Generative AI may create a boom in data licensing deals to train large language models (LLM).
While enterprises are training LLMs and looking for competitive advantage, new revenue streams are going to appear in surprising places. And the fun is just beginning.
The Associated Press has partnered with OpenAI to license its archive dating back to 1985 for two years. In return, AP gets AI expertise and technology. In addition, Shutterstock said it will license its archive to OpenAI to train models. Google has partnered with Adobe Firefly for its Bard generative AI platform, which has just been updated so it can handle more than 40 languages.
Add it up and the breakneck pace of generative AI is going to create a data licensing boom that's just getting started. Today, these deals are focused on consumer applications as image and text archives are licensed. Don't be surprised if media and movie studios ultimately find new revenue streams for archives.
But the real money is going to be made for specialized LLM usage. Consider the following possibilities.
- Financial data providers such as Standard and Poor's and Bloomberg could either use their data to train proprietary LLMs or license datasets. Reuters could bundle financial data and news archives.
- Industries such as oil and gas could train LLMs with first party data. Once oil giants leverage their own first party data, they could license a more generic version.
- Software providers could license anonymized data across their platforms for LLM training. Think retail, healthcare, and telecom.
- Process mining companies could license industry process data for LLM training.
- A company like Sears Home Services is sitting on decades of appliance repair data that can be licensed.
- Companies that monitor conditions on Earth and track climate can license data for generative AI applications.
The possibilities here are endless and any company that doesn't want to train its own models will license data sets. Enterprises have been collecting data for years and now there's an avenue to monetize what they've collected.
A look at some enterprises and what they're doing with AI.