Patent filings in AI reach record levels

Patent filings in AI reach record levels


6 min read

Filings rebound after 2008 crisis

Post the 2008 capital crisis, patent filings dipped in most industries, except Cleantech. Since then, a seven-year growth period led to record numbers, with about 650,000 applications filed yearly at the USPTO. The data shows businesses still view patenting as a worthwhile investment, despite strengthened trade secret protections and narrowed patent eligibility for computer-implemented inventions.

Top patent holders dominate the landscape

Established technology leaders hold a considerable share of patents, creating barriers for new competitors. Companies once indifferent or opposed to patents now see the benefits of a strong patent portfolio. Even those committed to open-source efforts invest in patent protection for competitive technologies.

The 2014 Supreme Court decision in Alice Corp. v. CLS Bank affected software-related invention patents. The two-part test led to high rejection rates for software-related patent applications. However, since 2016, there's been a significant increase in patents issued for bioinformatics and computational biology, with allowance rates exceeding pre-Alice levels. This trend is expected to continue and spread to other fields, like FinTech.

Growing foreign patent filings

Globalisation continues to impact patent applications, with the share of US entities decreasing as a percentage of total applications. The growth rate of foreign applicants is particularly high in AI, IoT, FinTech, Blockchain, and Automotive sectors. In Automotive, US-based applications made up less than 50% of applications since 2007. This shrinking share of US-originated applications could result in US-based innovators losing their home-field advantage over foreign competitors.
Foreign Applicants Increase Due to Offshore Outsourcing

US companies' offshore outsourcing of innovation activities might be causing the growing share of foreign applicants. MForesight's June 2018 report highlights the shift from "invent here, manufacture there" to "invent there, manufacture there." The report states that innovations often come from the factory floor, and moving manufacturing overseas fuels product and process innovations.

Innovation Hot Spots Across the US

California dominates patent applications, with 80,000 yearly since 2014. However, innovation hot spots are also found in states with smaller populations.

States with over 10,000 applications/year:

  • Illinois

  • Massachusetts

  • Michigan

  • New York

  • Texas

  • Washington

States with 5,000-10,000 applications/year:

  • Arizona

  • Colorado

  • Connecticut

  • Florida

  • Georgia

  • Minnesota

  • New Jersey

  • North Carolina

  • Ohio

  • Oregon

  • Pennsylvania

By analysing patent applications per capita, some states are as innovative as California, or even more so, like Massachusetts. On the other hand, populous states like Texas and Florida file fewer patent applications than expected based on population size.

AI innovation competition is global, with many countries recognising its national importance. The growing global competition for AI dominance is seen in the rising percentage of USPTO patent applications filed by foreign entities. AI enables machines to learn from experience and perform human-like tasks. Governments, including the US, have set national agendas to win the AI race.

To better understand AI patenting trends, we defined the following technology clusters within the AI industry:

  • Artificial Intelligence Systems

    • Machine Learning Systems

      • Utilising Classifiers

      • Utilising Clustering

      • Utilising Neural Networks

      • Utilising Pattern Recognition

      • Utilising Predictive Analysis

      • Utilising Support Vector Machines

    • Knowledge Based Systems

  • Application Areas/Industry

    • Computer Vision

    • Entertainment

    • Robotics

    • Healthcare

    • Natural Language Processing

    • FinTech

    • Digital Marketing

    • Automotive

    • Education

    • Cognitive Security

    • Augmented/Virtual Reality

Top AI innovations focus on Pattern Recognition, Predictive Analysis, and Neural Networks. IBM, Microsoft, and Google are the top filers of US patent applications in these areas.

AI Patent Landscape: IBM Dominates, Small Companies Show Strength

Artificial Intelligence

  • Facebook, Google, IBM, Microsoft, Samsung

    Level I

    • Application Area/Industry: Facebook, Google, IBM, Microsoft, Samsung

    • Artificial Intelligence Systems: Google, IBM, Intel, Microsoft, Samsung

      Level II

      • Application Areas/Industry

        • Augmented/Virtual Reality: Facebook, Google, Magic Leap, Microsoft, Sony Corp.

        • Automotive: Ford, General Motors, Hyundai, IBM, Toyota

        • Cognitive Security: Cisco, Cylance, Google, General Electric, IBM, Intel, Microsoft, NEC Corporation, Qualcomm, Symantec Corporation

        • Computer Vision: Adobe Systems, Facebook, Google, IBM, Microsoft, Samsung, Siemens

        • Digital Marketing: Adobe Systems, Facebook, Google, IBM, Microsoft

        • Education: Fanuc Corp., Ford, Fujitsu, IBM, Intuit, Microsoft, Oath, Tata Consultancy Services

        • Entertainment: Facebook, Google, IBM, Microsoft, Rom Guides

        • FinTech: Cognitive Scale, Facebook, Google, IBM, Microsoft

        • Healthcare: General Electric, IBM, Microsoft, Samsung, Siemens

        • Natural Language Processing: Facebook, Google, IBM, Microsoft, Samsung

        • Robotics: Fanuc, Google, IBM, Microsoft, Samsung

      • Artificial Intelligence Systems

        • Knowledge Based Systems: Facebook, Google, IBM, Microsoft, Samsung

        • Machine Learning Systems: Google, IBM, Intel, Microsoft, Samsung

IBM's dominance in AI patent applications is striking. This aligns with IBM's focus on using its AI platform, Watson, to offer AI-related solutions for various business problems and its embrace of the patent system.

Smaller Entities Shine, Global AI Competition Heats Up

Several smaller companies have captured top patent-filing spots in specific application clusters. For example, Magic Leap, a start-up developing a head-mounted virtual reality retinal display, files numerous patent applications in the Augmented/Virtual Reality cluster. Cognitive Scale builds industry-specific augmented intelligence solutions, focusing on the financial services industry. Its patent filings in the FinTech cluster reflect this focus.

AI innovation competition is global, and many countries recognise its national importance. Over 20 countries have released AI strategy documents, outlining policies to foster AI innovation and adoption for their economies and social welfare, addressing privacy, security, and ethics.

The rising percentage of patent applications filed in the USPTO by foreign entities shows increasing global competition for AI dominance. Although filings from China remain a small percentage, the rapid pace of application filing is notable. Patent applications filed by Chinese entities rose over sixfold between 2012 and 2018. This acceleration may continue as China heavily invests to become the world leader in AI by 2030.

Considering the emphasis other countries place on winning the AI innovation race, and the declining percentage of AI-related patent applications filed in the USPTO by American entities, American innovators may find themselves unable to compete in their own backyard.

Patent filings are increasing across all reviewed AI techniques. However, when data is tagged based on the AI application type, a different picture emerges. Filings have dropped over the last year for AI applications traditionally associated with business-method classifications at the patent office (e.g., FinTech, Education, and Entertainment), contrasting with the overall increase across other applications.

This trend resembles other software-related innovations, for which filings have significantly dropped since the Supreme Court's Alice decision, curtailing the scope of software-related patentable subject matter.

Although AI-related innovations often involve software, their allowance rate differs strikingly from similar innovations in FinTech. This divergence in allowance rates is tied to the Art Units examining these industries' respective patent applications. While FinTech applications are often examined in business method Art Units, AI patent applications are examined in Art Unit 2122 (AI & Simulation/Modelling), where the allowance rate is nearly double that of the business method Art Units.