69% or more of AI deals since 2012 have gone to startups still in the early stages.
We used patent application activity to gauge the internal R&D efforts of startups on our AI list. We analyzed over 1,150 active AI startups (that have not exited) and found that only 21% of the startups applied for a patent since 2009 (note: there is a lag in publishing applications, so patent data from 2014 to present is still processing). Only 11% have been granted at least one patent in the United States since 2009.
The data is as of 4/17/17.
The reason for the light patent activity could be that most companies are still in the early stages of development (69% or more of AI deals since 2012 have gone to startups still in the early stages), or that most startups are currently leveraging open-source AI libraries like TensorFlow, or are simply not using patent protection for their algorithms or techniques. And, of course, the most skeptical view is that many of these AI companies are not really AI at all.
Nonetheless, the AI startups highlighted below are some of the top patents holders in the field. A knowledge of their patent portfolios is useful when evaluating AI companies, in order to determine technological IP and defensibility. Additionally, startups with patents are tagged in the public Artificial Intelligence Collection here.