Enfish v. Microsoft is the 2016 Federal Circuit decision holding that software claims can be patent-eligible when they are directed to a specific improvement in the way a computer operates, rather than to an abstract idea that merely uses a computer as a tool. Decided on May 12, 2016, the case gave software patent owners their first strong post-Alice path to defeating an eligibility challenge at Step One of the Supreme Court’s two-step test. Enfish won the eligibility fight but lost on infringement: the Federal Circuit affirmed that Microsoft’s ADO.NET technology did not infringe the asserted claims, so the ruling stands as a landmark on Section 101 law rather than a damages victory.
The Patents at Issue
Enfish held U.S. Patent Nos. 6,151,604 and 6,163,775, both claiming priority to a March 1995 application. The patents described a database built around a “self-referential” table. Conventional relational databases scatter data across multiple tables linked by shared columns, with structures that have to be defined up front and are difficult to change. The Enfish design put all data entities in a single table whose columns were themselves defined by rows within that same table. Because the table described its own structure, it could be reconfigured on the fly.
The specification identified three concrete benefits over conventional relational databases: faster searching through a new indexing technique, better storage of non-text data like images, and more flexible configuration without extensive upfront modeling. Those technical details did the heavy lifting when the case reached the Federal Circuit.
Enfish sued Microsoft, alleging that ADO.NET infringed five claims across the two patents. The district court granted summary judgment to Microsoft on every issue: it held the claims patent-ineligible as directed to the abstract idea of “storing, organizing, and retrieving memory in a logical table,” found some claims anticipated under 35 U.S.C. § 102, and found one claim not infringed. On appeal, the Federal Circuit reversed the eligibility ruling, vacated the anticipation finding, and affirmed non-infringement as to ADO.NET.
The Alice/Mayo Framework the Court Was Working With
Section 101 of the Patent Act allows patents on any “new and useful process, machine, manufacture, or composition of matter.” The Supreme Court has read that grant to exclude laws of nature, natural phenomena, and abstract ideas. Two decisions built the current test. Mayo Collaborative Services v. Prometheus Laboratories (2012) struck down a medical diagnostic patent for adding nothing meaningful to a natural law about drug metabolism. Alice Corp. v. CLS Bank International (2014) applied the same reasoning to software, invalidating claims on a computerized escrow system as the abstract idea of intermediated settlement run on a generic computer.
The resulting two-step test asks:
- Step One: Is the claim directed to a law of nature, natural phenomenon, or abstract idea? If not, it is eligible and the analysis ends.
- Step Two: If it is, does the claim contain an “inventive concept” that makes it significantly more than the excluded idea itself?
After Alice, district courts invalidated software patents at high rates, often by characterizing any software function at a high enough level of generality to make it sound abstract at Step One, then finding nothing inventive at Step Two. That was the environment Enfish’s appeal entered.
What the Federal Circuit Held
The Federal Circuit reframed Step One as a real inquiry rather than a rubber stamp. The question, the court said, is whether the claims focus on a specific improvement in computer capabilities or on a process that is itself abstract and merely happens to run on a computer. Courts, the panel warned, should not describe claims at such a high level of abstraction that any software claim becomes ineligible.
Applied to Enfish’s patents, that inquiry came out in the patent owner’s favor. The claims, the court held, were “directed to a specific improvement to the way computers operate, embodied in the self-referential table.” The specification’s detailed contrast with prior relational databases mattered: it described how the old approach worked, why it was limited, and how the self-referential model solved those limitations. The court pointed to “the specification’s disparagement of conventional data structures, and the clear focus of the claims on the benefits of the disclosed self-referential model,” as indicators that the claims targeted an improvement in computer functioning. Because the claims cleared Step One, the court never reached Step Two.
The Improvement-to-Computer-Functionality Distinction
Enfish’s most durable contribution is the line it drew between two kinds of software claims. Claims that improve how the computer itself works — a new data structure that speeds up searching, a memory management technique that reduces storage, a networking protocol that moves data faster — target the machine’s own capabilities and are eligible at Step One. Claims that use a computer as a tool to perform some other task, such as automating a known business process or running a mathematical formula on generic hardware, invoke the computer merely as a means to an end and are likely abstract.
The distinction sounds clean, but applying it depends on what the specification says. The Federal Circuit looked for a described technical problem in the prior art, an explanation of why existing approaches fell short, and identified technical benefits of the claimed solution. Vague claims of “improved efficiency” do not qualify. The specification has to show the mechanism by which the software changes what the computer can do or how well it does it. Quantitative benchmarks are not required; qualitative improvements described with enough technical detail to show they are real rather than conclusory can be enough.
How the USPTO Applied Enfish
The USPTO responded within days. A memorandum to examiners stated that “a claim directed to an improvement to computer-related technology (e.g., computer functionality) is likely not similar to claims that have previously been identified as abstract by the courts,” and told examiners they could find a claim eligible at Step One without reaching Step Two.
The reasoning was later folded into the 2019 Revised Patent Subject Matter Eligibility Guidance, which restructured Step One into two prongs. Under Prong Two of the revised Step 2A, examiners assess whether a claim reciting an abstract idea nonetheless integrates it into a “practical application.” One listed consideration is whether the claim “reflects an improvement in the functioning of a computer, or an improvement to other technology or technical field.” The guidance cites Enfish by name. The current framework sits in Sections 2103 through 2106.07 of the Manual of Patent Examining Procedure, which directs examiners to “determine whether the claim purports to improve computer capabilities or, instead, invokes computers merely as a tool.”
What Enfish Means for Drafting Software Patents
For anyone writing a software patent application, Enfish points to the specification as the place most of the work happens. Identify a specific technical problem with existing computer systems. Explain why current approaches do not solve it well. Describe how the software changes the computer’s operation to fix the problem. Generic language about efficiency or automation will draw a Section 101 rejection. Concrete descriptions of data structures, processing steps that differ from conventional methods, or interactions with hardware are what move claims into eligible territory.
Claim drafting matters too. Claims that recite the technical architecture of the solution — data structures, processing steps, component interactions — hold up better than claims written at a high level of generality. The Enfish claims specified a logical table with rows defining column attributes in the same table, not just “a method of organizing data.” That specificity carried the decision.
Section 101 is only the first hurdle. Claims still need to satisfy novelty under Section 102, non-obviousness under Section 103, and the written description and enablement requirements of Section 112. Enfish does not guarantee a patent; it reopens a door that Alice had come close to shutting for software inventions that genuinely change what a computer can do.
Pending Legislation That Could Change the Analysis
The Alice/Mayo framework has drawn steady criticism for giving courts wide discretion to characterize claims at whatever level of abstraction produces the result they want. In May 2025, Congress introduced the Patent Eligibility Restoration Act of 2025 (H.R. 3152 in the House, S. 1546 in the Senate), which would narrow the categories of ineligible subject matter to specific exclusions such as mathematical formulas that are not part of a useful process, mental processes performed solely in the human mind, and unmodified human genes as they exist in the body. As of early 2026, both bills remain in the Judiciary Committee. If enacted, the statute would reshape the eligibility analysis Enfish addressed, though the case’s emphasis on technical specificity in the specification would remain sound practice under any framework.