The US Open, one of tennis's four majors, opened on August 23, 2026. This year the tournament's official site and app gain a new metric that puts a number on serve quality, along with a feature that surfaces the moments where a match turned. IBM and the United States Tennis Association (USTA) detailed the additions on August 24, and serve analysis alone will generate roughly 1.2 billion data points over the course of the event.
Twenty-one points on the body and racquet, tracked 50 times a second
The headline addition is Serve Quality, a score that rates how good an individual serve actually was. It covers all 254 singles matches. Limb-tracking technology captures 21 data points across the player's body and racquet at a rate of 50 times per second, fine-grained enough to register wrist flex and the way force generated in the legs travels up through the torso.
Those readings are organized around four aspects: efficiency, accuracy, consistency and ball toss. IBM Confluent handles the continuous stream of data coming off the court, and the resulting score reaches viewers in near real time. Serve speed has never been the whole story, and this gives that long-standing intuition a shared yardstick for the first time.
From who is ahead to why they got there
Likelihood to Win, already a fixture of the experience, expresses each player's chance of winning at any given moment. The calculation runs an AI-powered analysis over current and historical statistics, expert opinion and the momentum of the match.
Joining it this year is Key Moments, which collects the big swings and the turning points where momentum flipped. It also covers every singles match. Because the reasons behind movements in the win-probability graph can be read back afterward, anyone joining a match late can catch up on how it got to that point.
The site's front page has been rebuilt as well. The new Live Updates homepage lets fans pin their favorite players so that the matches and storylines they care about rise above the volume of tournament information.
A companion that answers in conversation, with photos and video
Match Chat has been reworked too. Fans can ask whatever comes to mind in their own words during a match and get a conversational answer on the spot. Some replies now arrive with relevant photos and video attached rather than text alone.
The foundation is IBM's watsonx Orchestrate. It combines several AI agents with fit-for-purpose models tuned for the task, trained on the USTA's editorial style and the vocabulary of tennis. Building it separately from a general-purpose chatbot was a deliberate choice, keeping the tone close to how the sport is actually covered. Its sources have been widened as well, so answers draw on live match data along with analysis and historical records.
The feature was built by IBM's own AI agent
One detail worth noting is that the limb-tracking work behind serve analysis was developed with IBM Bob, the company's AI coding agent, which is positioned to handle planning, implementation, testing and maintenance across the software lifecycle. It is a working example of IBM using its own AI development tooling to ship its own AI features.
Accuracy matters more than speed
IBM also released survey results the same day, conducted by Morning Consult on its behalf. Among tennis fans surveyed worldwide, 91 percent use sports apps during events, and 64 percent express high trust in AI-generated sports content. IBM reads the findings as evidence that the digital viewing experience is now judged on accuracy rather than speed.
IBM and the USTA have worked together since 1992, building digital experiences for more than 14 million tennis fans a year. Brian Ryerson, Senior Director of Digital Strategy at the USTA, said this year's work has changed what fan engagement can look like. Jonathan Adashek, Senior Vice President and Chief Global Affairs Officer at IBM, emphasized the ability to route millions of live data points from the court into workflows automatically and return usable insight within minutes to hours.
The 2026 US Open runs through September 13. All of these features are available on USOpen.org and in the official app, distributed through both the Apple and Android stores.
Summary
IBM and the USTA have rolled out three AI features for the 2026 US Open: Serve Quality, which analyzes serves across all 254 singles matches using 21 measurement points sampled 50 times per second; Key Moments, which explains why win probability shifted; and the conversational Match Chat, running on watsonx Orchestrate. IBM Bob was used to develop the limb-tracking component, so AI-built AI features are landing directly in fans' hands. How much of the underlying reasoning gets exposed remains an open question, but the direction here — sharpening the resolution of what viewers can see — is a sound one.
