IBM Puts Win Probabilities on Every Point at the US Open

IBM and the US Open are expanding their AI features for the 2026 tournament with personalized updates, an AI assistant and win probabilities that refresh after every point, serving a digital audience the tournament puts at more than 14 million users.

The US Open has become one of the most closely watched product demonstrations in enterprise software, and this year IBM is pushing further into the broadcast. The company and the tournament are expanding their use of artificial intelligence to deliver real-time match insights to fans, including personalized updates, an AI assistant and win probabilities that are recalculated after every single point.

IBM Senior Vice President Jonathan Adashek and former professional player Sam Querrey set out the plan on Bloomberg This Weekend with hosts David Gura and Christina Ruffini. Their pitch had two halves: a fan experience that keeps a very large digital audience inside the tournament's own apps and website, and a data layer that players and coaches can use to break down opponents and shape match strategy.

What the tournament is actually shipping

The headline feature is the point-by-point win probability. Tennis is unusually well suited to this: the sport is a sequence of discrete, scored events with a fixed structure, which means a model has a clean, high-frequency stream of state changes to react to. A break point saved, a double fault, a shift in first-serve percentage — each one moves the number. For a viewer, that turns an abstract sense of momentum into a figure that visibly ticks.

Alongside it sit personalized updates — the tournament pushing the matches, players and moments a given user has signalled interest in, rather than a single undifferentiated feed — and an AI assistant that fields questions in natural language. That last piece is the one with the clearest read-across to IBM's commercial business, because a conversational layer sitting on top of proprietary, fast-moving structured data is precisely the enterprise use case the company sells.

The audience figure is the part that gives the exercise scale. More than 14 million digital users follow the tournament, according to the details laid out in the Bloomberg Markets segment. That is a live population of people interacting with an AI product for two weeks under real conditions — unpredictable load, unpredictable questions, and no tolerance for latency during a fifth-set tiebreak.

Why a tennis tournament is a useful shop window

Sports sponsorships are usually judged on logo impressions. This one is judged differently. IBM is not renting attention at Flushing Meadows so much as running a public stress test of the same machinery it wants to sell to banks, insurers, retailers and manufacturers.

The argument to a corporate buyer is straightforward. If a system can ingest a continuously updating event stream, generate an inference against it in seconds, personalize the output across millions of concurrent users and survive a two-week peak without falling over, then the harder-to-demonstrate claims about reliability and throughput have already been made in public. Enterprise software buying is slow and reference-driven; a visible, functioning deployment shortens the conversation.

There is also a talent and perception dimension. IBM's brand problem for much of the past decade was not competence but relevance — the sense that the interesting AI work was happening elsewhere. A consumer-facing showcase that millions of people touch during a fortnight in New York addresses that in a way a white paper does not.

The player-side data is the less obvious half

Querrey's presence on the panel points at the second constituency. The same match data that produces a fan-facing probability number also produces a coaching artifact: serve patterns by court side, return depth, rally-length distributions, how an opponent's error rate behaves under pressure. Professional tennis has been assembling this material for years, but the constraint has usually been the analyst's time, not the data's availability.

An assistant that lets a coach ask a question in plain language and get an answer built from the tournament's own feed compresses that work. It also raises the familiar question of asymmetry: better tools tend to help the players and teams already able to pay for support staff, and the gap between a top-ten team and a qualifier is as much about analytical resource as it is about strokes.

Where the shares stand going in

IBM closed the most recent session at 235.68, up 0.85% from a prior close of 233.69, having traded between 233.12 and 238.72 on the day, as of 20:00 GMT on Friday, 21 August 2026. That was a firmer session than the broad market: the S&P 500 tracker (NYSEARCA: SPY) finished at $765.72, up 0.41%, and the Nasdaq 100 tracker (NASDAQ: QQQ) at $713.44, up 0.35%. The Dow 30 fund (NYSEARCA: DIA) closed at $532.22, up 0.89%.

None of that is attributable to a tennis tournament, and nobody should pretend otherwise. The relevant point for investors is timing rather than causation: the US Open showcase lands while the market is still sorting AI vendors into those with recurring, contracted revenue and those with impressive demonstrations. IBM has spent the current cycle arguing it belongs in the first group, on the strength of hybrid cloud and consulting engagements that convert AI interest into signed work rather than one-off pilots.

What to watch after the trophy is handed out

Three things will tell you whether this is more than a fortnight of good publicity.

• Whether the assistant survives contact with the public. Conversational products fail loudly. A high-profile hallucination during a marquee match would be a reputational cost, not just a bug report.
• Whether the deployment shows up in commentary on deal flow. The value of a showcase is measured in the pipeline it feeds, and that only becomes visible when the company discusses bookings and consulting demand.
• Whether rivals answer in kind. Major sporting properties are among the last genuinely mass-scale, real-time data environments available to a technology sponsor, and competitors have every incentive to buy their own.

For fans, the immediate change is smaller and more concrete: a number on the screen that moves with every point, and a way to ask the tournament a question instead of scrolling for the answer. For IBM, the tournament is a two-week argument, delivered to more than 14 million people, that its AI works when it is watched.

Frequently asked questions

What AI features are IBM and the US Open adding this year?

The tournament is expanding three things: personalized updates tailored to what an individual fan follows, an AI assistant that answers questions in natural language, and win probabilities that are recalculated after every point played. The same underlying match data also feeds tools that players and coaches can use to study opponents and plan match strategy.

How large is the US Open's digital audience?

More than 14 million digital users follow the tournament, according to details discussed on Bloomberg This Weekend. That scale is the point of the exercise for IBM: it puts an AI product in front of millions of concurrent users under real conditions, with heavy traffic peaks and no tolerance for delay during live play.

Who discussed the partnership publicly?

IBM Senior Vice President Jonathan Adashek and former professional tennis player Sam Querrey appeared on Bloomberg This Weekend with hosts David Gura and Christina Ruffini. Adashek represented the technology and commercial side of the partnership, while Querrey spoke to how deeper match data is used by players and coaching teams.

How does a point-by-point win probability work?

Tennis is scored as a sequence of discrete events with a fixed structure, so a model has a clean, high-frequency stream of state changes to react to. Each point, break point, double fault or shift in serving performance updates the match state, and the probability is recomputed against it and pushed to viewers in real time.

Where did IBM shares last close?

IBM finished the most recent session at 235.68, up 0.85% from a prior close of 233.69, with a day range of 233.12 to 238.72, as of 20:00 GMT on Friday, 21 August 2026. Over the same session the S&P 500 tracker SPY closed at $765.72, up 0.41%, and the Nasdaq 100 tracker QQQ at $713.44, up 0.35%.

Does a sports sponsorship like this affect a technology company's revenue?

Not directly and not immediately. The commercial logic is that a large, public, high-load deployment functions as a reference case for enterprise buyers evaluating similar AI systems. Whether it pays off shows up later, in consulting engagements and contracted work, rather than in any revenue line attributable to the tournament itself.

Sources

• AI Takes Center Court at the US Open — Bloomberg Markets

Photo: Adrien Olichon · Pexels Licence — source