Unitree UniStore: The World's First App Store for Humanoid Robots

The App Store Moment for Physical Robots

On May 7, 2026, Unitree Robotics opened UniStore to the public. The company describes it as the world's first humanoid robot task and action app store. The description is accurate — and the implications extend well beyond Unitree's product line.

The core mechanic is straightforward. A Unitree G1 owner opens the Unitree Explore app on their phone, browses UniStore, taps a skill — say, a Michael Jackson dance, a Mantis boxing routine, or a Charleston — and the motion pack installs to the robot. The robot can now execute that skill on command. No programming. No reconfiguration. One tap.

At launch, 24 motion packs are available, all free during an introductory period. The library covers expressive and demonstrative content — dances, jumps, martial arts inspired by traditional Chinese styles, and social reactions. Third-party developers can publish their own packs. UniStore currently supports the G1 humanoid, H1 humanoid, B2 quadruped, and Go2 quadruped, with a minimum requirement of Unitree Explore app version 1.9.0.

The motion library is the beginning. What Unitree is actually building is the infrastructure layer for a skills economy in physical robotics.

From Hardware Products to Software Platforms

The smartphone analogy is apt but worth unpacking precisely. When Apple launched the App Store in 2008, it did not make the iPhone more powerful in a hardware sense. It made the iPhone infinitely more capable over time by separating the platform from the applications running on it. The phone you bought on day one became a different device — more useful, more personalised, more integrated into daily life — without any change to its physical components.

UniStore applies the same logic to robots. The Unitree G1 you buy today is not the G1 you will own in two years, because the skill library it can draw from will be fundamentally different. Skills you could not imagine at purchase will be available for download. Capabilities that currently require custom robotics engineering will be one tap away.

This matters for three reasons.

First, it changes the value proposition of hardware. A robot with access to a growing skill store is worth more over time than a robot with fixed capabilities. This creates a retention dynamic — owners stay within the Unitree ecosystem not just because of the hardware but because of the accumulated skill library available to them.

Second, it creates a developer economy. Third-party developers who build and publish skills on UniStore are, in effect, building a new software category. Motion skills for physical robots — currently a niche requiring robotics engineering expertise — become a product that can be created, priced, distributed, and iterated on through a standard marketplace model. This is how the iOS developer economy was born.

Third, it accelerates capability development. Unitree does not need to engineer every useful robot skill itself. The developer community will build skills that Unitree's own team would never prioritise — niche industrial routines, cultural dances, accessibility applications, educational programs. The platform compounds in capability faster than any single engineering team could.

The Platform Race in Physical AI

Ole Lehmann, writing on X on May 13, 2026, framed the distinction sharply: when you download a skill for a software AI, it changes what happens on a screen. When you download a skill for a robot, it changes what a physical machine can do in your house, your warehouse, your office.

That distinction is real and important. The skills economy for software AI — Claude, Codex, Gemini plugins — operates entirely in the digital domain. Its outputs are text, code, images, analysis. UniStore's outputs are physical actions in the real world. A skill that teaches a robot to harvest tomatoes, sort warehouse inventory, or fold laundry is not a digital output. It is a change in what physically happens in a space.

This makes the platform race in physical AI qualitatively different from the platform race in digital AI. Whoever builds the dominant skill distribution layer for physical robots controls something closer to an operating system for the physical world than to an app store in the conventional sense.

Unitree is currently the only humanoid robot manufacturer with a live public skills marketplace. That is a significant first-mover position. The question is whether it can convert early distribution advantage into developer ecosystem depth before competitors — Boston Dynamics, Figure AI, Agility Robotics, 1X Technologies — build equivalent infrastructure.

Unitree's Position Heading into the UniStore Era

UniStore launches from a position of genuine market strength. Unitree was the world's largest humanoid robot manufacturer by shipment volume in 2025. Its annual operating revenue reached 1.708 billion yuan, a 335.4% year-on-year increase. The humanoid robot business is now the company's largest revenue segment.

The G1 humanoid — the primary UniStore platform — retails from $16,000, making it the most accessible full-size humanoid robot available for direct purchase. This matters for the app store model: developer investment in a platform is proportional to the addressable user base. A robot that costs $16,000 and is available on Walmart's US website has a meaningfully larger potential installed base than a robot requiring partnership negotiations and enterprise contracts.

Unitree is also preparing for an IPO of up to 4.2 billion yuan on Shanghai's STAR Market and opened its first direct retail store in Beijing's Wangfujing commercial district in May 2026. The retail store and UniStore together signal a deliberate consumer-market positioning — robots as consumer electronics, not industrial equipment.

What UniStore Means for Jobs and Automation

The skills-on-demand model has direct implications for workforce automation that extend beyond the current motion library.

Today's UniStore library is entertainment and demonstration focused — dances, martial arts, expressive reactions. This is appropriate for an early-stage platform where the priority is developer onboarding and user engagement. But the platform architecture is indifferent to the category of skill. The same one-tap installation model that delivers a Charleston dance can deliver a warehouse picking routine, a food preparation sequence, or a patient transfer protocol.

The Geppetto Jobs Impact Index tracks automation risk across 240 professions. Skills marketplaces like UniStore have the potential to accelerate automation timelines in specific profession categories by reducing the engineering friction between a robot platform and a deployable task capability. A warehouse operator currently needs a robotics integration project to deploy a picking skill on a humanoid platform. A future UniStore for task skills could reduce that to a procurement decision.

The professions most exposed to rapid capability expansion via skills marketplaces are those involving discrete, repeatable physical tasks in structured environments — exactly the categories scoring highest in the Jobs Index.

→ See automation risk scores: Warehouse Picker / Packer · Assembly Line Worker · Fast Food Worker

Unitree Robots on Geppetto

Geppetto indexes the full Unitree robot range, including the G1 humanoid — the primary UniStore platform — and the H1, Go2, and B2 platforms also supported at launch.

→ Unitree G1 — Full-size humanoid, from $16,000, UniStore compatible → Unitree H1 — Earlier generation humanoid, UniStore compatible → View all Unitree robots → Unitree in the supply chain — China's robotics ecosystem