Welcome to the HBR Executive Agenda for September 18, 2025. In This Issue: Should Your Company License Content to AI? How to Reinvent a Legacy Brand—and Win Should Your Company License Content to AI? Anthropic recently entered into a $1.5 billion settlement with book publishers and authors to resolve claims that it had illegally downloaded and trained its AI systems on millions of copyrighted books. The settlement has brought renewed attention to AI licensing deals—which is not just a question for media companies and publishers. There’s a rapidly expanding market for IP licensing, which means a wide range of companies with access to data can tap into this revenue stream. Two years ago, licensing deals with AI developers amounted to about $75 million, according to Dan Neely, cofounder and CEO of Vermillio, an AI rights management platform. This year the figure will be about $20 billion, he says, and next year it could be twice that. Most of the deals to date have involved large flows of content to train large AI models like Perplexity or ChatGPT, but that window is closing as the general learning for those models is nearly complete. The emerging market is for specialized AI models that need more-targeted data sets that will come from a broader universe of suppliers. Neely regularly fields requests from AI companies for very specific content needs, like photos of New Orleans before Hurricane Katrina devastated the city, or footage from 1990s American TV that could help train an AI agent to converse more fluently with middle-aged customers. There are countless other opportunities for companies with libraries of compelling content. Healthcare records, aggregated and anonymized, could revolutionize how AI identifies and suggests treatment for illness. Companies known for great customer service could license internal data that shows how they manage a continuous learning system. And in the field of advertising, companies could license performance data for their digital campaigns—what worked, what didn’t—to help optimize the agentic ad bots that are emerging in that field. Some of these new products will be subscription services that will pay royalties to the original licensors. So how do you know if your data is robust enough to license? Neely suggests taking the following steps: Evaluate the data you have. Would it potentially be valuable for an AI bot that is aggregating information from various sources? Financial companies, for example, have troves of analytic reports about equities that could help an investment bot find patterns and predictors that previously weren’t apparent. Lock down your ownership. Make sure you truly own the rights to your data. Book publishing, for example, can be complicated because both publishers and authors would need to approve any licensing agreement. In music, there are several royalty streams that would need to be considered. Consider getting help. It isn’t easy to do this on your own. There are many new companies (Vermillio among them) that can analyze and prepare your content for a potential deal. Among other things, this could entail stripping out from your data the content or images that you don’t have rights to. Set your terms. Pricing strategies can be determined by a range of factors, including the exclusivity of the data and the length of the license. You can license your content for a fixed period of time, as opposed to the one-and-done deals that we’ve seen to date. Many of these deals will be relatively small compared with the big transactions that have characterized the first phase of AI licensing. But as AI models launch new products and share their windfall, they could add up to big numbers. There are caveats, of course. You should avoid doing deals with platforms that don’t seem to have adequate guardrails for what they can do with your data. And you should hesitate to do deals that pay you only once for a license, as that could be giving away future upside as the technology evolves and creates new revenue opportunities. And of course, there are plenty of reasons to avoid this market altogether. Licensing your data to sloppy or unethical AI models could create security and privacy risks, as there have been cases of AI being used to de-anonymize data that had seemed safe. You might also feel your data and processes give you competitive advantage that you might not want to share via licensing. Keeping your data private could trump any possible revenue deal. AI licensing is not for everybody. But it’s becoming an opportunity to consider for far more players than just the media and entertainment giants. HBR Executive Live: How to Reinvent a Legacy Brand—and Win Join me for an exclusive Q&A with Lew Frankfort, the former CEO of Coach and author of the new book Bag Man, as he shares an inside perspective on the high-stakes art of reinventing a legacy brand. Frankfort guided the iconic brand from a $6 million niche leather goods manufacturer to a $5 billion global powerhouse—balancing tradition and innovation at every step. He’ll share how an obsession with consumer insights—born from his own haunting fear of failure—became the company’s secret weapon, and how Coach’s “magic plus logic” philosophy created the accessible luxury category that countless brands now chase. We hope you’ll join us for this rare opportunity to learn from a visionary who engineered the rebirth of a beloved legacy enterprise. September 24, 2025 11 am ET Register now
