How the US Open Keeps Upping Its Tech Game

More than one million fans will pass through the gates of the USTA Billy Jean King National Tennis Center this year, grabbing food, honey deuces, and packing into Arthur Ashe Stadium under the lights. But none of them will ever step into the room actually running the tournament. Inside the biggest stadium in tennis, servers and engineers are quietly processing every match, every point, in real time. We are in the IBM Data Operations room, and the people in this room are responsible for collecting data on all 17 courts that have matches happening right now. In this episode of NYC Innovates, we'll take an exclusive look at how the U.S. Open powers their unique digital experience for over 14 million people, break down the AI behind brand new features like serve quality and key moments, and whether you're watching at home or catching the action in person. We'll discover what's next for the future of the fan experience. Stay with us. For three weeks every August into September, Flushing, Queens becomes the epicenter of global tennis. The USTA Billy Jean King National Tennis Center, headlined by the 24,000-seat Arthur Ashe Stadium, hosts one of the most attended annual sporting events in the world. Beyond the tennis center, more than 14 million unique users engage through the tournament's app and website during the 395 main draw matches. But long before today's computational power, the US Open had already built a reputation as tennis's most willing innovator. In 1973, the tournament reshaped the sport by switching to the now iconic yellow tennis ball, adapted for one simple reason. It showed up better for television. In 1975, the US Open lit up for tennis's first ever night session, turning a daytime sport into a primetime broadcast, and setting the template other majors would eventually copy. In 2006, it became the first Grand Slam to let players challenge line calls electronically, putting instant replay in players' hands for the first time with the Hawkeye system. It's the same philosophy that still drives the tournament's approach to technology today. The story of the modern US Open isn't just what happens on the court, it's now what happens in the data behind it. Monica Ellingson leads IBM's sports and entertainment practice, the team responsible for getting all of that raw data from the court and into the hands of tennis fans. And it starts right here, inside IBM's operations center, tucked deep inside Arthur Ashe Stadium. These are basically what all of the components that we need to monitor the video that's being collected courtside, as well as scores, stats. We're also collecting biometric data and ball motion data that we're using to create a new stat called serve quality this year. This is the same technology that we're using for automated line calling on the courts, as well as serve speed. Those are the same cameras that we're using to collect biometric data. We also have statisticians along the court, and then even the chair umpire who is sitting courtside has a tablet that they are using to collect scores and vital statistics used to power all of our apps. All of that raw information funnels into the app's home screen, a feature called Live Updates. Live Updates is a personalized AI curated experience, and to put this in perspective, it is driving about 80% of the content within the at-home experience of the app. It adapts as a match unfolds, previews before it starts, live insights while it's happening, and a wrap-up once it's over. But the headline addition this year is something entirely new, serve quality. We identify 21 individual physical components of the serve, seven racket positions, and this is about 15,000 data points on average for a single rally in a singles match, and that's about 4.6 million data points per match on average. We used IBM Bob, which is a code generation, AI code generation tool to accelerate our development. So just to put this in perspective, we created a model of the biomechanics behind a serve within a couple of days, and then we expanded upon that and said, let's create a 3D visualization of the biomechanics of the serve, and it shows the court, the player, the ball. But then really, a lot of data doesn't really give you any insights, right? So we then trained that model by feeding into it academic papers, research from industry experts about what are the ideal physical layout of the body as well as kinetics behind a serve, and that helped us to identify the ideal state for each criteria of each phase of the serve. We then married that with an effectiveness score, so whether the serve was in or out or the outcome of that serve. Effectiveness plus efficiency then comes together into the serve quality score. We'll be looking at, you know, are the players going to want to see that data? You know, what is Wimbledon interested in that data? How can we evolve that and make it into new, enhanced fan experiences? So we're really just scratching the surface. The app has tracked live win probability for a few years now, and this year IBM added the why behind it. We've calculated using AI the likelihood to win of a matchup. When you look at this, you know, you're able to see the momentum of the match, the ebbs and flows, but then you might go back to a single moment and say, well, what happened here? So now we're adding that storytelling component by adding in key moments, which is an AI-generated summary of what actually occurred at that particular time. Here at the U.S. Open, we have access to, you know, 20 years of historical data, scores, stats, etc., and we also have access to the opinions of industry experts, punditry, and we're feeding all of that into these equations, you know, to be able to help us determine what really defines a momentum shift. And then there's match chat, a feature Ellingson describes less like software and more like a friend. The way I think about it is it's like, you know, that friend that knows everything about sports and, you know, having them sitting next to you, you could say, when was the last time these two players met up? You know, it's hard sometimes to find that specific information, but it's a companion that really allows you to ask just about any question about the match that's happening, and it's right there within a matter of seconds. Match chat doesn't run on a single AI. It runs on a team of them, all coordinated by IBM's Watson X Orchestry. It's why this year match chat can now answer with more than just text. We've always had a vision, for example, to incorporate media within match chat, so not only, you know, be able to ask your question, but for example, if there were a video highlight or a New research released by IBM this year found that 64% of global sports fans say they trust AI-generated sports content. That leaves more than a third who don't. A gap, Ellingson says, mostly comes down to awareness, not resistance. I think the people who don't don't realize that what they're seeing is AI-generated content. The reality is that you cannot deliver the level of personalized content without AI. With the volume of content that we're generating, that we're analyzing and transforming without AI, we just couldn't do it at scale. Over 14 million people. How do you personalize to that many people? This all lives within a multi-hybrid cloud environment, and this environment enables us to scale up to meet the high volumes of traffic. I mean, we're talking about going from very, very low traffic to within, you know, one moment, everything comes live and we have to scale up. There's a whole gamut of skill sets on my team. We cover from project managers, AI engineers, experts within the infrastructure space, iOS developers, Android developers, web developers, testers, the full gamut. And our goal is to be to talk to our clients, ideate with them, design it, develop it, support it, and host it. And that team doesn't build in a vacuum. Every year's roadmap starts as a conversation with the USTA itself. IBM has been the tournament's technology partner for three decades, reimagining how fans experience the sport year after year. It's very creative and collaborative. I mean, we really work together in terms of, you know, what's our vision for the next year? You know, what are the things that we want to improve, enhance, and what are the new experiences? Because we always have to have something brand new, you know, to further push the envelope of innovation. We're seeing even behavior shifting where, you know, people are using the app not as a second screen, but as a primary screen. And I'm talking about, you know, potentially, you know, the insights that we deliver within the app you cannot get within the broadcast. We certainly have some integrations, but, you know, I think that the behavior overall of fans is changing. You know, the appetite for data, the appetite for truly personalized content, and I think it's affecting the overall dynamic of how people are enjoying and following sports. Three weeks and more than a billion data points later, the US Open will crown its champions. And as this year's trophy is lifted, Ellingson's team will already be building what's next. I think it's just a matter of, you know, taking some of the insights as well as the advanced data that we have and seeing how can we tell better, more interesting stories? How can we package up media into that experience as well to make it even more immersive? So tell us, how do you envision the future of the digital fan experience? And what sports apps do you use the most? Let us know in the comments below. For Cheddar, I'm Chris Castellino.