The Patent Act has been interpreted to provide that there is liability for direct patent infringement only when a single person or entity performs all the elements of a patent claim, either by itself or vicariously, through an agent. Does that mean it is possible for someone to avoid infringement liability by performing some of the steps of a method patent himself and inducing someone else to perform the rest, or by inducing a group of persons to perform the steps, so that no one person does them all?

By a narrow 6-5 vote, the en banc U.S. Court of Appeals for the Federal Circuit in Akamai Technologies Inc. v. Limelight Networks Inc., 2012 WL 3764695 (Fed. Cir. August 31, 2012), answered those questions with a clear “no.” In doing so, the en banc majority struggled to reconcile basic principles of patent infringement, reinterpreted the Patent Act and overruled prior precedent, all in an effort to make it easier for owners of method patents to bring claims of induced infringement.


Federal courts have repeatedly ruled that direct infringement under § 271(a) of the Patent Act occurs only when a single actor commits all the acts necessary for infringement. Because direct infringement is a strict liability tort (no intent is required), this single-entity rule helps to prevent ensnaring actors who might have “no way of knowing that others were acting in a way that rendered their collective conduct infringing.” It is usually easy to identify a party responsible for direct infringement in the case of product or apparatus patents, because the entity that installs the final part, completing the claimed invention, is usually considered a direct infringer. However, there are no such presumptions for method patents, so that only parties who perform all the steps of the method are direct infringers. Therefore, dividing up the steps among two or more parties typically defeats a direct infringement claim.

In addition to direct infringement under § 271(a), § 271(b) of the Patent Act imposes liability for induced infringement, when one party causes, urges, encourages or aids another to infringe. But cases such as BMC Resources Inc. v. Paymentech L.P., 498 F.3d 1373 (Fed. Cir. 2007), hold that there cannot be induced infringement unless the induced party commits direct infringement. Putting those pieces together, one who induces another party to perform at least one of the steps of a method patent (or induces a group of others to perform separate steps) will defeat claims of direct and induced infringement. That was the presumed state of the law when the en banc Federal Circuit heard Akamai.

The Akamai court considered two consolidated appeals. One dealt with Akamai’s patent claiming a method of delivery of web content whereby some of the content for a web page is placed on a set of replicated servers and the content provider’s web page instructs web browsers to retrieve the content from those servers. Limelight maintains a network of servers that allow for content delivery to web pages, as in the patented method. However, because Limelight does not modify the web pages itself, but instead instructs content providers on the steps necessary to make that modification, the district court found Limelight had not completed all of the steps in the patented method and had not induced infringement.

In the second appeal, McKesson Corp. had patented a method for electronic communications between health care providers and patients. Portions of software sold by Epiq Systems Inc. to health care providers permitted such communication. However, Epiq does not perform any of the steps in the method patent, which are instead divided between the patients and health care providers themselves. As in Akamai, because the patients, not Epiq’s direct customers, performed the step of initiating communication, the district court found that Epiq was not liable.


The en banc majority was clearly concerned that, under existing law, in these circumstances “the patentee has no remedy, even though the patentee’s rights are plainly being violated by the actors’ joint conduct.” Several amici, including the American Intellectual Property Law Association, urged the court to remedy this perceived injustice by construing § 271(a) to recognize direct infringement by more than one actor. Under this view, direct infringement would occur if all the steps of the method are performed, regardless of the number of entities involved. Apportioning responsibility for direct infringement among those entities would be done through established tort law principles. Presumably a jury would assess the culpability of each actor whose conduct was a substantial cause of the infringement. And any party that induced one or more others to infringe would necessarily be liable for induced infringement.

But the en banc majority did not take this approach. It found no need to determine whether the single-entity rule for direct infringement should be discarded. Instead, it decided that, even if the single-entity rule properly applies to direct infringement, it need not be applied to induced infringement — in other words, that infringement might have different meanings depending on whether the patentee claimed direct or secondary liability.

While recognizing that “[t]he principle, that there can be no indirect infringement without direct infringement, is well settled,” the majority found that “[r]equiring proof that there has been direct infringement as a predicate for induced infringement is not the same as requiring proof that a single party would be liable as a direct infringer.” Therefore, “all the steps of a claimed method must be performed in order to find induced infringement,” but it is “not necessary to prove that all the steps were committed by a single entity.” On that basis, the court overruled BMC Resources and other cases that had held that there could be no induced infringement unless a single entity was liable for direct infringement.

Looking to the legislative history of the 1952 Patent Act, as well as tort and criminal law concerning aiding and abetting, the majority found that the statute had been designed to impose liability for inducement in cases of divided infringement. Congress did not intend to allow parties to “knowingly sidestep infringement liability simply by arranging to divide the steps of a method claim between them.” Because induced infringement, unlike direct infringement, requires intent, the court found no need to insist on a single-entity rule in order to protect innocent actors. Rather, “[i]f an entity has induced conduct that infringes a patent, there is no justification for immunizing the inducer from liability simply because no single party commits all of the components of the appropriative act.”

Judge Richard Linn’s dissent, joined by three other judges, argued that the statute plainly imposes a single-actor requirement for direct infringement and that infringement cannot have a different meaning in inducement cases. The majority, he wrote, “effectively rewrites” the statute to “accord patentees certain extended rights that a majority of this court’s judges would prefer that the statute covered.” The dissent noted that Congress had repeatedly amended § 271 to define additional acts of infringement (for example, supplying components of a patented invention for combination outside the United States), but had not criticized the single-entity rule. Indeed, Congress passed a significant amendment of the act just last year. “Broadening the doctrine of inducement, such that no predicate act of direct infringement is required, is a sweeping change to the nation’s patent policy that is not for this court to make.”

The dissent also found that, through foresighted claim drafting, patentees can take steps to prevent others from avoiding deserved liability for induced infringement. “A patentee can usually structure a claim to capture infringement by [a] single party.” Because the patentee defines the boundaries of the patent, the “claim drafter is the least cost avoider of the problem of unenforceable patents due to joint infringement.” Linn’s dissent would have affirmed the findings of the district courts rejecting claims of induced infringement. A separate dissent by Judge Pauline Newman took a different approach, proposing to abolish the single-entity rule for direct infringement.

For several reasons, even beyond the close vote in the Federal Circuit, Akamai would appear to be an excellent candidate for U.S. Supreme Court review. The stakes are high. Many method patents — for example, in the health care and information technology industries — can easily be performed by multiple parties, allowing evasion under pre-Akamai rules. On the other hand, users of technology may claim that they structured their affairs in reliance on pre-Akamai law, which allowed them to avoid infringement by arranging for multiple parties to carry out required steps of a patented method. Those settled expectations, they might argue, have been upset by the majority opinion.

The majority and the four-judge dissent also took different approaches to statutory interpretation, an issue that has interested the Supreme Court. While the dissent emphasized the statutory text, the majority apparently was influenced by its view of congressional policy. Proper construction of the statute, the majority wrote, should be informed by the assumption that Congress would not permit “ready evasion of valid method claims with no apparent countervailing benefits.” Some members of the Supreme Court have been critical of statutory interpretation that appears to deviate from the text. And, of course, if the Supreme Court overturns Akamai, Congress will have the last say.

Lewis R. Clayton is a litigation partner in the New York office of Paul, Weiss, Rifkind, Wharton & Garrison and co-chairman of the firm’s intellectual property litigation group. He can be reached at Rachale C. Miller, an associate at the firm, assisted in the preparation of this article.