Generative AI: Emerging IP Spotlight
Updated: Jan 3
Happy New Year, and welcome to the inaugural Emerging IP Spotlight. At Tradespace, we have the privilege of working with the largest, most R&D-intensive companies in the world. Our position at the intersection of IP generation and technology scouting gives us a unique window into emerging technologies. Our newsletter aims to provide investors, policy-makers, and operators with more efficient ways to find and commercialize emerging technology
Have a technology angle for us? Let us know.at firstname.lastname@example.org
Meet the Startups Operationalizing Generative AI
ChatGPT3, the latest iteration of OpenAI's chatbot, has rapidly brought Generative AI into the mainstream. So far much of the discussion has focused on its novelty or the near-term implications of GPT3 on things like secondary education.
However, our IP analysis reveals a growing cohort of startups already applying Generative AI to applications with the potential for much longer-term disruption. Below, we highlight 10 startups, each with a growing portfolio of IP covering a different Generative AI application:
Representative Startups using Generative AI for Different Applications
Western Companies in the Lead...For Now
Traditional Tech Giants like IBM, Google, Microsoft, and Intel accounted for much of the early IP generation in the space; however, industrial conglomerates such as Siemens, Bosch, and GE also made early investments
Chinese Companies - especially Tencent but also Baidu and Ping An - have more nascent portfolios that are beginning to rival or exceed their western peers
Also of Note: While Samsung is the only Korean company to break our top 10, there is an increasingly vibrant Korean Generative AI ecosystem that includes Seoul University, Lunit, 42Maru, AIZen, Deeping Source, and many others.
Role Reversal: Research Labs lag Companies in IP Generation
Large Companies Driving Early-Stage Generative AI Patenting: While research institutions undoubtably played a role in the emergence of Generative AI, large companies like IBM and Google accounted for 90% of the foundational IP generation.
Research Labs Starting to Catch Up: Research labs' share of IP generation has doubled in the last two years and now accounts for 24% of US IP generation.
One Theory: The initial infrastructure and resource requirements to develop and train Generative AI models were so high that only the largest, wealthiest companies could participate. Now, these same companies are making their models and toolkits available as a service, making it easier for researchers to access for their own projects.
Still, research lab IP generation remains relatively low for an emerging technology area, especially for US Government Labs:
A Final Note on AI Patents
Patent data is an imperfect lens through which to observe the world of AI. The standard for what AI innovations can be patented is notoriously unclear, and many companies elect to keep their AI as trade secrets. However, certain aspects of AI can and often are patented, providing us with key insights into emerging AI trends.
For example, data input arrangements and transceiver/receiver structure, such as how sensors and detectors are linked to networks, or how the networks collaborate, are often patented. More importantly, companies frequently patent the ways in which generative models can be applied to specific uses (e.g., telehealth chatbots), which gives us critical visibility into the industries that are adopting generative AI.
Was this analysis helpful? Sign Up Here to get an Emerging IP Spotlight for a different technology every other week