Google v. Oracle: Case Summary and Fair Use Ruling

In Google v. Oracle, decided in April 2021, the U.S. Supreme Court ruled 6-2 that Google’s copying of roughly 11,500 lines of code from Oracle’s Java platform to build the Android operating system qualified as fair use under federal copyright law. Justice Breyer wrote the majority opinion, joined by five other justices. The decision ended a decade of litigation and gave the software industry a strong signal that reimplementing the functional labels of an application programming interface, or API, can be lawful, though the Court deliberately declined to decide whether that code was copyrightable in the first place.¹

The Dispute: What Google Actually Copied

Java began as a Sun Microsystems programming language built to run across different computers. It caught on quickly, and by the time Oracle acquired Sun in 2010, millions of developers had learned Java’s system of commands and organizational structure. That acquisition gave Oracle the copyright to Java’s code and its API.

Java’s API contains two layers. Declaring code provides the names and organizational labels a programmer types to invoke a function, such as java.lang.Math.max to find the larger of two numbers. Implementing code performs the actual computational work behind the scenes. When Google built Android, it wrote its own implementing code from scratch. What it copied was the declaring code: the labels and the three-tier organizational structure Java developers already knew.¹

Google’s reasoning was practical. Programmers had already invested years learning Java. Keeping the familiar labels let them build Android apps without starting over. The 11,500 lines Google copied made up about 0.4% of the Java API’s 2.86 million total lines. Oracle saw that as theft. Google framed it as using a shared vocabulary to build something new.

How the Case Reached the Supreme Court

Oracle sued Google in the U.S. District Court for the Northern District of California in 2010, alleging copyright and patent infringement.² The district judge initially ruled that Java’s API structure was an uncopyrightable “method of operation” under federal copyright law.³ The U.S. Court of Appeals for the Federal Circuit, which had jurisdiction because the case originally included patent claims, reversed. On remand, a jury found Google’s use was fair. The Federal Circuit reversed again, holding as a matter of law that the copying was not fair use. Google appealed to the Supreme Court.

Why the Court Sidestepped Copyrightability

Two questions came before the justices: whether Java’s declaring code could be copyrighted at all, and whether Google’s copying was fair use. The majority answered only the second. Justice Breyer wrote that the Court would assume, purely for the sake of argument, that Oracle’s code was copyrightable, then move directly to fair use.¹

The Court justified that restraint by citing “rapidly changing technological, economic, and business-related circumstances” and its preference to “decide no more than is necessary to resolve this case.” The choice frustrated the dissent and left one of the biggest questions in software law unanswered, but it also allowed the Court to resolve the immediate dispute without unsettling decades of industry practice.

How the Court Applied the Four Fair Use Factors

Fair use is a defense that permits limited copying of copyrighted material. Federal law directs courts to weigh four factors, and the majority found all four favored Google.⁴

Purpose and Character of the Use

The central question here is whether the new use is “transformative,” adding something new rather than substituting for the original. The Court found Google’s use highly transformative. Java was built for desktop and laptop computers. Google repurposed the declaring code for smartphones, creating a mobile platform that let developers apply existing skills to a new environment.¹

Nature of the Copyrighted Work

Copyright protects creative expression more strongly than functional material. The Court placed declaring code near the functional end of the spectrum. Its labels and structure are “inextricably bound up” with the uncopyrightable idea of organizing tasks into named categories so programmers can find them. Its value, the Court said, comes from being a system developers have learned rather than from creative flourish.¹

Amount and Substantiality of What Was Copied

Google copied 11,500 lines, or about 0.4% of the Java API. The Court stressed that Google took only the declaring code needed to let programmers work in the new environment and wrote its own implementing code, where the actual work happens. Copying only what was necessary to achieve a transformative purpose weighed in Google’s favor.¹

Market Effect

The final factor asks whether the copying damages the copyright holder’s market. The Court found Android was not a substitute for Java SE, and that Oracle had not successfully entered the smartphone market with Java when Google built Android. Enforcing the copyright, the majority reasoned, would create a “lock-in” effect, letting Oracle control where programmers could apply skills they had already learned and limiting new software development.¹

The Thomas Dissent

Justice Thomas, joined by Justice Alito, dissented sharply. Justice Barrett did not participate, producing the 6-2 split.¹

Thomas argued that Congress had explicitly protected computer programs in the Copyright Act, defining them as instructions used to bring about a result in a computer, and that declaring code fits that definition. He rejected the majority’s line between declaring and implementing code as “wholly inconsistent with the substantial protection Congress gave to computer code,” noting that declaring code is the part developers actually work with.

On market harm, Thomas pointed to specific numbers. Amazon paid Oracle to license Java for Kindle devices before Android launched, then used the free Android alternative as leverage to negotiate a 97.5% discount on its Oracle license. Samsung’s contract with Oracle reportedly fell from $40 million to about $1 million. Thomas wrote that Google “decimated Oracle’s market” and built a mobile operating system running on more than 2.5 billion devices while earning tens of billions of dollars a year.¹

Thomas also rejected the lock-in concern. Apple and Microsoft built successful mobile operating systems without copying Oracle’s declaring code, he noted, so alternatives clearly existed. He compared Google’s conduct to a Broadway theater copying a script from a smaller theater just to attract actors who had already memorized their lines. Actors’ investment in learning a script, he wrote, does not give a competitor the right to copy it.

What the Ruling Means for Software Developers

The decision matters well beyond the parties. Reimplementing APIs is a routine practice in software development. Programming languages, operating systems, and cloud platforms regularly reuse the labels and organizational structures of existing APIs so programs can communicate and developers can move their skills between platforms. A ruling for Oracle would have called all of that into question.

The majority acknowledged this directly, citing testimony that “shared interfaces are necessary for different programs to speak to each other,” that reimplementing interfaces “is common in the industry,” and that the practice “fueled widespread adoption of popular programming languages.”¹

The ruling is not a blank check. It rests on the specific facts the Court emphasized: Google copied only declaring code, wrote its own implementing code, applied it to a genuinely different platform, and took only what was needed to let programmers use existing knowledge. A company that copies more than necessary, or copies into a directly competing product, would face a much tougher fair use argument.

What Google v. Oracle Did Not Decide

By assuming copyrightability rather than deciding it, the Court left the deeper question open: can anyone own the organizational structure of an API? The district court said no. The Federal Circuit said yes. The Supreme Court said it did not need to answer. Justice Thomas warned that the majority’s fair use analysis was so broad that it was “difficult to imagine any circumstance in which declaring code will remain protected by copyright,” effectively resolving copyrightability through the back door.¹

The practical room for enforcing a copyright on API structure after this ruling is narrow. Narrow is not zero, though, and a future case with different facts could reach a different result. What developers have now is a Supreme Court holding that copying the functional labels of an API to build something new for a different platform can qualify as fair use, decided on the record in front of the Court and no broader than it needed to be.

1
Supreme Court of the United States. Google LLC v Oracle America Inc
2
Justia. Oracle America Inc v Google Inc, No 13-1021
3
Office of the Law Revision Counsel. 17 US Code 102 – Subject Matter of Copyright In General
4
Office of the Law Revision Counsel. 17 USC 107 – Limitations on Exclusive Rights Fair Use