Artificial intelligence is reshaping the world of live sports, from player performance and broadcast production to fan engagement and content creation. Mark Venables explores how leading organisations use AI to transform the game on and off the field.
The pace and spectacle of live sport has long been its defining feature. But in a world where attention is fragmented and content is infinite, even sport must evolve. Today, that evolution is increasingly driven by artificial intelligence (AI). AI is transforming operations behind the scenes and reshaping how fans consume games, how teams prepare and perform, and how broadcasters deliver experiences that resonate far beyond the final whistle.
At the heart of this transformation lies a convergence of technologies – vision language models, machine learning, and advanced tracking systems – that make sport more searchable, personalised, and interactive. The challenge is not simply technological but cultural. And for those leading the change, the opportunity lies not in replacing the spectacle of sport but in enriching it.
Making every frame count
Live sport is a uniquely visual medium, but the video presents a formidable technical challenge. Every match involves dozens of camera feeds, high-resolution, and unpredictable, fast-moving action. Until recently, identifying where and how to apply AI within this workflow has been a daunting prospect.
“Sport is an inherently visual medium,” according to Marta Mrak, Principal Engineer at Sky. “High-quality video requires high resolution, bit depth, and frame rates across multiple camera feeds. This creates a massive bandwidth challenge for live broadcasting, which has made the integration of AI into live workflows extremely difficult until recently.”
That is now changing. Mrak describes a pivotal moment made possible by vision language models (VLMs) optimised for GPUs. “These models can understand video at a deep level, interpreting pixels and recognising actions without being trained for every specific object. We can now analyse many frames per second in real-time and create latent space representations that unlock a wide range of applications.”
Those applications are not limited to summarisation and search. “When video and text exist in the same latent space, we can retrieve archival footage as easily as current content,” she adds. “VLMs allow AI to uncover these moments visually and present them to fans instantly. This fundamentally changes how we connect with history and storytelling in sport.”
Data points and decision moments
In the background, data is being collected at an unprecedented scale. Every movement, pass, and shot creates a trail of information AI can analyse to produce statistical models, predictive insights, or real-time commentary prompts. However, data alone is not the goal; understanding and actionability are.
Javier Gil Fernandez, Head of AI at La Liga, explains the scale involved. “We have 16 tracking cameras installed in the 42 stadiums in our first and second divisions. These generate over three million data points per match. We use machine learning algorithms to analyse this data and produce predictive and statistical insights.”
Such data has become central to La Liga’s ‘Beyond Stats’ project, which Microsoft and Sportian delivered. It also feeds tools like Media Coach, a tactical platform freely available to all clubs. “It helps coaches and teams prepare tactics and game strategies, Fernandez explains. “This technology is one reason Spanish teams remain competitive in European competitions.”
For the National Hockey League (NHL), the ambition goes further. “Our scoring and tracking systems collect massive amounts of data, from infrared-based tracking tags in players’ jerseys and embedded in the puck, to optical tracking of skeletal movements and stick handling,” Dave Lehanski, EVP of Business Development and Innovation, says. “This has opened up new possibilities, including creating virtual content on platforms like Roblox, where we present live data as fully synthetic games, independent of any video footage.”
Yet the real innovation may be in how that data is delivered. “If a commentator wants to reference the last time an American-born rookie scored seven goals in a period, that answer currently takes too long to surface,” Lehanski continues. “By the time a request reaches a producer and returns, the moment is gone. If we can automate that and deliver real-time insights to commentators, it keeps the human creativity intact while significantly increasing engagement and efficiency.”
Redefining roles and responsibilities
While some industries fear job displacement from AI, sport remains a curious outlier. Players cannot be replaced, coaches are still critical, and the storytelling that surrounds the game is becoming more, not less, valuable.
“We are not competing against each other; we are competing for attention against Netflix, Hulu and HBO,” Fernandez says. “To succeed, we must build a collaborative community that shares knowledge and strengthens our position as an industry.”
Internally, that begins with culture. “Our primary objective right now is internal adoption,” he adds. We want to equip employees with the tools and mindset needed to work with AI. It is not just about tools because those will change. It is about changing workflows and daily practices, which have transformed significantly over the past three years.”
La Liga has created an internal AI department that reports directly to the President, reflecting the strategic weight of its mission. “We recently launched an internal AI adoption programme, which includes training, but more importantly fosters a culture of discussion and collaboration around AI,” Fernandez continues.
Lehanski echoes that sentiment. “We start with use cases,” he says. “It is easy to become enamoured with AI’s capabilities, but we must first identify our core goals and determine how AI can help us achieve them. Internally, we are going through a structured process of education, idea collection and testing to ensure everyone understands AI’s capabilities. Once they do, interest grows rapidly.”
What emerges is a reconfiguration of roles, not their removal. “One of the immediate impacts of AI is enabling many more people to produce and create content,” he adds. “Far from displacing jobs, we are putting powerful tools into the hands of those with a deep understanding of the sport, giving them the ability to produce their unique content.”
The new frontier of fan engagement
As AI unlocks the past and accelerates the present, it also reshapes the future of audience interaction. From personalised content to synthetic broadcasts, fans are no longer passive observers; they are participants in a dynamic ecosystem. “Two key themes for us are personalisation, first by audience segment and eventually at the individual level, and real-time delivery,” Lehanski says. Hockey is a continuous-play sport with few stoppages, so any content or insights we deliver must happen in real time, without disrupting the viewing experience.”
Gil Fernandez describes similar ambitions. “We are exploring many use cases. Internally, we use chatbots and avatarisation tools,” he says. “We created an AI influencer, Alex, who presents a weekly podcast on technology and artificial intelligence. It may sound playful, but it is an effective way of engaging audiences with serious content.”
Emerging platforms are also being considered. “We bring startups into our ecosystem and our clubs,” he continues. La Liga comprises 42 clubs across the top two divisions, and while we oversee the ecosystem, the clubs implement the solutions. Technology hubs across Spain—like Real Madrid’s, Barcelona’s, and initiatives like Madrid’s Delta—play a crucial role.”
Guardrails and global growth
The acceleration of AI in sports brings not just innovation but new risks. Piracy, legal compliance, and regulatory scrutiny all shape what can be done and how. “Piracy is a major issue across the sports industry,” Fernandez says. “We need to unite and leverage AI to combat it. While we may not eliminate piracy, we can reduce its impact through intelligent detection and enforcement mechanisms.”
Lehanski believes that compliance is equally pressing. “As an IP-driven business, we must control how our content is used,” he explains. “We work closely with our legal team to ensure we protect our rights while enabling creativity. It is a difficult but critical balance.”
Complexity is also added for those organisations operating in Europe. “Being based in the European Union introduces further regulatory considerations that could potentially slow adoption,” Fernandez adds. “Fortunately, our President is highly knowledgeable about AI, which gives us a head start.”
Technical considerations matter, too. A major technical challenge is deciding which AI models to use, especially given how quickly they evolve. “We cannot constantly reprocess massive video archives with new models,” Mrak says “This makes model selection and data compatibility essential. There is a need for more open models and software to ensure sustainability and interoperability across AI systems.”
Despite the challenges, the long-term vision is clear. As Mrak puts it, “The hardware is ready. Modern GPUs can handle demanding streams while extracting AI-ready embeddings. Now, we must ensure data – video, audio, metadata – is AI-compatible. We must become AI-ready to unlock the full potential of these tools.”
For Lehanski, the goal is scale. “Content is the foundation of everything,” he adds. “The ability to rapidly create personalised, localised, and multi-format content will be the key to global growth. Our arenas are no longer just performance spaces; they are content studios. We scale engagement beyond the live audience by capturing and distributing that content effectively. The potential is extraordinary.”
In the end, it may not be the technology that determines success but the mindset. “The most important thing is changing mindsets,” Fernandez concludes. The tools will evolve, but a culture of internal adoption and AI fluency will give the industry the foundation to move forward confidently.”
