Figure 03 Reveals 5 INSANE Upgrades… Tesla’s Robot Rival Is Getting Insane!

Added: 2026-06-04

[00:00:04]Figure just proved who's really winning the humanoid race. Figure AI built one AI brain, Helix O2, that controls the whole robot, fully autonomous. It revealed a seventh generation hand that can feel a touch as light as a paperclip. It unveiled Vulcan, a system that keeps Figure03 standing even after its parts fail. And it built a new factory that now cranks out one robot every hour. And while everyone's busy hyping up Optimus Gen 3, Figure04 is already designed and heading to the floor. Hey guys, Alfie here. Welcome back to AI Nexus. While most of the robot world was busy waiting on Tesla, Figure quietly stacked five [music] wins back to back. Let's go through all five, and by the end, you'll see exactly why this company is breaking away from the pack. First up, the brain. Figure calls it Helix O2. One AI model that controls the entire robot at the same time. the legs, the torso, the head, the arms, and every single finger all run by one system instead of a stack of separate programs bolted together. Figure threw out more than a 100,000 lines of handwritten code, and replace the whole thing with a single neural network. The payoff is full autonomy. Figure 03 can now walk to a dishwasher, unload it, cross a full kitchen, stack the dishes into cabinets, and then load the dishwasher again. 61 decisions in a row, 4 minutes straight, with no human touching anything. Normally, that is exactly where robots fail. One small mistake early on collapses the whole task. Helix O2 keeps track of the job and adjusts in real time instead of freezing. And here's the other half.

[00:01:31]Helix [music] O2 can feel. Touch sensors in the fingertips pick up a force as light as [music] 3 g. That is the weight of a single paperclip. Would you trust a robot to handle your good glasses with a grip that gentle? Tell me in the comments. But the real proof came on a live stream. On May 14th, Figure put three Figure03 robots named Bob, Frank, and Gary on a live feed sorting packages in a San Jose warehouse. The plan was 8 hours, just enough to prove one full shift. 8 hours passed. [music] The stream kept running, 24 hours, 72 hours.

[00:02:03]Then day 7, day 8, day 9, still going.

[00:02:06]In the first 72 hours alone, the three robots sorted around 88,000 packages with no logged failure. Gary alone sorted around 10,000. When Gary's battery ran low, Gary walked to the charging station alone, and another figure 03 took over the line. No human stepped in. Figure even ran a man versus machine contest, and a human worker barely won by less than a tenth of a second per package. You cannot fake any of that with editing. A short clip can hide the bad runs. A 9-day live stream shows every slip the moment it happens, and the slips barely came. Second move, the hand. And this is the one figure is most proud [music] of. The big idea here, dexterity is the single hardest problem in all of robotics. Plenty of companies can make a robot walk. Almost none can make a robot's hands work like yours. So, while everyone else chased better walking, Figure spent three quiet years on hands. The new one is a seventh generation hand on a third generation body. Read that again. The hand is four generations ahead of the robot wearing it. Each finger moves on its own. The thumb rotates across the palm and can reach the tip of every finger. the same trick that makes your hand so capable.

[00:03:12]There are more than 20° of freedom packed in palm cameras so the robot can see what its own fingers are doing even when the head can't and fingertip sensors that feel that same 3 g. For comparison, Tesla's Optimus hand sits around 11° of freedom today. Figure is past 20. If a robot can feel pressure lighter than a paperclip, would you let it handle glass in your kitchen? Drop your answer below. Third move, reliability. This one is called Vulcan.

[00:03:37]A robot that does not die when one part breaks. For years, the rule was brutal.

[00:03:42]A single bad joint in a humanoid's lower body meant the whole machine lost balance and [music] collapsed. Task over. A human had to walk in and drag it off the floor. Vulcan ends that rule.

[00:03:52]With Vulcan running, figure 03 can lose up to three joints or actuators and still keep standing, keep walking, [music] and keep working. Figure proved it live. An engineer triggered a fake knee failure while figure 03 was sorting packages. The robot did not fall and [music] did not freeze. Figure 03 shifted its balance, changed how it moved, and limped, [music] actually limped like an injured person, straight to the maintenance area on its own. A robot beside you loses a knee and [music] just keeps working like nothing happened. Cool or creepy? Let me know in the comments. And here's why this matters so much. Figure wants lights out factories where robots run all night with nobody watching. That only works if a robot can handle its own problems instead of stopping the entire line every time something goes wrong. While Figure03 is outsorting real packages, Figure has already finished designing the next robot. On May 13th, CEO Brett Adcock announced that the F.04, the fourth generation figure robot, has hit design lock. In plain terms, the design is finished and frozen, the parts are being ordered, and the robot is moving into pre-production. Adcock says F.04 04 is the biggest jump between generations the company has ever made with the engineering pushed to a whole new level.

[00:05:01]Now think about the pace here. Figure 01, then 02, then 03, and now a locked in O4. While figure 03 itself is barely out the door. Most companies are still trying to perfect their first humanoid.

[00:05:13]Figure is already building its fourth, fifth, and final move, the factory.

[00:05:17]Because a brilliant robot you cannot build does not change anything. at its bot Q plant in California. Figure went from making 1 figure 03 a day to 1 figure 03 every single hour in under 120 days. That is a 24 times jump in 4 months and more than 350 figure 03 robots have already shipped out the door. The numbers behind it are wild.

[00:05:39]Custom factory software running across more than 150 workstations. Over 50 inspection points along the line. End of line first pass yield already above 80% and climbing every week. the battery line hitting 99.3%.

[00:05:53]Over 9,000 actuators built and every single robot gets pushed through more than 80 tests before signoff, including burn-in sessions of thousands of squats, shoulder presses, and even jogging just to catch weak parts before they ever leave. This is the difference between a science project and a real company.

[00:06:10]Anyone can show one robot in a video.

[00:06:12]Building one an hour is a completely different sport. So, put all five together. The brain with Helix O2, the hands four generations ahead, the reliability of Vulcan, the next robot already locked in, and a factory pumping out one humanoid an hour. That is not one lucky breakthrough. That is the whole stack all moving at once. And that is exactly why Figure is pulling ahead of everyone right now. Which of these five updates impressed you the most?

[00:06:37]Tell me in the comments. After a 200 hour YouTube live stream where Figure03 sorted packages in real time, Figure is now preparing its humanoid robots for massive retail warehouse deployment.

[00:06:48]Then Limx Luna walked onto a runway and proved humanoids can also become public-f facing performers. This is the clearest sign yet that humanoids are leaving the demo phase and entering realworld deployment. But Boston Dynamics had a different message as Atlas started learning football and Unitry just showed a robot that can understand, clean, and adapt inside a messy human space. Figure AI just made a massive move that could reshape the future of American retail. On May 26th, 2026, the company signed a commercial agreement with Catalyst Brands to deploy humanoid robots across its distribution network. The first rollout will begin at Catalyst's Reno, Nevada distribution logistics center. And this is a big step because Catalyst is only Figure's second major commercial customer after BMW. So why is this deal such a big deal?

[00:07:34]Catalyst Brands is not a small retail company testing robots for fun. It was formed in early 2025 after Spark Group and JC Penney merged together. Today Catalyst controls some of the most recognizable American brands including JC Penney, Aerop Pastel, Brooks Brothers, Eddie Bower, Lucky Brand and Nautica. Altogether, the company operates around 1,800 stores and has nearly 60,000 employees across the US and Canada with major backing from Brookfield [music] Corporation and Simon Property Group. But the most important part is what Figures robots will actually do. Its next generation humanoids, likely the Figure03 model, are expected [music] to work inside the Joey Pouch sorting system where they will help with the difficult sorting and packing tasks that normally put heavy physical pressure on warehouse workers.

[00:08:20]Catalyst says this will allow human associates to move toward higher value work instead of spending long hours on the most exhausting [music] warehouse jobs and the timing makes this even bigger. E-commerce has created serious pressure on retail warehouses and companies across the country are struggling with labor shortages. Figure says its robot can already sort packages at near human speeds and its demos have shown more than 200 hours of autonomous operation. Now look at the money behind Figure. The company raised $675 million in series B funding in 2024 at a $2.6 billion valuation. Then in 2025, a $1 billion series C pushed its valuation close to $39 billion. Its investors include Nvidia, Microsoft, Intel Capital, and Jeff Bezos. Figure even moved away from OpenAI to build its own Helix vision language action model. And behind all of this, Brookfield's connection to both Figure and Catalyst may be the key. If the Reno deployment succeeds, Catalyst could eventually expand Figure Humanoids across its massive retail network, creating a new blueprint [music] for how major retail companies use humanoid robots. But while Figure is trying to prove humanoids can quietly reshape warehouse labor, China is showing a very different side of the robot race, one built for attention, performance, and public [music] trust.

[00:09:35]China just revealed a humanoid robot that walks the runway almost like a supermodel. [music] Its name is Limx Dynamics Luna and it is not being presented as another warehouse robot or factory worker. Limx showcased Luna at the Tao Influencer Festival in March 2026, [music] where it walked beside actress Michelle Chen and instantly became one of China's most eye-catching humanoid demos. What made Luna stand out was not just that it could walk, but how smooth and confident the movement looked. The robot used a clean heel totoe gate, natural hip sway, opposing arm swings, and poised runway posture that made it look closer to a real fashion model than a stiff machine. Then Luna performed an illusion turn, a gymnastics inspired spin that showed strong balance, [music] fast feedback, and precise body control with very little wobble. And this happens because Luna is built around a serious motion system. The robot stands about 5' 3 in tall, 160 cm, and weighs around 121 lb, 55 kg, with its battery installed. It uses 27 degrees of freedom and runs on [music] Limx's second generation CIS zero motion engine supported by the COSA cognitive operating system. Together, these systems help Luna move with smoother timing, better balance, and more human-like control. The price is around $41,000, which is aggressive for a full-size interactive humanoid. But Limx is not chasing heavy factory labor with Luna. This robot is designed for malls, theme parks, auto shows, museums, [music] brand events, and live stages.

[00:11:01]That is why it has premium textile cladding, soft rounded features, and a more elegant design that feels safer and more approachable around people. Its hardware also supports longer public performances. Luna has a 10,000 mia hours hot swappable battery with about 4 hours of operation. Limx says endurance improved by 150% while joint surface temperatures dropped by 30%. It can walk at 3.1 mph, 5 kmh, recover balance quickly and carry about 6.6 lb, 3 kg, in its arms. The software makes Luna even more powerful for entertainment. Its video to motion tool can turn uploaded dance clips into robot movement, while the AI task editor lets users create routines from simple text prompts. With swarm control, Limx says more than 200 Luna robots can move together with millisecond timing. While Tesla and Figure Chase Labor, Limx is chasing the attention economy, and Luna shows humanoid robots may win public trust on stage [music] first. And once robots start moving naturally on stages and runways, the next big question becomes obvious. Can that same body control survive something faster, messier, [music] and more unpredictable like sports? Boston Dynamics just enrolled Atlas in football school, and the goal is much bigger than a simple sports demo. Hyundai and Boston Dynamics recently launched a new series called School of Football: Can Football Teach a Robot to Move as part of Hyundai's FIFA World Cup 2026 campaign. The first episode, episode 1, the basics, shows Atlas beginning its football training by studying clips from past World Cup matches before stepping onto the practice field [music] to repeat the movements it just watched. The robot starts with simple football drills, including kicking the ball, shifting its weight, and adjusting its foot placement while moving. Atlas even performs small celebration poses after completing drills, lifting its arms, and dropping to one knee like a real football [music] player reacting on the field. The movement already looks smoother and more controlled than many older Atlas demonstrations. And that improvement comes from a major shift in both hardware and AI. The new Atlas is fully electric, replacing the older hydraulic system that was louder, heavier, and less [music] efficient. The electric design allows Atlas to move more quietly, react faster, and control its body with much better precision.

[00:13:11]Football is also one of the hardest movement challenges for a humanoid robot because it [music] requires balance, fast direction changes, quick reactions, and constant body coordination. Boston Dynamics is also moving toward more adaptive learning based movement instead of relying only on carefully scripted routines. Atlas now studies human motion more naturally and practices movements repeatedly to improve control and stability. That matters because the same balance and coordination skills used on the football field could later transfer into factories, warehouses, [music] and rescue environments. Hyundai bought Boston Dynamics in 2021 to push robotics into real world deployment, and Atlas [music] and Spot are already planned to appear at World Cup venues in 2026.

[00:13:52]Tesla, Figure, and Agility are all racing toward the same future, but Boston Dynamics still has decades of motion expertise behind Atlas, and now that experience is starting to show on the football field. But motion is only half the story. A robot can kick a ball or walk a runway, but the real test is whether it can understand a messy human environment and keep working when things suddenly change. Unitry just dropped a wild new demo that has everyone talking.

[00:14:16]On May 25th, [music] Unitry released footage showing its new WVLA 2.0 O model operating inside a messy conference room. Papers are scattered everywhere, cups are knocked over, and chairs are pushed out of place. Then the robot walks in and starts cleaning the room completely on its own without remote control, without human guidance, and without cutting the video into separate takes. But the most impressive part is what happens during the task. People keep changing the environment while the robot is working. They move objects around, block its path, and make the room harder to clean. Instead of freezing or waiting for instructions, the unitry robot quickly adjusts to the new situation and keeps going. That kind of real-time recovery used to feel far away in robotics, but now it is happening casually in a normal room. So, what makes WVLA 2.0 important? Vla stands for vision language action, which means the robot can see the environment, understand the task, and turn that understanding into physical movement. It uses cameras to scan the room, identify what counts as clutter, decide which object should be handled first, [music] and then plan the motion needed to complete the task. All of this happens in real time while the robot is still moving. Unitry describes this as embodied AI with a built-in world model.

[00:15:26]In simple terms, the robot is not just following a fixed list of instructions.

[00:15:30]It is trying to understand space, object positions, movement, and cause and effect in the real world. That is why it can continue cleaning even when people interrupt the scene or change the environment around it. The walking still looks stiff in some moments, almost like a drunk uncle trying to get home. But autonomy is the real story here. This cleanup demo may look small, but the signal is huge. Humanoid robots are slowly leaving controlled lab setups and stepping into messy human spaces. And that is the bigger pattern across all these updates. Whether it is figure in retail, Luna on stage, Atlas on the football field, or Unitry inside a messy room, humanoid robots are no longer just trying to look impressive. They are starting to prove where they actually belong. China just kicked off mass production of the T800 robot. One T800 rolling off the line every 15 minutes.

[00:16:18]Figure 03 showed us what the future looks like by working almost 200 hours straight on a YouTube live stream. AI just had a wild week in humanoids. This is only the beginning. You can now control the Unitry G1 robot just through your voice. G1 got a voice update. A humanoid just danced ballet with real performers. AI humanoid robots can already do incredible things. Back flips, cartwheels, [music] flying kicks, running across uneven ground, walking with a near human gate, lifting heavy boxes for hours. But here's the twist nobody saw coming. The biggest humanoid companies are no longer chasing those demos. The new race is mass production.

[00:16:55]Who can build these robots at scale? who can hit factory pace. That single metric is now the entire scoreboard. Tesla is retooling its Fremont factory this year, converting the old Model S and Model X lines into the first large-scale Optimus production line aimed at 1 million units per year. Musk is also breaking ground on a second factory at Giga, Texas, targeting 10 million robots annually.

[00:17:18]Figure is already in volume. The Bot Q facility in California is producing figure 03 at roughly one robot every 90 minutes with a 12,000 unit annual run rate and a goal of 100,000 robots over 4 years. Boston Dynamics started Atlas production in Boston in January with Hyundai planning a 30,000 unit annual factory in Savannah, Georgia. That's the field. And Engine AI just jumped in with both feet. Engine AI's 10,000 unit humanoid production line is now live in Shenzhen. The first T800 units are already rolling off the floor. The pace is one finished humanoid every 15 minutes. The factory sits inside [music] Hong Kong industrial zone in Nanchon district. Engine AI runs the full stack in-house research, hardware, manufacturing, quality control, delivery. The upgraded line lifts efficiency by 40%. Every T800 runs through 79 full-dimensional quality checks. The factory also simulates 46 different realworld working conditions during testing. That 15-minute pace is what changes everything. It means cheaper units, faster shipping, real factory data flowing back into the system every single day. Every T800 that rolls off the line teaches the next one how to move better. Engine AI is not stopping in Shenzhen. A second global manufacturing hub is locked in for Junga in Hanan Province. JD.com led the latest funding round with close to 1 billion UN in total capital behind the push. The Shenzhen line is just the opening lap.

[00:18:44]T800 is no longer a demo. T800 is a product on a schedule. But building robots fast only matters if regular people can actually use them. And that side of the story just got its first real proof. Unitry just dropped a demo that shows what living with a humanoid robot will actually look like. You give the robot a command, the robot listens, the AI inside understands what you said, and the robot does the action. That is the full loop. Unitry showed us exactly this with their G1 humanoid. One person talking, one robot reacting live. This is the part most people get wrong about future robots. People still picture it like chat GPT. Open an app, type a prompt, wait for the answer. That is not how a real home robot will work because nobody is going to pull out their phone every time they need water. Nobody is going to type out a prompt to clean the table. You will just talk the same way you talk to anyone living with you.

[00:19:36]Bring me the water. Clean the table.

[00:19:38]Move that chair. That is the natural way humans give instructions. And that is the only way a home robot can actually fit into daily life. This G1 demo is the first real proof that the natural way is coming. So how does this actually work inside G1? It is a chain of steps that happens in seconds. You speak, the microphones on G1 pick up the sound. The AI inside G1 turns your voice into text.

[00:20:01]Then the AI figures out what that text really [music] means as an action. Then a second layer of AI builds the full body motion for that action. Then the whole body controller [music] keeps G1 balanced while the motion plays out and G1 does the move. Every step has to work fast. Every step has to work together.

[00:20:18]Miss one and G1 either ignores you or falls over. G1 has the hardware for it.

[00:20:23]Unitry built in a four mic array, a small speaker, Wi-Fi 6, Bluetooth, and an 8 core CPU. Depth cameras and 3D lighter handle the sensing side. The EDU version can carry an Nvidia Jetson orange chip for the heavy AI work.

[00:20:37]Robots in your home will not run on prompts or remotes. They will run on your voice. And G1 just showed us the first taste. Talking to a robot in your living room is one thing. Putting that robot through a real shift is a completely different test. And one company just blew it wide open. Figure AI just did something no humanoid company has ever done. They showed us what their Figure 03 robot can really do. We have seen many robot demos.

[00:21:01][music] They look impressive. They go viral, but most of them are short. a few seconds, a few minutes at most, recorded, highly edited, cleaned up before you ever see them. Figure AI did the opposite. They showed us what a robot working in a real factory for hours actually looks like. On May 14th, Figure AI started a live stream of three Figure 03 robots named Bob, Frank, and Gary sorting packages at a warehouse in San Jose. The plan was 8 hours, just enough to prove Figure03 could handle one full shift. 8 hours came and went.

[00:21:31]The stream kept running. 24 hours, 48 hours, 72 hours, and counting. Day 7, day 8, the robots are still working. In the first 72 hours alone, figure 03 robots sorted around 88,000 packages with no logged failure. One robot named Gary sorted around 10,000 packages in 7 hours and 44 minutes. When the battery ran low, Gary walked to the charging station alone. Another figure 03 took over the line. No human stepped in. This matters because a short clip can hide everything. Bad runs get cut. Failures get edited out. Only the best take makes the video. A non-stop live stream cannot do that. Every stumble shows up live.

[00:22:10]Every misread package, every slip, every reset plays out in front of thousands of viewers. The stream has already crossed 2 million views. It got even more interesting when Figure ran a 10-hour man versus machine contest. A human worker named Ameerard barely beat the robots. 12,924 packages [music] for the human, 12,732 for the robots. The human won by less than a tenth of a second per package.

[00:22:35]The whole thing runs [music] on Figure's own AI system called Helix O2. Every decision happens on the robot itself. No remote operator, no telly operation, just figure 03, the AI inside, and the packages on the belt. Figure has said the stream will keep going until a robot physically breaks. That is the actual end condition. The era of polished demo clips is closing. The era of robots working real shifts on camera just opened. Sorting boxes for days on end is one side of the humanoid story. The next reveal goes the opposite direction. All grace, all polish, all in front of a live crowd. UB just unveiled the Walker C1 humanoid robot and the launch came with a ballet performance. UB revealed Walker C1 at the China International Supply Chain Expo 2026. This is the first time the world is seeing this robot and UBTE did not pick a normal launch event. The company put Walker C1 on stage with professional human dancers and ran full walts and ballet routines.

[00:23:33]Smooth turns, synchronized movement, real coordination with the people next to it. This is not just a flashy debut.

[00:23:39]Ballet is one of the hardest tests for a humanoid body. The robot has to balance on one leg, hold poses without trembling, move arms and legs at the right moment, match the rhythm of a human partner. One bad calculation and Walker C1 falls in front of a live audience. That is exactly why UBTE picked ballet for the unveiling. The whole point was to prove this new robot can move with the kind of precision a real service environment needs. Walker C1 is built for places like hotels, airports, and exhibition centers. UB named the new robot the first ever silicon spokesperson of the expo. Walker C1 handles guided tours, greets guests, gives information, walks people from one location to another. Walker C1 is an upgraded version of the Walker C from last year. The Walker C base stands at 163 cm tall and weighs around 43 kg. The body has 20° of freedom. Walking speed hits up to 6 km per hour. Walker C uses Ulam for autonomous navigation. RGB- D cameras and 3D structured light cameras handle sensing. A high precision IMU keeps the body steady. Multilingual dialogue is built in. UB is also going hard on production. The company is targeting 10,000 units of annual capacity by 2026. UB locked in 800 million Wen worth of [music] orders during 2025. A $37 million deal is on the books to deploy humanoids [music] at China Vietnam border crossings. The Industrial Walker S2 just hit unit number 1,000 off the