When the framing does the heavy lifting

Robot boxing is now officially a spectator sport. Unitree's G1 robots boxed in an organised tournament broadcast on Chinese state television, with human operators controlling every movement. A few months later, similar matches reached San Francisco, where spectators paid to watch VR-controlled robots fight inside a cage. EngineAI has now launched what it describes as the "world's first commercial free-combat competition", with teams competing on software rather than remote control. At RoboCup, a soccer tournament for robots, the emphasis has already moved in the same direction. More teams now share standard hardware and compete through AI. During this year’s FIFA World Cup, Boston Dynamics' Atlas walked the match ball onto the pitch.

These are different events, but they point in the same direction. I keep coming back to combat sports because they expose something unusual about spectatorship. Fighting has never been compelling simply because people hit each other. It works because every exchange carries the possibility of loss. A fighter risks pain, reputation, a career, sometimes permanent damage. That vulnerability gives every decision weight. It gives the audience somewhere to stand. You imagine what it would feel like to commit, to hesitate, to absorb the consequences.

The obvious response is that robots cannot offer any of that. They can be repaired. Nobody's family is waiting backstage. Nobody leaves with a concussion. Then again, spectatorship has never depended entirely on mortal stakes. Formula 1 drivers survive crashes that would once have killed them. Chess players risk no physical harm. Millions watch esports without confusing a digital defeat for a real one. What draws us in is not danger alone. It is the presence of someone whose decisions we can follow, anticipate and judge. A protagonist with enough agency that we begin projecting ourselves into the contest.

Robots have not reached that point. The current generation still looks more like engineering demonstrations than competitors. Even when they move well, intention remains difficult to read. You do not instinctively root for one or resent another. I cannot identify with any of them yet.

That may change. If it does, I do not think the surprising part will be that robots have become more like us. The surprising part will be what it reveals about us. If audiences can transfer that same investment to a machine, then the attachment was never sitting inside the flesh. It was attached to something more abstract. We were following choices made under pressure. We were responding to the appearance of consequence. The human body was simply the first place we learned to recognise the pattern.

When Agility Robotics announced its $2.5 billion merger with Churchill Capital Corp XI on 24 June, one line in the SEC filing caught my attention. The combined company, it said, would become "the only U.S. publicly listed pure-play humanoid company with proven, active commercial deployments." That wording is careful. It is also geographically precise. The filing never claims Agility is the first publicly listed humanoid company anywhere, only in the United States. A week later that distinction looked even narrower. Unitree received approval from China's securities regulator to pursue a Shanghai STAR Market listing, following reports of roughly $250 million in 2025 revenue, profitability with an estimated 35 percent adjusted margin. The Chinese company claimed it shipped more than 5,500 humanoid robots during the year. UBTECH has already been listed in Hong Kong since 2023, while AgiBot and EngineAI are also pursuing Hong Kong listings.

Agility's audited financials have not been publicly disclosed. The investor presentation contains preliminary, unaudited figures with an explicit disclaimer that they "are subject to change, revision, and adjustment" and "should not be relied upon as final financial statements." They also highlight more than $300 million in multi-year Digit v5 orders, while making clear those are contractual commitments subject to milestones rather than recognised revenue.

None of that makes Agility's claim inaccurate. It does make it incomplete in a way that matters if you're trying to price the sector rather than celebrate it. American investors are about to get their first pure-play humanoid listing through a company that has not yet disclosed audited financials. The S-4 registration statement containing those figures has not been filed. The closest thing the industry has to a public operating benchmark is emerging simultaneously in markets many US retail investors do not easily access, with a 363-page audited prospectus already on file. Much of the reporting treated Agility's listing as the category's defining public-market debut without noting that the company pursuing a listing at the same moment has already disclosed the numbers investors would normally use to price a sector. The comparison is missing because half of it hasn't been written yet.

Both pieces this week hinge on the same trick: the frame that decides what you see. Robot combat looks like a story about robots, but the bigger question is what spectatorship was ever really about. Agility's "first" looks like a milestone until you notice how narrowly the sentence was written.