Limelight Networks, Inc. v. Akamai Technologies, Inc., mercifully brings to a close the Supreme Court’s year-long session on patent law, which has featured five substantive patent-law cases in the last three months of the Term. The issue in Limelight has preoccupied the Federal Circuit for the last several years, and it comes before the Court in a long-standing, high-stakes dispute over Internet technology.

Akamai’s patent in this case involves the method by which web pages that include images upload those images to users.  To simplify a great deal, the patent contemplates a network of distributed servers (a “content delivery network” or “CDN”) on which the large image files reside.  So, for example, when a California browser asks for an image-laden page from a site like, the text and frame of the page (which is not all that much data) come from the server (which might be in Connecticut, for example). The images, however, come from whatever CDN server is most easily able to deliver them to California, often the server closest to the requesting person.  This solves two problems.  The first is the private problem of ESPN having to maintain a server so large that it can handle peak demand for image-laden pages (imagine the demand the evening of the Super Bowl as compared to mid-day on a routine business day). The second is the public problem of burdening the Internet backbone with transmission of large image files all the way across the country (or the world).

Limelight competes with Akamai, operating such a network, marketing its services to companies with a large web presence like Microsoft and DirecTV.  When Akamai sued Limelight, Limelight responded that its activities did not amount to infringement.  It might be (and apparently is) the case that the use of Limelight’s services necessarily uses Akamai’s patent; but because the customers take some steps of the patented method and Limelight takes others, no single party implements the entire patented method.

The district court found no infringement and held in favor of Limelight, and a panel affirmed (both applying settled Federal Circuit law).  But the Federal Circuit took the case en banc; reversing direction, it ruled for Akamai and found infringement in a splintered decision.  Largely rejecting its prior holdings, the Federal Circuit held that a party induces infringement not only when a single party acts out the entire method, but also when that party “advises, encourages, or otherwise induces others” to complete the infringing conduct. Responding to the disarray in the court of appeals’ decision, the Supreme Court granted review.

Having offered what I would like to think is a relatively objective background summary, this is the point where I should admit that I always have regarded the Federal Circuit’s doctrine in this area as bizarre. The doctrine first arose in a 2007 case called BMC Resources Inc. v. Paymentech LLP.  Essentially, that case held that a payments processing patent can’t be infringed if no single party implements the entire method.  That seemed crazy to me because the very premise of payments networks is to transfer funds rapidly and seamlessly among a web of interconnected parties, through countless distinct intermediaries.  I would have thought it common, and indeed desirable, for inventors to pursue technological improvements that facilitated those network activities, and I would have expected that many of those patents described joint conduct.  The Federal Circuit rule, as far as I could tell, seemed to say that patents of that nature would as a practical matter never be enforceable.  How much sense can that make?

To be sure, I’d never spent the time to investigate the question closely, but having read the briefs for this case I still can’t understand why so many judges have regarded this as a difficult, or even close, question.  So I think the most useful thing I can do here is summarize the arguments as they appear in the briefs and explain what I find so puzzling about them.

It’s not as if Limelight hasn’t retained capable counsel.  Limelight is represented by Aaron Panner and the Kellogg, Huber firm, which has argued as many cases in the Supreme Court in the last few years as any private firm in the country. So I start from the premise that he has skillfully presented the best argument for the no-joint-infringement position.

Not surprisingly, the linchpin of the argument is directly textual, and given my skepticism it is useful to lay out the words in extenso. The relevant text (Patent Act § 271(b)) provides that “[w]hoever actively induces infringement of a patent shall be liable as an infringer.” The word “infringe” self-evidently refers to Patent Act § 271(a), providing that “whoever . . . makes [or] uses . . . any patented invention . . . infringes the patent.”

Limelight argues that because “infringement” in Section 271(b) must have the same meaning as “infringe” in Section 271(a), there can be no infringement under Section 271(b) unless the induced activity, standing alone, amounts to infringement.  Here, Limelight’s customers essentially “tag” the files for identification in Limelight’s CDN.  Because that tagging is only a part of the patented method, and because Limelight doesn’t do the tagging (but instead operates the CDN), neither party implements the entire method. Accordingly, because the customers have not by themselves infringed, the activity that Limelight has induced is not itself “infringement.”  Hence, Limelight vehemently insists, there can be no infringement and thus no inducement liability.

The argument plainly is a weighty one; many Federal Circuit judges over the last several years have found it compelling, though it fell just short of persuading a majority of the en banc panel in this case.  The Solicitor General plainly finds it insuperable:  the Solicitor General’s brief in this case, filed in support of Limelight, argues that the no-joint-infringement rule is bad policy, but suggests that the text is so clear that the Court has no choice but to overturn the Federal Circuit’s judgment and await congressional relief.  Microsoft (one of Limelight’s customers) files a similar brief, taking no position at all on the wisdom of the policy, but suggesting that only Congress can prescribe liability here.

But I just can’t find this in the statute.  The statute suggests that what is reprehensible is “induc[ing] infringement.”  At least as far as we know at this point in the process, Limelight’s business model depends on the Akamai method.  Customers cannot use Limelight’s services without implementation of the patented Akamai method.  So how can it be so clear that Limelight does not “induc[e] infringement” when it sells the services that use the patented method? I just don’t see anything in the statute suggesting a lens through which infringement must be located in the acts of a single person.

And having now read the briefs, it seems plain to me that there is nothing that compels the Court to that result. None of its prior cases seems to shed any substantial light.  The able lawyers on both sides (Seth Waxman is counsel of record for Akamai) have proffered the mass of quoted snippets you might expect, but none of them reflects attention to the key situation:  actions by two parties that collectively, though not individually, implement a patented method.

The policy implications seem to weigh even more heavily in favor of Akamai.  As summarized above, it is self-evident that a large number of importantly innovative patents in the Internet era involve collaborative activity involving many parties connected at most indirectly. Limelight states that all patent-worthy methods can be redrafted to focus on the activity of a single entity.  But that argument is difficult to credit, and Limelight offers no explanation or examples that assuage that concern at all.

Moreover, the no-joint-infringement rule seems to make abusive patent avoidance trivially easy: all the rival need do is design a business model in which some single step of the method is taken by a customer or supplier and the patent becomes inapplicable.  Limelight suggests that this is admirable “inventing around” a patent, but if it still involves execution of the patented method it looks a lot more like abuse to me and a lot less like inventing around.

To make matters even worse for Limelight, Akamai urges persuasively that this case doesn’t even present the Section 271(b) question (whether Limelight induced infringement), because Limelight is so closely involved with its customers that Limelight can be held liable for direct infringement under Section 271(a).  In fact, a jury found Limelight guilty of infringement under Section 271(a) based on a finding that Limelight “directed” or “controlled” its customers. To be sure, it is not clear that question is before the Court (because it did not grant Akamai’s cross-petition), but it supports the judgment below; the Court well might find an answer to the problem in Section 271(a).  Indeed, IBM (filing a brief that like the Solicitor General and Microsoft regards the Section 271(b) textual argument as insuperable) argues that Section 271(a) is the best way out of the box, suggesting that this is a textbook case for Section 271(a) liability.

So at the end of the day I still can’t understand why this case is so hard.  But the past record of the issue suggests that the Justices will see the case as much closer than I do.  For me, the oral argument should reveal whether my preexisting sentiments have blinded me to the difficulties of the question.

Posted in Limelight Networks v. Akamai Technologies, Featured, Merits Cases

Recommended Citation: Ronald Mann, Argument preview: Ending the Term not with a bang, but a whimper, SCOTUSblog (Apr. 28, 2014, 12:00 PM),