Two weeks after dropping the GD01 transformable mecha on the world, Unitree is back with a launch that, on the surface, looks much smaller — and that I would argue is the more important of the two. The company has opened UniStore, a marketplace where third-party developers can publish, sell and share skills: pre-packaged behaviours, gestures, moves and full routines that a Unitree robot can download and perform on demand.
The pitch, if you squint, is almost exactly the one Epic Games used in 2017 for Fortnite emotes: here is a catalogue of dances, celebrations and signature moves; pick one, pay a small price, and your avatar performs it forever. Swap the avatar for a real piece of hardware standing in your living room and you have UniStore. The first listings include kung-fu kata sequences, breakdancing routines, a presenter-style bow, hospitality greetings, and a handful of more utilitarian behaviours like sit-to-stand recovery and tray balancing.
Unitree did not just open a content store. It opened a supply chain where the suppliers are every animator, choreographer and reinforcement-learning hobbyist with a laptop.
What UniStore Actually Is
Strip the gaming analogy back for a second. UniStore is a developer marketplace bolted onto Unitree's robot operating environment. Developers upload a skill — either a recorded motion sequence, a learned policy, or a higher-level scripted behaviour. The skill goes through Unitree's safety and compatibility checks. Once approved, it appears in the store, priced by the developer. End users browse it from the same companion app they already use to control their robot, buy with a tap, and the robot downloads it and is ready to perform it within seconds.
Unitree takes a cut. Developers keep the rest. The numbers Unitree quotes publicly — a 70/30 revenue split in favour of the developer — are, suspiciously, the exact ratio Apple and Google pioneered for app stores. That is not an accident. This is a deliberate copy of a playbook that has been proven, over fifteen years, to turn a hardware product into a platform.
Why the R1 Changes the Maths
A marketplace only works if there is an installed base big enough to attract serious developers. For most humanoid manufacturers that is still a long way off — Figure, 1X and Boston Dynamics between them have shipped a few hundred commercial units. Unitree is in a different position. The Unitree R1, launched earlier this year at a price point that finally crossed the line into the consumer bracket, is on track to be the first humanoid platform with a five-digit install base. That is the number where things start to be interesting for an outside developer.
Ten thousand units is not Fortnite. But it is more than the original iPhone shipped in its first few months, and that did not stop a generation of small studios from betting their careers on it. A choreographer in Seoul who spends two weeks recording a clean dance routine, lists it for $4.99 and sells two thousand copies has made a year of beer money from a side project. Multiply that across a few thousand creators and the catalogue grows faster than any in-house team at Unitree could ever produce.
The Bigger Idea: Manufacturers Should Stop Trying to Own the Software
I have been saying this in print for a while. In our piece on why the software running your humanoid should belong to everyone, the argument was that hardware is the part nobody can replicate cheaply, and software is the part everyone now can — because the cost of writing it has collapsed. The conclusion was simple: the manufacturer that lets go of the software wins the decade.
UniStore is not full open source. Unitree still owns the runtime, the safety layer and the distribution rails. But it is the most pragmatic intermediate step we have seen any major hardware vendor take — and it answers the obvious objection to pure open source, which is that someone needs to get paid for writing the software in the first place. UniStore says: the developers get paid, in cash, every time a robot performs their work. The manufacturer keeps the hardware margin. The user gets a constantly expanding catalogue without waiting for Unitree's next firmware update.
Everyone wins, in the precise sense that everyone has a financial reason to keep showing up.
Where the Value Actually Lives
The obvious starting point — dance routines, choreographed greetings, party tricks. Low stakes, high virality, perfect for the first wave of paying customers and the first wave of TikTok-ready creators.
Pre-recorded sequences from named instructors. The economic model here looks more like Patreon than the App Store — a Tai Chi master can monetise expertise that previously only travelled with their body.
Standing-from-fall recovery, guided morning routines, exercise companions. Niche but high-value — institutions will pay enterprise pricing for skills that consumer users would never touch.
Tray balancing, gait switching for uneven floors, low-light navigation packs. Where pure utility skills meet small business willingness to pay — the boring revenue that funds the whole catalogue.
The Risks Are Real, and Worth Saying Out Loud
I am genuinely positive on UniStore, but it is not without traps. Three are worth flagging early, before the first round of breathless press coverage forgets to mention them.
1. Curation and safety
A skill that makes the robot dance is harmless. A skill that makes the robot lift a child, pour a hot drink, or interact with an animal is not. Unitree's safety review process is the single most important — and least visible — part of the entire product. If they get the curation wrong, one viral incident is enough to put the marketplace on the front page for all the wrong reasons. The platforms that failed historically (early Android Market, early Steam Greenlight) all failed at this exact step.
2. The race-to-the-bottom on price
App stores compress prices ruthlessly. The choreographer charging $4.99 in month one will be undercut by a $0.99 clone in month four and a free-with-ads version in month eight. The pattern is so well-documented that there is no good reason to think UniStore avoids it. The question is whether Unitree builds in protections — minimum prices, exclusivity windows, verified-creator programmes — before the catalogue is too big to police.
3. Lock-in dressed up as ecosystem
Once you have bought fifty skills for your Unitree R1, you have a switching cost to a competing humanoid that did not exist before. That is good for Unitree and not necessarily good for the industry. The healthiest version of this market would be a common skill format that any humanoid vendor can implement — a USB-style interoperability layer. We are nowhere near that. UniStore could be the first step toward it, or it could be the moat that prevents it.
Competing With Giants Means Taking Bets Like This
Unitree is not, by global standards, a large company. It competes against Tesla, Figure, Apptronik and the entire Chinese state-backed humanoid industry. Differentiating on hardware alone, against that field, is a losing strategy. The only way a mid-cap robotics company secures a real position is by changing the shape of the market — by making the platform, not the unit, the thing customers buy into.
UniStore is exactly that bet. It says: the value of owning a Unitree robot is not the metal and the motors. It is the catalogue of things it can do, and the rate at which that catalogue grows. That argument compounds over time in a way pure hardware specs never will. A 2026 R1 with three years of UniStore behind it will be a meaningfully different product from a 2026 R1 with no ecosystem at all — and that difference is what locks customers in.
Our Verdict: The Smartest Move of Unitree's Year
The GD01 mecha got the headlines. UniStore deserves the analysis. One is a spectacle, an expensive piece of corporate marketing aimed at a few hundred customers who do not yet have a reason to exist. The other is a structural change in how a Chinese robotics company plans to make money over the next decade — and a quiet message to every other manufacturer that the software economy is now part of the competition, whether they wanted it to be or not.
Five years from now, the manufacturers still pretending they can own every line of code that runs on their hardware will look the way Nokia and BlackBerry looked in 2012 — defending a moat that stopped being the moat. Unitree just told the rest of the industry which way the wind is blowing. The fact that the first listings are kung-fu kata and breakdance routines does not change the shape of what they have actually built underneath. It just makes it easier to underestimate.
Big risks, big bets, big potential rewards. This is what real innovation looks like in a market crowded with giants — and it is exactly the right move for Unitree to be making right now.