Patents

No Inducement Without a Direct Infringer: Limelight Networks v. Akamai

A unanimous Supreme Court held that inducing infringement under Section 271(b) requires a predicate act of direct infringement, refusing to extend liability across divided method steps.

June 16, 2025
Server racks in a data center supporting a content delivery network
Akamai's patent claimed a method of distributing web content across a content delivery network, with one step performed by the provider's customers. Shutterstock
Educational content, not legal advice. This article explains general legal concepts. It does not create an attorney–client relationship. For your specific situation, consult a licensed attorney.

In Limelight Networks, Inc. v. Akamai Technologies, Inc., No. 12-786 (U.S. June 2, 2014), a unanimous Supreme Court resolved a structural question that had divided the Federal Circuit: can a defendant be liable for inducing infringement of a method patent when the patent’s steps are split among multiple actors and no single party performs them all? Writing for the Court, Justice Alito answered no. Inducement under 35 U.S.C. § 271(b) must rest on a predicate act of direct infringement, and where the steps of a method are divided such that no one has committed direct infringement, there is nothing to induce. The decision reversed an en banc Federal Circuit that had tried to plug a perceived gap in the law—and it reframed how patentees must think about claims that contemplate cooperation between a service provider and its users.

At a glance

  • Court and date: Supreme Court of the United States, decided June 2, 2014; docket No. 12-786. Justice Alito wrote for a unanimous Court, reversing and remanding the en banc Federal Circuit.
  • Patent: Akamai is the exclusive licensee of a patent claiming a method of delivering electronic content over a content delivery network (CDN), reducing congestion by storing content on multiple servers.
  • The divided step: Limelight operates a competing CDN and performs most of the claimed steps, but its customers—not Limelight—perform the “tagging” step that designates which content is served from the network.
  • The Federal Circuit’s theory: The en banc court held that Limelight could be liable for inducement even if no single actor was liable for direct infringement, so long as all steps were collectively performed.
  • Holding: A defendant cannot be liable for inducing infringement under § 271(b) when no one has directly infringed under § 271(a) or any other provision. Liability for inducement is predicated on direct infringement.

The divided-infringement gap

Method patents create a distinctive vulnerability. A claim that recites a series of steps is directly infringed under § 271(a) only when a single actor performs every step—or is legally responsible for the performance of all of them. When the steps are naturally split between a company and its customers, a competitor can structure its operations so that it performs most steps itself and leaves one or two to its users. If no single entity performs the entire method, traditional direct-infringement doctrine finds no infringer.

That was Limelight’s posture. It carried out the bulk of Akamai’s claimed CDN method but left the tagging step to customers, who decided which content to route through the network. Under the Federal Circuit’s then-governing single-entity rule, that division meant no one directly infringed: Limelight did not perform every step, and the customers were not acting as Limelight’s agents or under its direction and control. Akamai’s patent appeared to describe exactly what Limelight was doing, yet the structure of the activity slipped between the doctrinal categories.

The Federal Circuit’s workaround—and why it failed

The en banc Federal Circuit tried to solve the problem through inducement. It held that a defendant who performs some steps of a method and encourages others to perform the rest could be liable under § 271(b) for inducing infringement, even if no single party was liable for direct infringement. The court effectively decoupled inducement from a completed act of direct infringement, treating the collective performance of all steps as sufficient.

The Supreme Court rejected the move as incompatible with the statute’s architecture. Section 271(b) makes liable one who “actively induces infringement of a patent.” But “infringement,” the Court explained, has a settled meaning tied to § 271(a): a method is infringed only when all of its steps are performed. If no actor has committed direct infringement—because the steps are divided and no single party is responsible for all of them—then there is no infringement in existence to be induced. Inducement liability is, by its nature, derivative; it presupposes a completed wrong by someone. The Federal Circuit’s rule would have created the anomaly of inducement liability for conduct that, on the law as the court itself had defined it, no one was committing.

The Court drew an instructive analogy to criminal aiding-and-abetting: one cannot aid and abet a crime that was never committed. The same logic governs § 271(b). A defendant who encourages a course of conduct that, in total, never amounts to direct infringement by anyone cannot be said to have induced infringement.

What the Court deliberately left open

The decision is as notable for its restraint as for its holding. Akamai had urged the Court to revisit the Federal Circuit’s narrow single-entity standard for direct infringement under § 271(a)—the rule that attributes one actor’s conduct to another only in limited agency-like circumstances. The Court declined. It assumed, without deciding, that the Federal Circuit’s § 271(a) standard was correct and rested its analysis on the consequences that follow if it is. The Court expressly noted that the Federal Circuit would be free, on remand, to reconsider the § 271(a) question.

That invitation proved important. On remand, the en banc Federal Circuit broadened the test for when one entity is responsible for another’s performance of a method step. It held that direct infringement can be attributed to a single actor not only through a principal-agent relationship or a contractual obligation, but also where the actor conditions participation in an activity or receipt of a benefit on performance of the step and establishes the manner or timing of that performance. Under that expanded standard, the dispute ultimately resolved in Akamai’s favor—a reminder that Limelight closed the inducement route while pointing toward a revived direct-infringement theory.

Open questions

Limelight settled that inducement needs a direct infringer but left the boundaries of the direct-infringement attribution doctrine to be worked out below. The post-remand “directs or controls” test—particularly the “conditions a benefit and dictates the manner or timing” branch—remains fact-intensive, and its outer limits continue to be litigated. How much control over customers’ conduct is enough to attribute their steps to a provider? When does a standard terms-of-service relationship cross from arm’s-length dealing into the kind of direction that supports single-entity liability? These questions decide the fate of countless method claims that contemplate user participation, and Limelight itself answered none of them.

Implications

  • Inducement is derivative. Without a predicate act of direct infringement by some single responsible actor, there is no § 271(b) liability—full stop.
  • Draft method claims for a single actor. Claims whose steps must be performed by different parties are structurally fragile; where possible, recite steps performable by one entity.
  • Attribution is the new battleground. After Limelight, the decisive question in divided-infringement cases is whether one party “directs or controls” another’s performance under the Federal Circuit’s expanded § 271(a) standard.
  • System and apparatus claims offer alternatives. Where a method claim invites division of steps, a corresponding system claim performed by a single accused entity may avoid the divided-infringement trap.
  • Encouragement alone is not enough. Instructing customers to perform a missing step does not create inducement liability unless the customers’ performance is itself attributable to a direct infringer.

Frequently asked questions

Why couldn’t Limelight be liable for inducement? Because inducement under § 271(b) requires an underlying act of direct infringement, and none existed. Limelight performed most of the claimed steps, but its customers performed the tagging step, and no single party performed or was responsible for all the steps. With no direct infringement, there was nothing for Limelight to induce.

Did the Supreme Court decide when one party’s conduct counts as another’s? No. The Court assumed the Federal Circuit’s existing § 271(a) single-entity standard was correct and declined to revisit it, expressly leaving that question for the Federal Circuit on remand. The Federal Circuit later broadened the attribution test.

What changed after the remand? On remand, the en banc Federal Circuit expanded direct-infringement attribution, holding that a single entity is responsible for others’ performance where it conditions participation or a benefit on performing a step and establishes the manner or timing of that performance. That broader standard ultimately favored Akamai.

Authorities and sources