New USPTO Director Immediately Signals Strong Patent-Friendly Shift in Software Patent Eligibility

The USPTO Appeals Review Panel, led by the newly confirmed Director of the US Patent and Trademark Office, has vacated a rejection based on alleged patent ineligibility (§101) in DeepMind’s U.S. Patent Application No. 16/319,040, which covers methods for training machine learning models to preserve performance across multiple tasks. In doing so, the Panel in Ex parte Desjardins recognized that the invention reflects a practical improvement to computer functionality, not merely an abstract algorithm. This decision marks a notable departure from prior Office practice and signals a more favorable climate for software and AI-based patent applications.

Strategy Impacts

For technology company executives, this shift represents a significantly broader opportunity to safeguard core software and algorithmic assets that might once have been unattainable on a startup budget. By securing expanded intellectual property protection, companies can not only defend their innovations against competitors but also establish clearer differentiation in crowded markets. Enhanced patent portfolios can support more robust licensing and cross-licensing strategies, open new revenue streams, and serve as a valuable bargaining chip in strategic partnerships. In addition, well-defined IP rights can materially increase enterprise valuation, whether in the context of raising venture capital, negotiating strategic investments, or positioning for acquisition.

For technology investors, the decision provides a clearer framework for risk mitigation and value creation. With patent assets more attainable, investors can deploy capital into AI- and software-driven ventures with greater confidence that the underlying innovations can be defended as proprietary assets rather than easily replicable features. This shift strengthens the monetization potential of portfolio companies, supports more predictable exit opportunities, and encourages deeper investment in cutting-edge technologies that rely heavily on proprietary algorithms and data-driven methods.

Proceed with Caution

However, caution is warranted. First, this shift in policy may only endure for the tenure of the current administration — applications filed a year or two from now under this administration may not be examined until the next administration, which could easily swing the pendulum back to be unfriendly to patent applicants. Second, while the USPTO Director’s guidance can influence how examiners approach subject‑matter eligibility, courts are not bound by the USPTO Director’s decisions, and judicial scrutiny under §101 remains rigorous. Judges apply their own interpretations of 35 U.S.C. §101, often with stricter standards than the agency, which means that a patent allowed by the USPTO may still be invalidated during litigation. The Federal Circuit’s decision in Rideshare v. Squires (also decided this week) illustrates this point: although the USPTO had allowed claims directed to a ride‑matching algorithm, the court invalidated the patent, finding that the claims were framed too broadly around the business goal of connecting drivers and riders rather than specific technical improvements that enabled the system to function. This underscores the gap that can exist between agency practice and judicial review, and the importance of drafting claims that emphasize concrete technological improvements.

Emerging Clarity on Software Patentability

At the same time, both the USPTO and the courts are showing signs of convergence around the 2016 federal appeals court decision in Enfish v. Microsoft as the leading precedent for software patent eligibility. Enfish established that software innovations can be patentable when they improve the functioning of technology (e.g., a computer) itself, rather than merely automating a business method using overall techniques that are well known. Recent USPTO guidance (including the Director’s decision in Ex parte Desjardins) cites Enfish repeatedly, and courts—including Rideshare—have used it as the benchmark for distinguishing between abstract business concepts and genuine technical solutions. This alignment suggests that Enfish has become the common language for analyzing software patentability, even if the USPTO and judiciary apply it with different emphases.

The Enfish case (and many other cases on patent eligibility) can be understood this way: solving a business problem alone—like matching riders with drivers—is not patentable, because business goals are considered abstract ideas. But if the invention solves a technical problem that makes the business solution possible—such as designing a new database structure that processes matches faster, or a network protocol that reduces latency—then the patent may be valid. In other words, the law rewards inventors who improve the “tools” (the technology) rather than just the “tasks” (the business objectives). When evaluating whether a software innovation is patentable, the key question is: does it make technology (such as a computer system) itself work better in a specific, demonstrable way? If so, it has a stronger chance of surviving both USPTO examination and judicial review.

Ultimately, these recent decisions and overall trends underscore the need for carefully engineered patent applications that emphasize concrete technical improvements and are resilient against both USPTO and court review. While the path forward is not without challenges, applicants who invest the effort to align their innovations with clear, demonstrable technological improvements stand an improved chance of securing durable protection. With thoughtful drafting and attention to converging standards, the immediate outlook is bright for innovators to build investable patent assets protecting the unique contributions that set them apart from the competition.

To explore how to strategically position your software innovations for both USPTO success and long-term enforceability, please contact a member of Synchrony IP.