties already paid: a licensee that successfully invalidates cannot ordinarily recover royalties paid before the challenge, so the escalation and termination provisions do real work.

Misuse and antitrust at the edges

Patent misuse is an equitable defence that renders a patent unenforceable until the misuse is purged. It requires an impermissible broadening of the patent's scope with anticompetitive effect. Its practical footprint has narrowed considerably.

Illinois Tool Works Inc. v. Independent Ink, Inc., 547 U.S. 28 (2006), eliminated the presumption that a patent confers market power for tying analysis, requiring proof of power in the relevant market. Princo Corp. v. International Trade Commission, 616 F.3d 1318 (Fed. Cir. 2010) (en banc), required a nexus between the alleged misconduct and the patent's leverage, holding that alleged suppression of a competing technology by pool members did not constitute misuse of the asserted patents.

Statutory limits also matter. Section 271(d) provides that a patentee does not commit misuse merely by refusing to license, by seeking to enforce, or by conditioning a licence on the acquisition of a separate licence unless the patentee has market power in the tying product.

Antitrust exposure in licensing has instead concentrated around specific structures: horizontal agreements dressed as licences, reverse-payment settlements after FTC v. Actavis, Inc., 570 U.S. 136 (2013), pools that eliminate competition between substitutes, and standard-setting conduct including FRAND commitments. The general shape of the modern rule is that a patent licence is analysed like any other contract: legitimate exploitation of a lawful exclusive right is fine; using the licence as a vehicle for agreement among competitors is not.

Zenith Radio Corp. v. Hazeltine Research, Inc., 395 U.S. 100 (1969), retains practical importance on royalty base: conditioning a licence on payment of royalties on unpatented products is misuse when it is coerced, though a total-sales base that the licensee genuinely prefers for convenience is permissible.

Bowman and the self-replicating problem

Bowman v. Monsanto Co., 569 U.S. 278 (2013), addressed what happens when the patented article makes copies of itself. A farmer bought commodity soybeans, planted them, and harvested a crop containing the patented trait; he argued exhaustion.

The Court disagreed unanimously. Exhaustion permits the purchaser to use or resell the article purchased; it does not permit the purchaser to make new articles. Planting and harvesting made new infringing seeds, and making is a distinct right under § 271. The Court confined its holding carefully, noting that it did not address every self-replicating technology, and that circumstances in which replication is a necessary and incidental consequence of an otherwise lawful use might be treated differently.

The principle generalizes: exhaustion covers the article, not the right to manufacture more of them. It matters for biological materials, cell lines, software distribution keyed to instantiation, and any technology where use produces copies.

Lease, licence, or sale: the characterization that decides everything

Because only a sale exhausts, the question of whether a transaction is a sale becomes the hinge on which a whole business model can turn. Parties who understand this restructure toward retained ownership; parties who do not paper a sale and call it something else.

Courts look at substance, and the factors are familiar from the tax and secured-transactions worlds:

  • Does title transfer? Not what the document says title does — what happens in fact.
  • Is there a single up-front payment approximating the article's value? A one-time payment equal to the purchase price is the strongest indicator of a sale, whatever the label.
  • What is the term relative to the article's economic life? A "lease" for the full useful life is a sale.
  • Is there a nominal purchase option? A $1 option at the end of the term is a sale.
  • Who bears risk of loss, maintenance, insurance, and obsolescence? A transferee who bears all of these has the incidents of ownership.
  • Must the article be returned, and is return actually enforced? A return obligation that nobody polices is decoration. Courts notice.
  • Is the transferor's continuing role real? A managed service in which the provider monitors, maintains, replaces, and eventually reclaims the equipment is a genuine service. A "service" that consists of shipping a box and sending an invoice is a sale.

The point is not that retained-ownership structures are a trick. They are a real business choice with real consequences: the owner keeps the residual value and the disposal obligation, carries the asset on its balance sheet, bears obsolescence risk, and takes on service obligations it may not want. Firms that adopt equipment-as-a-service models for exhaustion reasons alone usually discover the other consequences within a year. Firms that adopt them because the service model is genuinely better — predictable revenue, tighter customer relationships, control over the installed base — get the exhaustion benefit as a bonus.

One caution. A structure that is a sale in substance will not be saved by the label, and a party that has told its auditors, its tax advisers, and its lenders that a transaction is a sale will find those statements quoted back to it. Consistency across the whole record is what makes a characterization hold.

Have-made rights and the foundry problem

The phrase "have made" occupies four words in a grant clause and can decide whether a licence is usable at all.

A licence to "make" authorizes the licensee to manufacture. It does not clearly authorize the licensee to engage a third party to manufacture on its behalf. In industries that depend on contract manufacturing — semiconductors above all, but also pharmaceuticals, medical devices, and most consumer hardware — a licence without have-made rights may be commercially empty.

Licensors resist have-made rights for a legitimate reason: an unbounded have-made right can become a back-door sublicence. A contract manufacturer with have-made authority might make products for the licensee and, using the same authorization, make products for others.

The provisions that resolve the tension:

  • Have-made rights limited to products made to the licensee's specifications, for the licensee's account, and sold by or for the licensee. This is the standard formulation and it addresses the licensor's real concern.
  • A requirement that the contract manufacturer be bound in writing to use the licensed technology only for the licensee's products.
  • A reporting obligation identifying contract manufacturers, so the licensor can see who is making licensed product.
  • An express statement that have-made rights are not sublicence rights and confer nothing on the manufacturer beyond what it does for the licensee.

The related question is whether the manufacturer's own sale to the licensee exhausts. It generally does not create a problem: the manufacturer is making for the licensee within the licensee's authority, and the article's journey to a customer is a single authorized chain. Where it does create a problem is when the manufacturer overproduces and sells the surplus — the "grey channel" issue that plagues consumer electronics. Those sales are outside the have-made authorization, are unauthorized, and do not exhaust. Proving which units came from where is the practical difficulty, which is why serialization and reporting obligations earn their keep.

Exhaustion in the settlement context

Settlements of patent litigation almost always include a release, a covenant not to sue, or a licence — and the choice among them has exhaustion consequences that settling parties routinely overlook because they are focused on ending the case.

A release covers past conduct. It does not authorize future conduct and does not, by itself, exhaust anything going forward. It also does not clear articles already in the field: a release of past infringement damages does not make articles already sold "authorized" for exhaustion purposes, though as a practical matter a patentee that has released the manufacturer will find suing its customers unattractive and sometimes barred by the settlement's own terms.

A covenant not to sue for future conduct is, per TransCore, functionally a licence. Draft it as a licence: define the products, the field, the territory, and the term with the same care.

A licence granted in settlement should say what happens on termination, what happens to affiliates, and whether it runs to successors. Settlement licences are often drafted in a hurry against a trial date and inherit none of the structure a negotiated licence would have.

One further trap. A settlement covenant given to a defendant frequently extends to the defendant's customers "with respect to the accused products." Read that phrase carefully. If the accused products are components, a customer covenant may exhaust system claims across an entire industry. If the patentee intends to preserve claims against downstream implementers, it must say so with the specificity Helferich required — distinct claims, expressly reserved, directed to different actors.

Marking, notice, and the damages you can actually collect

A licensing programme that gets exhaustion right can still lose most of its damages to 35 U.S.C. § 287. The marking statute provides that a patentee who makes or sells a patented article, or licenses another to do so, may recover damages for infringement only from the date the infringer had notice — and constructive notice is given by marking the article with the patent number or a web address under the virtual marking provision.

The trap for licensors is that a licensee's failure to mark is attributed to the licensor. If a licensee sells unmarked patented articles, the patentee's damages against a wholly unrelated infringer may be limited to the period after actual notice. Courts apply a reasonableness standard to the licensor's efforts, but "we told them to mark" is thin if nobody checked.

Three provisions follow:

  • An express marking obligation in every licence covering an apparatus claim, with the marking format specified.
  • A verification right, exercised at least annually, with samples or photographs.
  • A cure and indemnity: if the licensee fails to mark and the licensor loses damages as a result, the licensee bears the loss.

Marking does not apply where only method claims are asserted, since there is no article to mark — a distinction that occasionally changes which claims a patentee chooses to assert.

Damages themselves are governed by 35 U.S.C. § 284, which sets a floor of a reasonable royalty. A licensing programme generates the evidence that will later set that royalty. Every licence signed becomes a comparable, and a portfolio of low-rate settlement licences signed to avoid litigation costs will be produced by the next defendant as evidence of the patent's value. Licensors who understand this distinguish carefully in the documents between litigation settlements and arm's-length commercial licences, and they think about the rate they are establishing, not just the cheque they are receiving.

Assignment, standing, and who may sue

35 U.S.C. § 261 makes patents assignable by written instrument and provides for recordation. The distinction that matters for licensing is between an assignment and a licence, because it determines who can sue.

A patent owner may sue alone. An exclusive licensee with all substantial rights is treated as the effective owner and may sue alone. An exclusive licensee with less than all substantial rights may sue only by joining the patentee. A non-exclusive licensee has no standing to sue at all, ever, regardless of what the licence says.

Whether a licensee holds "all substantial rights" turns on the totality: the scope of the exclusive grant, the right to sublicense, the right to sue and to control litigation, the right to settle, the patentee's retained rights, reversion provisions, and the patentee's retained economic interest. A licence labelled "exclusive" that reserves the right to practise, requires consent to sublicense, and gives the patentee control over litigation does not convey all substantial rights.

The practical consequences are worth stating for both sides:

For a licensee that intends to enforce, the licence must grant exclusivity in a defined field, the right to sue in its own name, control over the litigation, the right to settle, and a covenant that the patentee will join as a party if joinder is necessary — with a mechanism for the licensee to bear the cost.

For a licensor that does not want to be dragged into its licensee's disputes, the reverse: retain the right of first refusal on enforcement, retain settlement approval where a settlement would grant rights beyond the licensee's field, and require indemnity for the cost of involuntary joinder.

Standing defects are frequently discovered in the middle of litigation, and they are expensive. A licence drafted with the enforcement scenario in mind avoids a category of motion practice that has nothing to do with the merits.

A short taxonomy of common errors

Reviewing licensing programmes, the same errors recur:

The covenant that should have been a limitation. Restrictions written as promises rather than as boundaries on the grant. The single most common and most consequential drafting failure in the field.

The undefined field. "Consumer applications." "Non-medical uses." "The automotive market." Each one is a future dispute.

The missing have-made right. A licence the licensee cannot actually use because it does not own a factory.

Silence on sublicensing. Neither granted nor prohibited, which means litigated.

Silence on sublicence survival. The head licence terminates; nobody knows what happens to the twelve sublicences beneath it.

The single royalty rate running past expiration. A Brulotte problem that a step-down schedule would have avoided.

The unmarked licensee. Damages quietly lost across an entire programme.

The "exclusive" licence that is not exclusive enough to sue. Discovered on a motion to dismiss.

The territorial restriction on purchasers. Post-Impression Products, decorative.

The settlement covenant that extends to customers. Exhaustion across a supply chain, granted without anyone intending it.

Each of these is cheap to fix at drafting and expensive to fix afterward. That asymmetry is the whole argument for spending an extra week on a grant clause.

Related documents