Unitree’s GD01 Just Changed Robotics Forever

Added: 2026-05-17

[00:00:00]For decades, mechs lived in movies.

[00:00:02]Pacific Rim, Transformers, Avatar.

[00:00:05][music] Giant machines you climb inside, pilot with your own body, and unleash on the world. Everyone [music] watched them.

[00:00:12]Nobody actually built one. Until now.

[00:00:14]Unitree just unveiled the GD01, >> [music] >> the world's first mass-produced, ridable, transforming mech that you can actually buy. The founder climbed inside, gripped the controls, walked it across the floor, and punched through a brick wall. It weighs 500 kg with you in it. It transforms from biped to quadruped mid-ride. And it starts at $650,000.

[00:00:38]This is not a movie. This is reality, and it's happening right now. Let's set the stage. Because to understand why the GD01 is such a big deal, you need to understand how long human beings have been obsessed with this idea, and how many times reality failed to deliver.

[00:00:54]The concept of a piloted mechanical giant goes [music] back decades.

[00:00:58]Mazinger Z in the '70s, Gundam in the '80s, Pacific Rim in 2013. The fantasy was always the same. A machine that moves like you, responds like you, but hits like a building. Hollywood spent billions making that look real on screen. And every time a new mech movie dropped, the same question appeared in comment sections everywhere. When are we actually building these? Engineers heard that question. They tried to answer [music] it. And for a long time, the answer was not yet. The physics are brutal.

[00:01:29]>> [music] >> Bipedal locomotion at scale is already one of the hardest problems in robotics.

[00:01:34]Add a human pilot, hundreds of kilograms of steel, real-time balance control, and the engineering complexity does not double. It multiplies. For most of robotics history, a real ridable mech was not just expensive. It was genuinely unsolved. Until a Chinese company that started by making robot dogs decided it was time to change that. Here is the thing. Unitree was not the first to try this, not [music] even close. There is actually a small graveyard of companies that attempted ridable mechs before them, and almost every single one hit the same walls. In 2012, a Japanese company called Suidobashi Heavy Industry built the Kuratas. It was enormous, it looked incredible, and it could technically be piloted from inside. It even had a weapon system that fired when the pilot smiled, but it moved at about 10 km/h on wheels, never walked on legs, and cost around 1.3 million dollars. It was more of an art project than a functional machine. A very expensive, very cool art project. Then there was Prosthesis, a Canadian mech built for racing. No motors, no hydraulics assist, completely human-powered, amplifying the pilot's own leg movement. Impressive engineering concept, brutal to actually operate, and it never found a market beyond [music] niche demonstrations.

[00:02:51]Megabots in the United States built Eagle Prime, a hydraulic battle mech that famously fought the Kuratas in a staged robot duel [music] in 2017. The fight was entertaining. The company filed for bankruptcy shortly [music] after. The pattern is consistent.

[00:03:05]Impressive demos, real engineering effort, and then nothing. No mass production, no commercial path, no actual product that anyone could buy and use. That is exactly what makes Unitree's announcement different. They are not showing a concept. They are not staging a fight. [music] They opened a price list. To understand why Unitree might actually pull this off where everyone else failed, you need to know who they are and how fast they have been moving. Unitree started by doing something nobody thought was possible at the time, making robot dogs affordable.

[00:03:37]Boston Dynamics had the Spot, a capable quadruped that cost around 75,000 dollars. Unitree came in with a comparable machine at a fraction of that price, and sold it to researchers, developers, and hobbyists worldwide.

[00:03:50]They are interested in making impressive things accessible. Then they moved into humanoids. The G1 became one of the most talked about robots in the space, >> [music] >> agile, cost-effective, and built for real environments. But what really turned heads was what they did with the software. They open-sourced a vision language action model, essentially giving any researcher or developer on the planet the ability to build serious AI capability on top of their hardware.

[00:04:16]That is not a move a company makes when it is trying to protect its lead. That is a move a company makes when it is confident enough to grow the entire ecosystem and [music] still come out on top. Then came the IPO filing, 335% revenue growth in 2025, a $610 million listing on China's Star Market, and the same week they launched the GD01, they also opened Unit Store, the [music] world's first humanoid robot app store.

[00:04:41]Running theme here, Unitree does not launch one thing at a time. [music] They drop everything at once and let the world catch up. All right, let's actually talk about the machine itself.

[00:04:51]The GD01 is classified by Unitree as a transformable civilian vehicle, not a prototype, not a concept, a product.

[00:04:59]That classification matters because it tells you exactly how Unitree is positioning [music] it, not as a research project or a one-off demo piece, but as something with a defined use case, a price tag, and a production line behind it. Here is what we know from the demo. The pilot, Unitree founder Wang Xingxing himself, climbs into an open cockpit sitting at the center of the machine. There are no doors, no enclosed cabin. You are sitting inside a structural frame of shiny red panels, silver support bars, and hydraulic lines running down the arms and legs. [music] It looks exactly like what you imagined as a kid watching mech anime, except it is [music] real. It is in a workshop, and someone is actually sitting in it.

[00:05:38]Once inside, Wang grips the controls and sets the machine in motion. [music] It walks forward on two legs with smooth, deliberate steps, not the jerky, cautious shuffling you might expect from something this heavy. Controlled, intentional [music] movement. Then comes the moment that everyone clipped and reposted everywhere. The GD01 walks up to a pile of bricks and punches straight through it. No dramatic windup, no theatrical pause, just 500 kg of mass and structural force doing exactly what physics says it should do. But the most technically impressive moment in the demo is not the punch. It is the [music] transformation. Mid-demo, the GD01 lowers itself backward and shifts into a quadruped form, four legs, still carrying the rider, and continues moving in that configuration. The switch looks rapid and smooth. That dual-mode locomotion is genuinely difficult to engineer at this scale. Bipedal walking and quadrupedal movement >> [music] >> require completely different balance strategies, different weight distributions, and different control algorithms. Getting both working cleanly in a single 500 kg machine with a human inside is not a small thing. On specs, Unitree has kept details fairly tight so far. Total loaded weight sits at approximately 500 kg. Starting price is 3.9 million yuan, which converts to roughly $650,000 for the base model. That price covers a complete, ready-to-pilot system, not a kit, not a deposit on a future product.

[00:07:07]A finished machine heading into mass production. What we do not yet know is battery life, maximum speed, precise payload limits beyond the pilot, and how the control system handles edge cases like uneven terrain or emergency stops.

[00:07:21]Given how little Unitree shared beyond the 1-minute video, there is still a lot of ground to cover in future reveals.

[00:07:27]But what they did show was enough to make the point. This is real, it works, and it is going into production. Okay, so the obvious question, who is writing a $650,000 check [music] for a ridable mech? The honest answer right now is not many people, and Unitree knows that. This is not a mass-market product in the way the G1 or their robot dogs are. At [music] that price point, the buyer pool is small by definition, but small does not mean non-existent. And the use cases Unitree is pointing [music] toward actually make sense when you think them through. The deeper point though is this, Unitree is not trying to sell a million GD 1s. They are planting a flag.

[00:08:07]They are saying that this category exists, that they own it, and that they got here first. The commercial volume comes later. The statement comes now.

[00:08:15]Every company that tried this before hit a wall. Unitree walked through it, [music] literally. If this is what Unitree is shipping in 2026, the only real question is, what are they building right now that we have not seen yet?