Tesla Bot Killer? Figure’s New Robot Enters BMW Factory… Tesla Should Be Worried

[00:00:02]BMW just [music] put a humanoid robot back on its factory floor, and this time the job is much harder. The Figure 03 robot has arrived at BMW Group plant Spartanberg in South Carolina. It follows a successful 10-month pilot with the older Figure O2 model. Now, Figure03 takes on a tougher logistics task called sequencing, and BMW is not the only place these machines are showing up.

[00:00:26]Figure just signed a fresh commercial deal with Catalyst Brands. The company also staged a strange public spectacle to prove a point. It put figure 03 against a human worker in a 10-hour sorting duel. The human won, but only by a razor thin margin. That result [music] exposed something uncomfortable about the robot. Humanoid robots are moving into real [music] work faster than most people expected. The factory floor is where that pressure shows up first. The BMW partnership already has a real track record behind it. Last year, the Figure O2 robot helped build more than 30,000 BMW X3 vehicles. It worked for about 10 months inside the Spartanberg body shop.

[00:01:04]Its main job was inserting sheet metal parts for the welding process. That task demands both speed and accuracy. It can also wear a human body down over a long [music] shift. The collaboration proved humanoids are no longer just lab experiments. Now, figure 03 moves to a completely different part of the plant.

[00:01:21]It handles a logistic sequencing job instead of welding prep. Parts arrive unsorted inside large containers. Figure 03 picks them up and sorts them into a sequencing trolley. That trolley [music] then travels to a defined collection point. From there, the parts reach assembly workers just in sequence. It is a common step in car production and a hard one to automate. Sequencing sounds simple, but it breaks most traditional robots. The problem is that parts never arrive the [music] same way twice. They shift around inside the container. They rotate to odd angles. Sometimes they sit partly hidden behind other parts. A robot running fixed hard-coded motions cannot handle that kind of mess. Every single pick forces [music] the robot to read the scene again. Figure 03 has to actually look before it grabs. Then it makes small corrections on the fly moment by moment. The challenge gets even harder because two things happen at once. The robot must grip parts while also moving its feet and body. It shifts [music] its weight to hold its balance and keep its reach. Figure calls this loco manipulation and it [music] has bottlenecked factory logistics for generations. Picking and walking at the same time is brutally difficult. That exact combination is what fixed factory [music] arms simply cannot do. Figure 03 arrives with a stack of upgrades over the last model. It uses soft outer components designed [music] to make it safer around people. It charges wirelessly which keeps it available for longer stretches.

[00:02:44]>> [music] >> It can even hold spoken conversations through speech-to-pech audio. Its hands got the most serious rework of all.

[00:02:50]Tactile sensors in the fingers help it feel each part it touches. Small cameras sit right inside the palms for extra precision. All of this runs on Figur's own AI model, Helix O2. That brain works fully on board with no remote computer guiding it. Helix O2 ties the robot's whole body together as one system. This lets Figure03 handle a delicate part while stepping and repositioning at once. It can [music] then switch to pulling a heavy metal cart down the line. Precision and raw force now live inside one workflow. Together, these features push Figure [music] 03 well past a simple lab demo. If you are enjoying this breakdown, make sure to subscribe for more. We cover the biggest AI and robotic stories every week so you never miss what's next. BMW is only one front in Figur's rapid expansion. The company just signed a commercial agreement with Catalyst Brands. Catalyst runs well-known retail names like JC Penney, Aerop Pastel, and Brooks Brothers. Under the deal, Figure Humanoids will move into distribution and logistics work. [music] The rollout starts at a distribution center in Reno, Nevada. The goal is to automate the physically demanding jobs across the supply chain. Figure frames it as freeing workers to shift toward higher value tasks. There is a quieter detail that matters just as much here. This is the first commercial bridge between two Brookfieldbacked companies. Brookfield is a shared investor in both Figure and Catalyst. That kind of overlap hints at how quickly this could scale. Figure says humanoids are the only form factor able to slot in this way. One flexible robot form could serve many different brands at once. That flexibility is exactly why holding companies are paying attention. Then came the moment that got everyone talking online. In midMay 2026, Figure ran a live streamed contest for anyone to watch. It pitted Figure03, nicknamed Bob, against a human intern named Ame Gerard. Both did the same warehouse sorting task for 10 straight hours. They grabbed each package and turned it to find the barcode. Then they placed it down so it could keep moving, and there was no hidden operator behind the robot. Bob ran its Helix O2 brain fully on board under the same pressure.

[00:04:55]The humans still won the match by a narrow [music] 192 packages. The final score was 12,924 to 12,732.

[00:05:05][music] That works out to a gap of just 0.04 seconds per package. But the human win came at a real physical price. A needed water, breaks, and rest throughout [music] the shift. He finished with blistered fingers and a forearm he said felt basically broken.

[00:05:20]Bob held a steady pace near 2.83 date 3 seconds [music] per package the entire time. Later, a full fleet of robots ran over 145 hours non-stop, sorting more than 179,000 packages [music] through battery swaps. Figures CEO Brett Adcock called the whole thing a tortoise and hair moment. He even suggested this may be the last time a human wins. So, how does Bob keep that pace without ever tiring? The answer is the Helix O2 brain, which figure launched in January 2026. Instead of one control system, it runs three layers at different speeds.

[00:05:54]The slowest layer, called system 2, handles reasoning at 7 to n herz. It looks at the scene and decides what should happen next. The middle layer, system 1, runs far faster at 200 hertz.

[00:06:06]It turns that plan into real body movement. Then system 0 works at a full kilohertz every single millisecond. It controls balance and force. So the robot never moves like a stiff machine. That bottom layer trained on more than 1,000 hours of human motion. Figure says it replaced over 100,000 lines of handcoded software. The design borrows the fast and slow idea from how human reflexes work. Speed at the bottom layer is what makes the robot feel alive. That is how Bob keeps working as one smooth connected system. And Figure is not slowing down to enjoy this win. The company is already building the robot after Figure03. On May 13th, 2026, CEO Brett Adcock announced Figure04 cleared its critical design review. That means the design passed a major internal checkpoint. Adcock also said Figure04 is in full design lock. He added that parts for it are already shipping. The timing here is worth paying attention to. The whole industry is still waiting on Tesla's Optimus Gen 3. Musk says that design is finalized [music] and the robot is already walking. That matters because Tesla wants to tie robot production to its factory scale and AI hardware. [music] That realworld data could give Tesla a big edge over rivals.

[00:07:20]Tesla is targeting production around summer 2026 with bigger scaling into 2027. Elon Musk finally said [music] something concrete about Optimus and the answer explains almost everything. It explains why the robot has not shown up [music] yet. It explains what Tesla is really building behind closed doors. And for the first time, it puts real dates on mass production. So, here's the question. This whole story hangs on. Why is there still no Optimus Gen 3? Start with the hype. Because the hype has been enormous. Musk and Tesla have described Optimus as the company's biggest product ever. Tesla has even argued the robot's long-term value could eventually surpass the entire electric vehicle business.

[00:08:02]That is a wild claim from a company built on cars. Optimus is supposed to be the product that redefineses what Tesla even is. And yet, Optimus Gen 3 is still nowhere to be seen. The reveal has reportedly been delayed multiple times.

[00:08:17]Every delay fed the same question. Is Optimus real progress or expensive theater? But behind the scenes, Tesla made a move that shows how serious this project really is. The company converted production space at the Fremont factory for Optimus. [music] That includes the lines that previously built the Model S and the Model X. Read that again. The lines that once built two flagship Teslas [music] now belong to a robot. Tesla is using that converted space to support initial assembly of Gen 3 Optimus robots. A car maker does not hand lines to a robot on a whim. That is a bet with real money on the table. It signals that Optimus has moved past the demo phase inside Tesla.

[00:08:56]The robot is being treated like [music] the next product line, not a science experiment. and products need factories.

[00:09:03]Here's the part most people underestimate. Building a humanoid robot is a long, brutal process. [music] Building the line that builds the robot is even harder. Optimus contains [music] roughly 10,000 unique parts that all have to work together. Every one of those parts is a possible bottleneck.

[00:09:19]One weak supplier or one flawed component can choke the entire ramp. And [music] unlike cars, there is no playbook for any of this. Car makers lean on more than a century [music] of refined supply chains and mature tooling. Humanoid robots get none of that history. Cars solved these problems [music] generations ago. Humanoids are solving them for the first time. Musk has framed this as building an entirely new manufacturing system. Experts point out that Tesla is inventing new automation techniques in real time. The actuator supply chain barely exists yet.

[00:09:51]Quality control standards for humanoids are being written from scratch. Every element needs fresh engineering from the dextrous hands to the AI integrated movement. That is why the new Fremont line is the real story here. Tesla is not just assembling robots in that building. Tesla is building the system that builds the robots. Musk has said the early phase is about getting the assembly process [music] right, not chasing volume. Musk has also warned that early output will be quite slow while the kinks get worked out. Early units are expected to handle simple factory tasks before expanding into harder roles. Slow and careful now means fewer disasters later. That is the entire strategy. Then there was another angle and it had nothing to do with robots at all. SpaceX. [music] The SpaceX IPO reports had already created a massive wave of attention. Investors [music] were watching it. Tech media was covering it. Musk's business empire was already sitting in the spotlight. That timing matters. A full Optimus Gen 3 reveal could have swallowed every major [music] tech headline for weeks. It could have pulled attention away from SpaceX right when that attention was most valuable. So, keeping the robot under wraps [music] made sense. All the noise stayed focused on SpaceX instead of being split between two massive Musk [music] stories at the same time. It is the same reason movie studios avoid releasing two blockbusters on the same [music] weekend. Musk's focus was also stretched across Space X, X AI, and Cursor. [music] That is a lot of pressure on one calendar, even for Elon Musk. And then comes the China angle.

[00:11:22]When Musk finally broke his silence, he was replying to a theory about this exact situation. [music] An ex-user named Dr. Jack suggested Tesla delayed the Optimus 53 demo on purpose. The theory claims Tesla realized it was further ahead on production than expected. So, the company allegedly went [music] quiet to stop competitors from copying its designs. Dr. Jack even predicted a major reveal with many working robots [music] on stage. The prediction cited earlier comments from Tesla executive Lars Moravey. And in the humanoid race, the copycats [music] everyone fears most are Chinese. But here is why Optimus Gen 3 might be closer than it looks. The proof [music] came straight from Musk himself. Days ago, Musk posted a photo standing with the Optimus production team inside the Fremont factory. Musk stood with arms crossed, [music] surrounded by workers in hard hats and safety vests. That is not a render or a concept slide. That is a real crew on a real line. Then one day later came the reply that laid out the plan. On July 1st, Musk answered Dr. Jack's theory directly on X. Musk wrote that Optimus production will be extremely slow at first because everything is new. Then came the line that framed the whole project. This is not like making a car. That reply partially shut down the ahead of schedule theory. It also confirmed the ramp will follow a classic manufacturing scurve. Output starts slow, then scaling speeds up once the process matures. Musk flagged that pattern in earlier warnings, [music] long before this tweet. Musk has even called early output rates literally impossible to predict.

[00:12:58]But buried inside all that caution sits the actual timeline. According to Tesla's first quarter earnings commentary, limited production begins in late July or August 2026. That means pilot production at Fremont starts this summer on the converted line. A second much larger Optimus factory is already under construction at Giga Texas. That [music] plant is dedicated to Optimus from day one, not converted from something else. High volume production is expected around summer 2027, [music] and the long-term target is staggering, 10 million robots per year. Analysts expect the slow scurve to accelerate hard once the early problems get solved.

[00:13:37]Some already frame Optimus as a potential trillion dollar product line.

[00:13:41]A trillion dollar label sounds absurd [music] until the volume math starts working. The strategy looks deliberate.

[00:13:47]Prioritize reliability and steady improvement over rushed volume. So yes, Optimus Gen 3 could finally step into the light very soon. The Fremont line comes online this summer and the first production numbers will follow.

[00:14:00]Investors and fans will be watching for the first real capability demonstrations. The bigger picture is simple. Tesla is trading spectacle for reliability, at least for now. Musk has long pitched Optimus as a robot that transforms labor itself. If the ramp works, 2026 becomes the year that vision left the lab. If it stalls, the skeptics get another year of ammunition. By the time Giga Texas hits stride in 2027, the whole debate changes.