Picture this. Your encryption startup—call it Sekure Kart—is finally humming. You undercut the big payment processors by twenty percent, you are signing new customers every day, and you cannot hire fast enough. Then your CFO knocks, looking pale, and slides a letter across your desk. It is not from a so-called patent troll. It is from one of the giant processors you have been quietly stealing customers from, and it says, in the polite menace that lawyers do so well, that your product infringes their patent and you should stop or prepare to be sued.

Your stomach drops. Patent litigation is famously expensive and famously unpredictable—a single case routinely runs into the millions before a jury ever hears a word. But before you panic, and certainly before you write a check, you need to understand what patent infringement actually is. Because here is the reassuring secret at the heart of this whole intimidating field: infringement is, fundamentally, a checklist exercise. A patent's claims are a numbered list of requirements. If your product contains every single requirement on that list, you infringe. If it is missing even one, you do not. That is most of the game.

This article is the friendly, plain-English tour of that game. We will explain what infringement means under the governing statute, 35 U.S.C. § 271; how courts figure out what a patent claim actually covers (a process with the slightly grand name "claim construction"); the difference between copying a claim word-for-word and infringing something close enough to count (the "doctrine of equivalents"); the special problem of inventions whose steps are split across several actors; when infringement becomes bad enough to triple the damages; and the defenses an accused infringer can raise to make the whole problem go away. By the end you will be able to read a cease-and-desist letter like the one on Sekure Kart's desk and have an intelligent first reaction instead of a panic attack.

If you want the deep, courtroom-by-courtroom mechanics of how an infringement lawsuit actually unfolds—pleadings, discovery, summary judgment, trial, appeal, and post-trial motions—this piece is the on-ramp, and our comprehensive guide to patent infringement litigation is the highway. For the broader "what is a patent and how do you get one" picture, see our general information concerning patents and utility patent basics. And if you are launching a product and want to avoid ever receiving Sekure Kart's letter in the first place, the disciplined front-end practice is a clearance review, covered in our guide to conducting a freedom-to-operate analysis for new products. This article assumes you already know roughly what a patent is and zeroes in on the single question: what does it mean to infringe one?

The One Idea You Need First: A Patent Is a Right to Exclude, Defined by Its Claims

Before we touch infringement, fix one concept firmly in mind, because everything else hangs from it. A U.S. patent does not give its owner the right to make anything. It gives the owner the right to stop other people—the legal phrase is the "right to exclude." A patent is, in essence, a government-backed "keep off the grass" sign. And like any property line, it has a precise boundary. That boundary is set by the patent's claims.

If you flip to the end of any patent, past the diagrams and the long technical description, you will find a numbered list of single (often very long) sentences. Those are the claims. Each claim is a self-contained definition of one version of the invention, written as a list of elements (also called "limitations"—the words are interchangeable). The rest of the patent—the part lawyers call the specification, which includes the drawings and the written description—is the inventor's promise to teach the public how to make and use the invention. In exchange for that teaching, the public hands the inventor a time-limited monopoly. That is the basic bargain of patent law, rooted in the Constitution's grant to Congress of power to "promote the Progress of Science and useful Arts" by securing exclusive rights to inventors (U.S. Const. art. I, § 8, cl. 8).

Here is why the claims matter so much for infringement. The claims, and only the claims, define what you can be sued for. The dazzling invention described in the specification is irrelevant to infringement except insofar as it helps explain what the claims mean. As patent lawyers like to say, "the name of the game is the claim." A useful mental model: think of each claim as a recipe. To infringe, your dish has to contain every ingredient in the recipe. Leave out the saffron, and—however delicious your paella—you have not made that recipe.

This is the principle the patent bar calls the all-elements rule (or the all-limitations rule), and it is the single most important rule in this whole article. A patent claim is infringed only if the accused product or process contains each and every element of the claim. Miss one, and there is no infringement of that claim. We will come back to this rule constantly.

One more piece of vocabulary worth installing now, because it does quiet but decisive work in litigation: the difference between independent and dependent claims. An independent claim stands on its own. A dependent claim refers back to another claim and adds something ("The container of claim 1, wherein the strip is blue"). Because a dependent claim contains everything in the claim it depends on plus the extra limitation, it is necessarily narrower—and so harder to infringe—than its parent. This is why a sophisticated infringement analysis always starts with the independent claims: if the accused product clears those, it clears every claim that depends from them too.

The Governing Statute: 35 U.S.C. § 271

Infringement is a creature of statute. The operative section is 35 U.S.C. § 271, and its first three subsections describe the three flavors of infringement you will hear about again and again. (Several later subsections—(e), (f), and (g)—cover important specialty situations, and we will get to those once the main framework is in place.)

§ 271(a): Direct Infringement—The Strict-Liability Heart of It

Section 271(a) is the core. It provides that "whoever without authority makes, uses, offers to sell, or sells any patented invention, within the United States or imports into the United States any patented invention during the term of the patent therefor, infringes the patent."

Read that list of verbs slowly, because each is a separate trigger: making, using, selling, offering to sell, or importing. Do any one of them, within the United States, without permission, to something covered by the claims, and you are a direct infringer. You do not have to do all five. You do not even have to sell anything—merely using a patented machine in your own factory can infringe. (And note the territorial limit baked into the verbs: the prohibited act generally has to happen in the United States. A device made and used entirely in another country does not infringe a U.S. patent—though, as we will see, importing the product of a patented process can.)

Now here is the part that surprises almost everyone, including sophisticated businesspeople: direct infringement is a strict-liability offense. Intent is irrelevant. You do not have to know the patent exists. You do not have to have copied anything. You can independently invent the very same thing in your garage, in good faith, having never heard of the patentee—and you still infringe if your product reads on the claims. Patent law, unlike copyright law, has no "independent creation" defense to direct infringement. The accused infringer need only have performed one of the prohibited acts on something covered by the patent.

Let's ground this with an invented example. Suppose Acme Corp. holds a patent whose Claim 1 covers "a beverage container comprising (a) a cylindrical body, (b) a removable lid, and (c) a temperature-indicating strip affixed to the exterior of the body that changes color above 140°F." If our friends at Beta LLC sell a travel mug that has a cylindrical body, a removable lid, and a heat-sensitive color strip on the outside, Beta directly infringes Claim 1—even if Beta's engineers had never seen Acme's patent and designed the mug from scratch. Why? Because Beta's mug contains every element (a), (b), and (c). That is the all-elements rule in action.

For more on what a "patented invention" even is and how claims come to exist, see our pieces on utility patent basics and the gentler patent basics—a plain-English guide.

§ 271(b): Induced Infringement—Helping Someone Else Do It

Not everyone who profits from infringement is the one physically making or using the patented thing. Enter induced infringement under § 271(b), which makes liable "whoever actively induces infringement of a patent."

Inducement is about encouraging, persuading, or instructing someone else to infringe. Classic example: a company sells a perfectly legal kit of parts but publishes an instruction manual telling customers to assemble those parts into a patented device. The company never makes the infringing device itself; its customers do. But by actively encouraging that infringement, the company can be on the hook. This is a favorite theory against suppliers, distributors, and software vendors whose products do nothing wrong on the shelf but become infringing the moment the user follows the directions.

Two things make inducement importantly different from direct infringement. First, there must actually be an underlying act of direct infringement by someone—you cannot induce an infringement that never happens. The Supreme Court drove this home in Limelight Networks, Inc. v. Akamai Technologies, Inc., 572 U.S. 915 (2014), holding there can be no induced infringement without a predicate act of direct infringement. Second—and this is the crucial difference from strict-liability direct infringement—inducement requires knowledge and intent. The accused inducer must know of the patent and must knowingly intend to bring about the infringing acts. The Supreme Court settled the knowledge standard in Global-Tech Appliances, Inc. v. SEB S.A., 563 U.S. 754 (2011), holding that inducement requires knowledge that the induced acts constitute patent infringement—and that this knowledge can be satisfied by "willful blindness," meaning the defendant subjectively believed there was a high probability the patent existed and deliberately avoided confirming it. Mere negligence or recklessness is not enough. (Global-Tech is also a vivid cautionary tale: the defendant had a competitor's product copied overseas and obtained a freedom-to-operate opinion without telling its lawyer the design was a knockoff—precisely the kind of deliberate avoidance the Court found damning.)

One more wrinkle worth knowing: a defendant's good-faith belief that the patent is invalid is not a defense to inducement. In Commil USA, LLC v. Cisco Systems, Inc., 575 U.S. 632 (2015), the Court held that infringement and validity are distinct issues, and believing a patent is invalid does not negate the knowledge of infringement that inducement requires. (The right move for a defendant who thinks the patent is junk is to attack its validity—in court or at the Patent Office—not to keep selling on a private hunch.)

§ 271(c): Contributory Infringement—Selling the Special Part

The third flavor, contributory infringement under § 271(c), targets a narrower act: selling (or importing) a component of a patented invention—or a material or apparatus for practicing a patented process—knowing it was especially made or adapted for infringing use, and where that component is not a staple article of commerce suitable for substantial non-infringing use.

That last clause is the key to the whole doctrine. Contributory infringement only reaches components that are essentially good for nothing but infringing. If the part you sell has a real, substantial, legitimate non-infringing use, you are off the hook for contributory infringement no matter how many infringers happen to buy it. The classic illustration: selling ordinary screws cannot be contributory infringement even if someone uses them to build a patented machine, because screws have countless lawful uses. But selling a uniquely shaped widget that fits only into a patented apparatus and serves no other purpose can be.

Like inducement, contributory infringement carries a knowledge requirement: the seller must know the component was made or adapted for use in infringing the patent. And, like inducement, it requires an underlying direct infringement by someone—the Supreme Court made that prerequisite clear long ago in Aro Mfg. Co. v. Convertible Top Replacement Co., 365 U.S. 336 (1961). (Aro is also the origin of the durable—and frequently litigated—line between permissible repair of a patented article you legitimately own and impermissible reconstruction of a new one, a distinction that still decides cases about replacement parts and refurbished goods.)

So, to keep the three straight:

  • Direct infringement (§ 271(a)) — you make, use, sell, offer, or import the claimed invention. No intent required. Strict liability.
  • Induced infringement (§ 271(b)) — you knowingly encourage someone else to infringe. Requires knowledge and intent, plus an underlying direct infringement.
  • Contributory infringement (§ 271(c)) — you sell a special-purpose component with no substantial non-infringing use, knowing it is for infringement. Requires knowledge, plus an underlying direct infringement.

Both inducement and contributory infringement are species of "indirect" infringement—they make liable someone other than the direct infringer—and neither can exist without a direct infringer somewhere in the picture. That dependency is not a technicality; as the next section shows, it is the pressure point on which an entire category of modern patent disputes turns.

Divided Infringement: When No Single Actor Does It All

Here is a problem the drafters of § 271 never fully anticipated and the courts have wrestled with for two decades. The all-elements rule says infringement requires every step (or element) to be present. But what if a method claim's steps are spread across different actors—say, one step performed by a company's server and another by its customers? If no single entity performs all the steps, is anyone a direct infringer? And if there is no direct infringer, then—because indirect infringement needs a predicate direct infringement—no one can be liable for inducement either. The whole edifice can collapse on this gap. This is the problem of divided (or joint) infringement, and it is the bread and butter of patent disputes over software, telecommunications, networking, and the internet, where claimed methods are naturally split between a provider and its users.

For system and apparatus claims, the Federal Circuit took some of the sting out of the problem in Centillion Data Systems, LLC v. Qwest Communications International, Inc., 631 F.3d 1279 (Fed. Cir. 2011), which held that to "use" a claimed system, a party need not physically control every element; it need only "put the invention into service"—that is, control the system as a whole and obtain benefit from it. A customer who runs a service that pulls together a back-end and a front-end can "use" the whole system even though the provider owns half the hardware.

Method claims proved thornier. After years of doctrinal whiplash, the Federal Circuit settled the standard in its en banc decision in Akamai Technologies, Inc. v. Limelight Networks, Inc., 797 F.3d 1020 (Fed. Cir. 2015) (en banc) (the sequel to the Supreme Court's 2014 Limelight decision). The court held that the acts of multiple actors can be attributed to a single accused infringer—making that one party a direct infringer of a method claim—in two situations: (1) where the accused party directs or controls the others' performance, and (2) where the actors form a joint enterprise. Critically, Akamai expanded what "directs or controls" means: it now reaches a defendant that conditions participation in an activity or receipt of a benefit on performance of the claimed steps and establishes the manner or timing of that performance. In Akamai itself, Limelight required its customers to perform the final "tagging" step in order to use Limelight's content-delivery service and gave them step-by-step instructions for doing so—enough, the en banc court held, to attribute the customers' step to Limelight.

What does this mean on the ground? For a patentee, divided infringement is a hazard to be drafted around. The savvy move is to write method claims so that every step is performed by a single actor (typically the actor you most want to sue), avoiding claims that require cooperation between, say, a server operator and an end user. For an accused infringer, by contrast, divided infringement remains a genuine non-infringement defense: if the steps really are split among independent parties—no direction, no control, no joint enterprise, no conditioning-plus-manner—then there is no single direct infringer, and the inducement claim that depends on it fails with it. It is one of the more elegant ways to win a software patent case without ever conceding that the technology is similar.

The Two-Step Analysis: First Read the Claims, Then Compare

Every infringement question, no matter how complex the technology, resolves into a two-step framework that the Federal Circuit has applied for decades. (The Federal Circuit, by the way, is the specialized appeals court in Washington, D.C. that hears all patent appeals; under 28 U.S.C. § 1338, patent cases start exclusively in federal district court and funnel up to the Federal Circuit, not the regional circuits.)

Step one: construe the claims. Figure out what the words of the claim mean—what they cover and what they exclude. This is "claim construction."

Step two: compare. Apply the now-construed claim to the accused product or process and ask whether every element is present, either literally or by equivalents.

Step one is a question of law for the judge. Step two is a question of fact, usually for the jury (or the judge in a bench trial). Keep that division in mind; it explains a lot about how patent cases are litigated, which our comprehensive litigation guide explores in depth.

Step One: Claim Construction (the "Markman" Process)

Claim construction is where patent cases are very often won and lost. It sounds dry, but think about how much turns on a single word. If Acme's claim says the temperature strip changes color "above 140°F," does a strip calibrated to 138°F infringe? It depends entirely on how the court construes that limitation. Battles over the meaning of words like "comprising," "substantially," "attached," "a," and "or" have decided multimillion-dollar cases. More than one defendant has escaped liability—and more than one patentee has lost a fortune—on the difference between "a" meaning "one" and "a" meaning "one or more."

The procedure has a nickname: the Markman hearing, after the Supreme Court's decision in Markman v. Westview Instruments, Inc., 517 U.S. 370 (1996). Markman established that claim construction is a job for the judge, not the jury. The judge holds a hearing, reviews briefing from both sides about what the disputed terms mean, sometimes hears expert testimony, and then issues a claim construction order (a "Markman order") declaring the meaning of each disputed term. That construction becomes part of the jury instructions; the jury then takes the judge's definitions as given and decides the factual question of whether the accused product meets them. So decisive is this ruling that a great many cases settle within days of the Markman order—once the parties know what the words mean, they usually know who is going to win.

How does a judge decide what a claim word means? The roadmap comes from the Federal Circuit's landmark en banc decision in Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005). The governing principles:

  • Ordinary meaning to a person of ordinary skill. Claim terms are generally given the ordinary and customary meaning they would have to a person of ordinary skill in the relevant art at the time of the invention. (That hypothetical "person of ordinary skill in the art"—often abbreviated POSITA—is patent law's reasonable person: a skilled but unimaginative practitioner in the field.)
  • The intrinsic record comes first. The most reliable evidence is the patent's own "intrinsic record": the claims themselves, the specification, and the prosecution history (the back-and-forth between the applicant and the U.S. Patent and Trademark Office while the patent was being examined). The specification, Phillips said, is usually the single best guide—"the best source for understanding a technical term is the specification from which it arose."
  • The patentee can be its own lexicographer. If the inventor expressly defined a term in the specification, that definition controls, even if it departs from ordinary usage.
  • Extrinsic evidence, handled with care. Dictionaries, treatises, and expert testimony can help, but courts treat them cautiously and never let them override the intrinsic record. Phillips specifically warned against letting abstract dictionary definitions hijack the analysis.

One often-overlooked layer governs appellate review of all this. In Teva Pharmaceuticals USA, Inc. v. Sandoz, Inc., 574 U.S. 318 (2015), the Supreme Court held that while the ultimate construction is still reviewed de novo, any subsidiary factual findings a district judge makes about extrinsic evidence (for example, what a skilled artisan understood a term to mean) are reviewed only for clear error. In practice, that gives a careful trial judge meaningful insulation when the construction rests on disputed technical facts—and it gives litigants a reason to build a real evidentiary record at the Markman stage rather than treating it as pure legal argument.

There is also a powerful doctrine lurking in the prosecution history called prosecution history estoppel (or the "disclaimer" rule). If, during examination, the applicant narrowed a claim or surrendered a particular interpretation to get the patent allowed over prior art, the patentee cannot later recapture that surrendered scope in litigation. It would be unfair to tell the Patent Office "my invention does not cover X" to win the patent, then turn around and sue someone whose product is X. We will see this doctrine again when we get to the doctrine of equivalents, where it does its most important work.

If you want to understand how examiners and applicants generate that prosecution history in the first place—and how the give-and-take can come back to haunt a patentee—see our companion pieces on responding to patent office actions and overcoming obviousness rejections.

Step Two, Part A: Literal Infringement

Once the judge has told everyone what the claim means, the comparison begins. The first and cleanest way to infringe is literal infringement: the accused product or process contains, word-for-word, each and every element of the claim as construed.

Back to chairs, the patent lawyer's favorite teaching prop. Suppose a claim covers "an apparatus for sitting, comprising four legs attached to a seating member, and a curved back member attached to said seating member." To literally infringe, an accused chair must have (1) four legs attached to a seat, and (2) a curved back attached to that seat. A chair with four legs and a curved back? Literal infringement. A three-legged stool with no back? Not even close—it is missing two elements. A four-legged chair with a perfectly straight back? No literal infringement, because the claim requires a curved back and the accused chair does not have one. Remember the all-elements rule: missing one element means no literal infringement, full stop.

A quick but important note on the word "comprising." In patent-claim grammar, "comprising" is an open-ended, magic word meaning "including at least the following, but possibly more." So a claim to a device "comprising A, B, and C" is infringed by a product that has A, B, C and also D, E, and F. Adding extra features does not let you escape infringement. (Contrast "consisting of," a closed phrase meaning "only the following"—but most claims use "comprising," so adding bells and whistles is generally no defense.) This trips up a lot of would-be designers-around: you cannot dodge a claim merely by piling on extra components. The flip side, though, is just as important and far more useful to defendants: because "comprising" only adds permission to include more, it never excuses omitting a required element. If your product leaves out limitation C entirely, all the open-ended drafting in the world will not save the patentee.

Step Two, Part B: The Doctrine of Equivalents

Now for the subtle part, the doctrine that keeps patent litigators employed. What happens when an accused product is almost but not quite a literal match—it changes one element just enough to avoid the exact claim language, but the change is trivial? If literal infringement were the only test, every infringer could escape by making a cosmetic, insubstantial substitution. The law refuses to let form triumph over substance. Enter the doctrine of equivalents.

Under the doctrine of equivalents (often shortened to "DOE"), a product can infringe even if one or more claim elements are not literally present, so long as the differences between the accused element and the claimed element are insubstantial. The doctrine exists, as courts have long explained, so that the value of a patent is not destroyed by minor, unimportant changes that contribute nothing of substance.

The Supreme Court's foundational modern statement of the doctrine is Warner-Jenkinson Co. v. Hilton Davis Chemical Co., 520 U.S. 17 (1997). Warner-Jenkinson confirmed that the doctrine of equivalents lives on but imposed a crucial discipline on it: the doctrine must be applied element by element, not to the invention as a whole. This is sometimes called the "all-elements rule for equivalents." You cannot say "well, the overall gadgets are pretty similar, so it's equivalent." You must take each missing or differing claim element and ask whether the accused product has an equivalent of that specific element. The all-elements rule, in other words, governs both literal infringement and equivalents.

How do courts decide whether a substitute is an "equivalent"? The most common test is the venerable function-way-result (or "triple identity") test: an accused element is equivalent if it performs substantially the same function, in substantially the same way, to achieve substantially the same result as the claimed element. (The Federal Circuit continues to apply this framework; see, e.g., Siemens Medical Solutions USA, Inc. v. Saint-Gobain Ceramics & Plastics, Inc., 637 F.3d 1269, 1279 (Fed. Cir. 2011).) Courts also frame the inquiry as whether the difference is "insubstantial." Because equivalence is a factual question requiring particularized, element-by-element proof—usually through expert testimony, not lawyer argument—courts regularly grant summary judgment of no equivalents when a patentee offers only conclusory similarity instead of a limitation-by-limitation showing.

Back to our chair. Suppose the claim requires a back "attached to" the seat, and the accused chair's back is instead joined to the seat through an intermediate bracket rather than directly. Literally, the back may not be "attached to" the seat as construed. But under the doctrine of equivalents, the patentee can argue the bracket performs the same function (holding the back up), in the same way (a rigid mechanical connection), to the same result (a usable chair back). If a jury agrees the difference is insubstantial, the chair infringes by equivalents even though it dodged the literal language.

The Guardrails on Equivalents: Why You Can't Use the Doctrine to Swallow the Claims

The doctrine of equivalents is powerful, which is precisely why the law fences it in with several important limits. If equivalents could stretch indefinitely, patent claims would lose their public-notice function—nobody could ever be confident their product was clear of a patent. Three guardrails matter most:

1. Prosecution history estoppel (the Festo rule). This is the big one. When an applicant narrows a claim during prosecution to get around prior art—for instance, amending "a fastener" to "a screw" to overcome a rejection—the patentee generally surrenders the territory it gave up and cannot reclaim it through the doctrine of equivalents. The Supreme Court addressed this in Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., 535 U.S. 722 (2002). Festo rejected a rigid rule that any narrowing amendment automatically barred all equivalents for that element. Instead it adopted a flexible, rebuttable presumption: a narrowing amendment presumptively surrenders the equivalents between the original and amended claim language, but the patentee can rebut the presumption by showing, for example, that the equivalent was unforeseeable at the time, that the amendment bore only a tangential relation to the equivalent, or that there was some other reason the patentee could not reasonably have described the equivalent. In plain terms: if you gave up ground to get your patent, you usually don't get to grab it back by calling it an "equivalent."

2. The "all-elements" limitation (from Warner-Jenkinson). As noted, equivalence is judged element-by-element. A patentee cannot invoke the doctrine in a way that effectively reads an entire claim element out of the claim. If applying the doctrine would mean an element is simply ignored—a problem courts sometimes describe as "vitiating" a limitation—the doctrine cannot apply.

3. The dedication-disclosure rule and the prior-art ceiling. Subject matter that the patentee disclosed in the specification but did not claim is generally considered "dedicated to the public" and cannot be recaptured through equivalents. And the doctrine can never be stretched so far that it would cover something that was already in the prior art—you cannot use equivalents to claim what you could not have patented in the first place.

The practical upshot for an accused infringer like Sekure Kart: if you can show that the patentee narrowed the relevant claim element during prosecution, you have a strong argument that prosecution history estoppel bars the patentee from reaching your design via equivalents. That is one reason the prosecution history is such a treasure trove in patent defense.

When Infringement Gets Expensive: Willfulness and Enhanced Damages

Suppose infringement is established. How much does it cost? Ordinary damages are bad enough (more on those below), but the statute lets a court triple the award in egregious cases. Under 35 U.S.C. § 284, a court "may increase the damages up to three times the amount found or assessed." These are called enhanced damages, and they are reserved for willful infringement.

For years, willfulness was governed by a rigid two-part test from the Federal Circuit's In re Seagate decision that required, among other things, proof of "objective recklessness." The Supreme Court swept that framework aside in Halo Electronics, Inc. v. Pulse Electronics, Inc., 579 U.S. 93 (2016). Halo held that enhanced damages are a matter of the district court's discretion, to be exercised in cases of egregious infringement behavior—the kind of "willful, wanton, malicious, bad-faith, deliberate, consciously wrongful, flagrant, or—indeed—characteristic of a pirate" conduct that stands out from the typical case. Halo also lowered the burden of proof to a preponderance of the evidence (rather than the more demanding clear-and-convincing standard) and made willfulness easier to reach a jury on, while emphasizing that enhancement is not automatic even where willfulness is found.

In the years since Halo, the Federal Circuit has organized the inquiry into two distinct steps, and the distinction matters. First comes the willfulness question: did the accused infringer know of the patent (or willfully blind itself to it) and nonetheless deliberately or intentionally infringe? That is a fact question for the jury, and the patentee must prove it by a preponderance of the evidence (Eko Brands, LLC v. Adrian Rivera Maynez Enterprises, Inc., 946 F.3d 1367, 1378 (Fed. Cir. 2020)). Note what willfulness no longer requires: the patentee need not prove wanton, malicious, or bad-faith conduct at this stage. Second, only if the jury finds willfulness, comes the enhancement question: the judge decides, in her discretion, whether the conduct is egregious enough to warrant increasing the award and, if so, by how much—weighing case-specific considerations (the so-called Read factors: whether the infringer copied, whether it formed a good-faith belief of non-infringement or invalidity, its behavior as a litigant, its size and financial condition, the closeness of the case, duration of the misconduct, remedial action, motivation to harm, and any attempt to conceal). See SRI International, Inc. v. Cisco Systems, Inc., 14 F.4th 1323, 1330 (Fed. Cir. 2021). A jury can find willfulness and the judge can still decline to enhance a dollar—or can triple the verdict outright.

The key practical lesson from Halo: culpability is judged at the time of the infringing conduct, not based on litigation-invented defenses. An "objectively reasonable" defense cooked up for trial no longer immunizes a company that barreled ahead at the time with no good-faith basis. A company that knew about a patent and pressed on anyway is exposed. A company that learned of the patent and promptly investigated, formed a reasonable belief that it did not infringe or that the patent was invalid, and acted on that belief, has a much stronger position. Note, too, the floor the Federal Circuit has set: mere infringement, or even a failure to adopt an available non-infringing alternative, is not by itself enough to make infringement willful—there must be something deliberate about it.

This is why the response to a cease-and-desist letter matters enormously. If Sekure Kart receives that letter and simply ignores it, continued sales could later look willful. If instead it promptly engages competent patent counsel, gets a reasoned analysis, and documents a good-faith belief of non-infringement, it builds the kind of record that defeats willfulness. (One statutory caveat: under 35 U.S.C. § 298, the failure to obtain advice of counsel cannot, by itself, be used to prove willful infringement—so a company is not penalized merely for declining to get an opinion letter. But affirmatively obtaining and following sound advice remains a powerful shield.) For a deeper exploration of how good-faith diligence can defang a willfulness claim—drawn from the trademark world but conceptually parallel—see the shield of good faith.

The Other Subsections: § 271(e), (f), and (g)

The three main flavors—direct, induced, contributory—cover the vast majority of cases. But § 271 has a few more subsections that catch situations the core provisions miss, and they surprise even experienced businesspeople. Three are worth knowing.

The Hatch-Waxman safe harbor and "artificial" infringement, § 271(e). This provision does two opposite things. First, it creates a safe harbor: using a patented invention solely to develop and submit information to the FDA (think generic-drug testing) is not infringement, which lets generic and biosimilar makers do their development work before a patent expires. Second, and counterintuitively, it makes the mere filing of an abbreviated new drug application (an ANDA) seeking to market a generic before patent expiry an act of "artificial" infringement—the legal fiction that launched the entire modern pharmaceutical-patent litigation industry. If you work anywhere near drug development, this is the subsection that runs your world.

Exporting the parts, § 271(f). Recall that § 271(a) is territorial—acts must happen in the United States. Companies once tried to exploit that by manufacturing all the components of a patented invention in the U.S., shipping them abroad, and doing final assembly overseas, where the U.S. patent could not reach. Section 271(f) closes that loophole: supplying components from the United States for combination abroad can infringe. But the Supreme Court drew a sharp limit in Life Technologies Corp. v. Promega Corp., 580 U.S. 140 (2017), holding that supplying a single commodity component of a multicomponent invention is not enough to trigger liability under § 271(f)(1), which requires "a substantial portion of the components."

Importing the product of a patented process, § 271(g). What if someone uses your patented process in another country—beyond the reach of § 271(a)—and then imports the resulting product into the United States? Section 271(g) makes importing, selling, or using the product made by a patented process an act of infringement, even though the process itself was practiced abroad. It is the back-door protection for process patents in a global supply chain, and it dovetails with the ITC remedies discussed below.

The Defenses: How an Accused Infringer Fights Back

A cease-and-desist letter is an accusation, not a judgment. The accused infringer has a deep arsenal of defenses, which generally fall into a few families.

1. Non-Infringement

The most common—and often the best—defense is simply: we don't infringe. Remember the all-elements rule. If the accused product or process is missing even one claim element (and there is no equivalent for it), there is no infringement. This is exactly the move Sekure Kart hopes to make. If the asserted patent's claim requires a particular method of "random number generation" and Sekure Kart uses a genuinely different method, then—assuming the difference is substantial and not captured by the doctrine of equivalents—Sekure Kart does not infringe.

This is why the very first thing a sophisticated defendant does is commission a detailed, element-by-element technical comparison ("claim charts") showing precisely which claim elements its product lacks. A clean non-infringement position is the most satisfying defense because it ends the case without ever having to attack the patent's validity. And as we saw above, divided (or joint) infringement is a specialized sub-flavor of this defense: where the steps of a claimed method are split across multiple independent actors so that no single entity performs all of them—and there is no direction, control, or joint enterprise under Akamai—there may be no direct infringement at all, which is fatal to the case.

2. Invalidity

A close second is to attack the patent itself: even if we do what the claim says, the claim should never have issued. A patent that is invalid cannot be infringed. Importantly, issued patents enjoy a statutory presumption of validity under 35 U.S.C. § 282, and an accused infringer must prove invalidity by clear and convincing evidence—a higher bar than the preponderance standard used to prove infringement (Microsoft Corp. v. i4i Ltd. Partnership, 564 U.S. 91 (2011)). The main invalidity theories:

  • Anticipation (lack of novelty), § 102. The claimed invention was already disclosed, in a single prior-art reference, before the patent's critical date. To anticipate, a single reference must disclose every element of the claim—mirroring the all-elements rule, but comparing the claim to the prior art instead of to the accused product. (Patent lawyers capture the symmetry with a tidy maxim: "that which infringes if later, anticipates if earlier.")
  • Obviousness, § 103. Even if no single reference shows the whole invention, the claim is invalid if the differences between it and the prior art would have been obvious to a person of ordinary skill at the time. The framework comes from Graham v. John Deere Co., 383 U.S. 1 (1966), and the modern, flexible approach to combining references comes from KSR International Co. v. Teleflex Inc., 550 U.S. 398 (2007). Our guide to overcoming obviousness rejections unpacks this analysis in detail.
  • Inadequate disclosure, § 112. The patent fails the written-description, enablement, or definiteness requirements—for example, the claims are so ambiguous that a skilled person cannot tell with "reasonable certainty" what they cover (Nautilus, Inc. v. Biosig Instruments, Inc., 572 U.S. 898 (2014)).
  • Ineligible subject matter, § 101. Laws of nature, natural phenomena, and abstract ideas are not patentable. This defense has reshaped software and business-method patents in the wake of Alice Corp. v. CLS Bank; see our deep dive on patent eligibility after Alice.

Crucially, an accused infringer can challenge validity not only in court but also at the Patent Office itself through inter partes review (IPR) and other post-grant proceedings before the Patent Trial and Appeal Board—often a faster, cheaper venue with a lower burden of proof (preponderance, rather than clear-and-convincing). Many defendants pursue parallel IPRs while the district court case proceeds, and a successful IPR can dissolve the lawsuit entirely.

3. Inequitable Conduct

This is the "atomic bomb" of patent defenses, because success renders the entire patent unenforceable. Inequitable conduct means the patentee (or its attorney) breached the duty of candor owed to the Patent Office during prosecution—typically by withholding material prior art with intent to deceive. The Federal Circuit tightened this defense considerably in Therasense, Inc. v. Becton, Dickinson & Co., 649 F.3d 1276 (Fed. Cir. 2011) (en banc), requiring proof, by clear and convincing evidence, of both (1) but-for materiality (the Patent Office would not have allowed the claim had it known) and (2) a specific intent to deceive (mere negligence, even gross negligence, is not enough). Because it must be pleaded with the particularity of fraud under Federal Rule of Civil Procedure 9(b), it is hard to win—but devastating when it lands. Our companion article on finding evidence of inequitable conduct explores how this defense is built and defended; for the prosecutor's-eye view of avoiding it, see inequitable conduct in patent prosecution—navigating the atomic bomb of patent law.

4. License and Exhaustion

If the patentee gave permission, there is no infringement—simple as that. Permission can be an express license (a contract granting the right to practice the patent) or an implied license (inferred from the circumstances of a sale). Closely related is the doctrine of patent exhaustion, which prevents a patentee from "double-dipping" down a supply chain. Once a patentee (or its authorized licensee) sells a patented article, the patent rights in that article are exhausted—the buyer can use and resell it freely, and the patentee cannot use patent law to control what happens next. The Supreme Court extended this principle dramatically in Impression Products, Inc. v. Lexmark International, Inc., 581 U.S. 360 (2017), holding that a sale exhausts patent rights regardless of post-sale restrictions the patentee tries to impose, and that even an authorized sale abroad exhausts U.S. patent rights. (Compare Bowman v. Monsanto Co., 569 U.S. 278 (2013), where the Court held that exhaustion did not let a farmer make new infringing copies of patented seeds by planting and harvesting—buying a patented article does not authorize you to manufacture new ones.) Anyone licensing patents in or out should read our practical guide on how to license your patent.

5. Equitable Estoppel (but Not Laches)

If a patentee led an accused infringer to reasonably believe it would not enforce the patent—through assurances or a long, conspicuous failure to act—and the infringer relied on that to its detriment, equitable estoppel can bar the claim. This often arises with standards-setting organizations and FRAND commitments; see our piece on standard-essential patents and FRAND licensing. Note, however, that the related defense of laches is no longer available against a claim for patent damages: the Supreme Court eliminated it in SCA Hygiene Products Aktiebolag v. First Quality Baby Products, LLC, 580 U.S. 328 (2017), reasoning that the Patent Act's six-year damages limit already supplies the relevant time bar.

The Remedies: What a Winning Patentee Gets

If the patentee proves infringement and survives the defenses, two basic remedies are on the table: money and an injunction.

Monetary Damages

The statute, 35 U.S.C. § 284, guarantees damages "adequate to compensate for the infringement, but in no event less than a reasonable royalty." That floor—a reasonable royalty—is the amount the parties would hypothetically have agreed to in an arm's-length negotiation just before infringement began. Courts run this hypothetical negotiation through the famous (and famously long) list of factors from Georgia-Pacific Corp. v. United States Plywood Corp., 318 F. Supp. 1116 (S.D.N.Y. 1970), weighing things like comparable licenses, the parties' competitive relationship, and the value the patented feature actually adds.

A patentee that competes directly with the infringer may instead (or additionally) seek lost profits—the profits it would have earned on sales the infringer stole. The classic test comes from Panduit Corp. v. Stahlin Bros. Fibre Works, Inc., 575 F.2d 1152 (6th Cir. 1978), requiring proof of demand for the patented product, the absence of acceptable non-infringing substitutes, the capacity to meet the demand, and the amount of profit.

Modern damages law has grown notably more rigorous about apportionment—isolating the value of the patented feature from everything else in a multi-component product. The Federal Circuit polices "entire market value" theories closely, generally requiring the patentee to tie damages to the smallest salable patent-practicing unit and to apportion out the value of unpatented features (see VirnetX, Inc. v. Cisco Systems, Inc., 767 F.3d 1308 (Fed. Cir. 2014)). Two timing limits also matter: damages reach back no more than six years before suit (§ 286), and if the patentee sells products covered by the patent, it generally cannot recover pre-notice damages unless it marked those products with the patent number (§ 287). That marking rule catches a surprising number of patentees off guard; for the nuts and bolts, see understanding patent marking requirements.

On top of compensatory damages sit the enhanced damages (up to treble) we discussed under Halo, and—in "exceptional" cases—attorney fees under 35 U.S.C. § 285. The Supreme Court loosened the fee standard in Octane Fitness, LLC v. ICON Health & Fitness, Inc., 572 U.S. 545 (2014), defining an exceptional case as one that simply "stands out from others" in the strength of a litigating position or the unreasonable manner in which it was litigated—a standard that cuts both ways and can punish overaggressive patentees as readily as bad-faith infringers.

Injunctions

The most fearsome remedy is an injunction—a court order to stop making, using, or selling the infringing product. Because a patent is a right to exclude, injunctions were once nearly automatic for prevailing patentees. No longer. In eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388 (2006), the Supreme Court held that patent plaintiffs must satisfy the traditional four-factor test for injunctive relief: (1) irreparable injury; (2) inadequacy of money damages; (3) a balance of hardships favoring an injunction; and (4) that the public interest would not be disserved. The practical consequence: a patentee that competes head-to-head with the infringer usually still gets its injunction, while a non-practicing entity that only licenses its patents (and could be made whole with a royalty) often does not. eBay fundamentally changed the leverage dynamics in patent disputes—and is a big part of why so much enforcement has migrated to the ITC, which we turn to next.

A Special Forum: The International Trade Commission

One more piece of the infringement landscape deserves attention because it often surprises people: importation. Recall that § 271(a) makes importing a patented invention an act of infringement. But patentees facing infringing imports have a powerful alternative to district court—the U.S. International Trade Commission (ITC).

Under Section 337 of the Tariff Act of 1930 (19 U.S.C. § 1337), the ITC can investigate the importation of articles that infringe a U.S. patent and, if it finds a violation, issue an exclusion order barring those products at the border—enforced by U.S. Customs. The ITC is popular for two reasons: it is fast (investigations move on a tight statutory timeline) and its remedy—keeping infringing goods out of the country entirely—can be far more potent than a damages award, especially after eBay made district-court injunctions harder to get. The ITC cannot award money damages, and its decisions on infringement do not bind district courts, so patentees frequently litigate in both forums at once. One quirk worth flagging: the ITC has grappled with whether induced infringement can support a Section 337 violation when the direct infringement occurs only after importation. The Federal Circuit answered yes in Suprema, Inc. v. International Trade Commission, 796 F.3d 1338 (Fed. Cir. 2015) (en banc), holding that the ITC may bar articles that are used to directly infringe post-importation at the inducement of the importer. For global enforcement strategy across borders, see our global patent litigation strategies.

Putting It All Together: Sekure Kart Revisited

Let's return to that cease-and-desist letter and walk Sekure Kart through the analysis like a seasoned patent lawyer would.

First, read the claims. Not the splashy summary in the letter—the actual numbered claims of the asserted patent. Identify the independent claims (the standalone ones) and list out their elements. Suppose Claim 1 requires, among other things, a step of "random number generation."

Second, construe the key terms. What does "random number generation" mean in this patent? Read the specification and the prosecution history. Did the patentee define the term? Did it narrow that element during prosecution to overcome prior art? If the patent only ever describes standard, deterministic methods—and the prosecution history shows the applicant distinguished its method from others—that history may both narrow the literal scope and trigger prosecution history estoppel limiting the doctrine of equivalents.

Third, compare element-by-element. Walk through every element of Claim 1 against Sekure Kart's actual product. If Sekure Kart's number-generation method is genuinely, substantially different—not just cosmetically relabeled—then Sekure Kart lacks that element literally. And if prosecution history estoppel bars the patentee from reaching Sekure Kart's method through equivalents, there is no infringement of Claim 1, full stop. That detailed technical report from the engineering team—the 42 pages of flowcharts explaining 27 concrete differences—is exactly the evidence that builds a non-infringement defense. (And if the asserted claim happens to be a method claim whose steps are split between Sekure Kart's servers and its customers, the divided-infringement analysis from Akamai offers a second, independent escape hatch.)

Fourth, consider the patent's validity as a backup. Even if there were an infringement risk, is the patent vulnerable to an obviousness or eligibility challenge? Could an IPR knock it out?

Fifth, manage the willfulness clock. Because Sekure Kart now has notice of the patent, continuing to sell without a reasoned basis could expose it to enhanced damages later. Promptly engaging patent counsel and documenting a good-faith non-infringement (or invalidity) position is not just good lawyering—it is the Halo shield in action.

If Sekure Kart does this homework and the analysis holds, the company can respond to the letter from a position of strength: here is why we do not infringe, here is the technical basis, and if you would like a fight, we are ready. Whether to send that response, negotiate, seek a declaratory judgment, or simply hold the line is a strategic call best made with counsel—and one our comprehensive litigation guide addresses at length.

Key Takeaways

  • Infringement is a checklist. A patent's claims are a list of elements. Under the all-elements rule, you infringe a claim only if your product or process contains every element—literally or by equivalents. Miss one, and you do not infringe that claim.
  • There are three core statutory flavors—plus specialty subsections. Section 271(a) covers direct infringement (strict liability—no intent required). Section 271(b) covers induced infringement (knowledge and intent). Section 271(c) covers contributory infringement (a special-purpose component with no substantial non-infringing use). Subsections (e), (f), and (g) handle FDA testing/ANDA filings, component exports, and imports made by a patented process. Indirect infringement always requires an underlying direct infringer.
  • Divided infringement is the modern battleground. Where method steps are split among actors, there is a direct infringer only if one party directs or controls the others or they form a joint enterprise (Akamai, en banc 2015). For defendants, a genuine split is a clean non-infringement defense; for patentees, it is a drafting warning.
  • The analysis is two steps. First, the judge construes the claims (the Markman process, guided by Phillips, with subsidiary fact findings reviewed for clear error under Teva). Then the factfinder compares the construed claim to the accused product.
  • Literal infringement is a word-for-word match; the doctrine of equivalents (Warner-Jenkinson) catches insubstantial changes via the function-way-result test—but is fenced in by prosecution history estoppel (Festo) and the element-by-element rule.
  • Willfulness can triple the damages. Under Halo, willfulness (deliberate infringement, judged at the time of the conduct) is a jury question; enhancement is then a discretionary judicial decision weighing the Read/SRI factors—so how you respond to notice matters enormously.
  • Defenses are deep. Non-infringement (including divided infringement), invalidity (anticipation, obviousness, § 112, § 101), inequitable conduct, license/exhaustion, and equitable estoppel each can defeat a claim.
  • Remedies range from a reasonable royalty floor and lost profits to enhanced damages, attorney fees in exceptional cases (Octane Fitness), and injunctions under the four-factor eBay test. The ITC offers a fast, border-exclusion alternative for infringing imports.

Frequently Asked Questions

Do I have to copy a patent to infringe it? No. Direct infringement under § 271(a) is a strict-liability offense. You can independently invent the very same thing, in complete good faith, never having heard of the patent, and still infringe if your product contains every element of a claim. Patent law has no "independent creation" defense to direct infringement—unlike copyright law. Knowledge and copying do matter for indirect infringement (inducement and contributory) and for willfulness, but not for basic direct infringement.

What is the difference between literal infringement and the doctrine of equivalents? Literal infringement means the accused product matches the claim language word-for-word, element-by-element. The doctrine of equivalents reaches products that change one or more elements only insubstantially—the function-way-result test asks whether the substitute performs substantially the same function, in substantially the same way, to reach substantially the same result. The doctrine prevents infringers from escaping liability through trivial, cosmetic changes, but it is limited by prosecution history estoppel and the rule that equivalence is judged element-by-element (Warner-Jenkinson; Festo).

Why does claim construction matter so much? Because the meaning of a single word can decide the entire case. Claim construction is the judge's job (the Markman process), and the construed meaning becomes the yardstick the jury uses to measure infringement. Whether a strip that triggers at 138°F infringes a claim reciting "above 140°F," or whether a bracketed connection counts as "attached," turns on how the court construes those terms—using the claims, specification, and prosecution history as its primary guide (Phillips v. AWH).

What is divided infringement, and why do software companies care so much about it? Divided (or joint) infringement arises when the steps of a patented method are performed by different actors, so no single entity does all of them. Because direct infringement normally requires one actor to perform every step—and because indirect infringement needs a direct infringer—a true split can mean no one is liable. Under the en banc Akamai decision, a single party can still be a direct infringer if it directs or controls the others (including by conditioning a benefit on their performance and dictating the manner or timing) or if the actors form a joint enterprise. Software, networking, and internet companies care intensely because their claimed methods are so often divided between a service provider and its users.

What makes infringement "willful," and what happens if it is? Willful infringement is deliberate or intentional infringement by a party that knew of the patent (or willfully blinded itself to it), judged by the infringer's conduct and state of mind at the time, not by defenses cooked up later. Under Halo Electronics, a jury decides willfulness by a preponderance of the evidence, and the judge then has discretion to enhance damages up to three times the compensatory award for sufficiently egregious conduct. The best protection is to take any notice of a patent seriously, investigate promptly, and form (and document) a reasonable good-faith belief of non-infringement or invalidity. Simply declining to get an opinion of counsel cannot, by itself, prove willfulness (35 U.S.C. § 298).

Can I be sued just for selling a part? Possibly—but only in narrow circumstances. Contributory infringement under § 271(c) reaches the sale of a component that is especially made for an infringing use and has no substantial non-infringing use, where you know it is for infringement. If the part is a "staple article of commerce" with real lawful uses (think generic screws, common chemicals, general-purpose chips), you are not a contributory infringer. You might separately face an inducement claim if you actively encourage others to infringe—or, if you ship components abroad for assembly, a claim under § 271(f).

What is the strongest defense to a patent infringement claim? There is no single "strongest" defense—it depends on the facts—but a clean non-infringement position is often the most attractive, because it ends the case without having to attack the patent's validity: you simply show your product lacks at least one claim element (or, for divided method claims, that no single actor performs all the steps). Invalidity (the patent never should have issued) is the next major front and can be pursued in court or through an inter partes review at the Patent Office. Inequitable conduct is the most devastating when it succeeds (it makes the whole patent unenforceable) but the hardest to prove. Exhaustion or a license can provide a complete defense if you bought the product from an authorized source.

How long does a patent last, and can I infringe after it expires? A utility patent generally lasts 20 years from its earliest non-provisional filing date (subject to maintenance fees and possible term adjustments). Once a patent expires, the invention enters the public domain and anyone may freely practice it—there is no such thing as infringing an expired patent. You can, however, owe damages for infringement that occurred while the patent was in force, reaching back up to six years before suit. For the lifecycle of a patent, see utility patent basics and general information concerning patents.

What are the consequences if a court finds I infringed? At minimum, compensatory damages—no less than a reasonable royalty, and possibly lost profits if you competed with the patentee. In egregious cases, damages can be tripled, and attorney fees may be awarded in "exceptional" cases. The court may also enter an injunction stopping you from making, using, or selling the product (under the four-factor eBay test), and the ITC may exclude infringing imports at the border. For the broader landscape of IP enforcement consequences, see what are the consequences of pirating intellectual property.

Related Articles

This article provides general information about U.S. patent law and is not legal advice. Patent infringement analysis is fact-intensive and the law continues to evolve; for advice about a specific patent, product, or dispute, consult qualified patent counsel. To discuss a matter with mclaw.io, reach out to admin@mclaw.io.