An AI-powered robot has beaten elite table tennis players in official competitive matches for the first time. The achievement, published in Nature on Wednesday, marks the first autonomous robot to reach expert-level performance in a real-world competitive sport under the actual rules of the game.

The robot, named Ace, was developed by Sony AI in Zurich. Under official International Table Tennis Federation rules, Ace won three out of five matches against elite amateur players who practice an average of 20 hours per week and have been playing for over a decade. Against two professional Japanese league players, Ace lost both matches but clawed back one game out of seven.

The results represent a breakthrough for physical AI. Previous milestones in artificial intelligence have come from virtual environments: DeepMind's AlphaGo beat the world champion at Go, OpenAI's Dota 2 bots defeated professional esports teams, and Sony's own Gran Turismo Sophy outraced human drivers in a video game. Ace is different. It operates in meatspace, where balls spin at up to 1,000 radians per second, shots arrive in under half a second, and the physics are real.

How It Works

Ace consists of three systems. A multi-camera perception network tracks the ball from multiple angles across the entire court, zooming in on the ball's printed logo to estimate spin and axis of rotation in milliseconds. A deep reinforcement learning brain, trained through 3,000 hours of simulated play, makes shot decisions in real time rather than relying on pre-programmed responses. And an eight-jointed robotic arm on a movable base executes those decisions with the speed and precision required for competitive play.

The spin detection system may be the key innovation. Analysis of Ace's matches showed it won points not by overpowering opponents but through superior control, successfully returning 75% of spinning balls across a wide range of spin types. It also scored multiple direct points on serves, something previous table tennis robots could not do.

The Impossible Shot

During one match, Ace intercepted a ball early and applied backspin in a way that stunned observers. Former Olympic table tennis player Kinjiro Nakamura, watching the match, said: 'No one else would have been able to do that. I didn't think it was possible. But the fact that it was possible means that there is a possibility that a human could do it too.'

That observation may be the most significant part of the research. The robot did not just compete with humans. It expanded the known range of what is physically achievable in the sport. Elite players are already studying Ace's techniques.

Where It Failed

Ace is not invincible. It lost both matches against professional players, winning just one of seven games. Elite player Rui Takenaka identified the weakness: 'If I used a serve with complex spin, Ace also returned the ball with complex spin, which made it difficult for me. But when I used a simple serve, Ace returned a simpler ball. That made it easier for me to attack on the third shot.'

Professionals also noted the psychological dimension. Ace has no eyes to read, no body language to interpret, and no emotional reaction when games reach 10-10. Peter Durr, the Sony AI director who led the project, acknowledged this advantage: 'The players want to see the eyes of their opponent. And the eyes of Ace are all around the court and they don't show any intention or feeling.'

What It Means for Robotics

Jan Peters, a professor of intelligent autonomous systems at the Technical University of Darmstadt, called the project 'truly impressive' but cautioned that table tennis success does not solve broader robotics challenges like object manipulation. To be useful for the general public, 'a lot of good old-fashioned engineering is needed,' he said.

But Peters added a prediction: 'There will be a moment in the next decade which will change the world as much as ChatGPT did in 2022. That moment may be closer to now than to 2036.'

The significance of Ace is not that it plays table tennis. It is that it proves a robot can perceive unpredictable changes in a physical environment, decide how to respond, and execute that response at human speed under competitive pressure. That capability has applications in manufacturing, surgical robotics, autonomous vehicles, and any domain where millisecond decisions must translate into precise physical action.

DeepBlue beat Kasparov. AlphaGo beat Sedol. Ace beat elite table tennis players in the real world. The gaps between these milestones are getting shorter.