Is the sports industry sitting on an AI licensing goldmine?

While industries like media are already pioneering AI licensing deals, are data-rich sports organisations lagging behind? 

The Value in AI Licensing 

Last year, it was reported that Reddit signed an estimated $60 million annual deal with Google to provide access to Reddit’s vast database of content and data, for the purposes of training Google’s AI models. 

Off the back of this, an interesting report by sports business commentator Aaron Miller estimated that, by his calculations, a theoretical AI data licensing deal with the NBA could fetch around $87 million annually, rivalling the league’s blockbuster partnerships such as its $125 million Nike deal.

Major sports leagues like the English Premier League and the clubs that compete in them create vast amounts of potentially lucrative data. Each club could play 50+ games per season, as well as generating training and performance analytics behind closed doors. Other data-rich sports like Formula 1, awash with vast quantities of technical and performance data, are generating more data than ever. Plus, the advent of relatively low cost SaaS platforms and wearables has seen traditionally less wealthy sports beginning to accrue their own potentially valuable datasets. These organisations should be ideal candidates for AI developers seeking datasets to train models and develop sports technology applications, such as tactic prediction models (e.g. Google DeepMind’s TacticAI), performance analysis tools or talent scouting tools. AI can even be combined with virtual reality to simulate low-intensity virtual training for athletes, showcased by NFL Washington Commanders quarterback Jayden Daniels. 

Potential Challenges 

Despite exciting potential, monetising data in sport isn’t straightforward. Challenges include:

• Ownership and Licensing rights 

Ownership (or more accurately contractual and commercial “control” of data generated) is likely to be knotty – with third party tech providers (e.g. Opta), leagues, clubs and even the players themselves potentially having an interest in, or control over, the data and how it may be used or exploited. Negotiating and agreeing the terms on which such data may be collectively licensed to and exploited by a third party for AI training purposes with vendors (and the parameters on any such use), is likely to be a significant challenge. 

• Confidentiality 

Training data, unaggregated performance metrics, and tactical analytics are valuable but confidential assets. Sports organisations may hesitate to expose such granular details to third parties, fearing competitive disadvantages if certain information was leaked. The appetite to take risks with confidential data may also vary between sports. Formula 1 races for example, can be decided by the smallest of margins, with highly technical performance data being potentially crucial, so teams will be reluctant to share any form of data with third party vendors.

• Player Rights 

The commercial use of player personal data, including biometric and performance statistics has already sparked controversy. “Project Red Card” for example saw hundreds of professional athletes in the UK claiming organisations have been exploiting their personal data without consent or compensation. The same principles could apply to player data used in datasets licensed to train AI models.

• Data Quality 

Some sports organisations, leagues and clubs may struggle to consolidate and manage vast quantities of data they already have or generate, making it difficult to simply ‘plug in’ a dataset. It is likely that many of these organisations may not yet have a handle on the data they generate or a co-ordinated data governance programme in place to enable them to exploit it. Many may need to dedicate resources to collating, curating and fine-tuning datasets before they could be licensed to train AI models, something that might not be at the top of the priority list for certain organisations, particularly those with tight budgets.

Opportunities 

As the AI data licensing market continues to grow, the sports industry (given the wealth of data it creates) is clearly well-positioned to explore its potential. Similar to the likes of Reddit in the media industry, sports organisations such as the Premier League could unlock a potentially lucrative revenue stream and set the benchmarks for dataset licensing across professional sports. However, careful navigation of ownership rights, confidentiality, player consent and data governance would be essential.

Thoughts for Rights Holders or AI Developers

If you would like advice on licensing content for AI training, contracting for AI tools, the EU AI Act or anything AI-related, please reach out to your usual Bristows contact or get in touch here. 

Additionally, for an overview of the key areas to consider when licensing content to train AI, see our recent article here.