In partnership with

The Gold standard for AI news

AI keeps coming up at work, but you still don't get it?

That's exactly why 1M+ professionals working at Google, Meta, and OpenAI read Superhuman AI daily.

Here's what you get:

• Daily AI news that matters for your career - Filtered from 1000s of sources so you know what affects your industry.

• Step-by-step tutorials you can use immediately - Real prompts and workflows that solve actual business problems.

• New AI tools tested and reviewed - We try everything to deliver tools that drive real results.

• All in just 3 minutes a day

Join 1M+ pros

The Robot That Showed Up For Work Every Day: Inside Figure AI's Five-Month Manufacturing Marathon

Listen, I know what you're thinking. Another breathless article about how robots are coming for our jobs, complete with obligatory Terminator references and hand-wringing about automation. But here's the thing: something genuinely historic just happened at a BMW factory in South Carolina that nobody's really talking about.

For five straight months—ten hours a day, five days a week—a humanoid robot named Figure 02 has been showing up to work on BMW's production line. Not for a demo. Not for a carefully staged photo op. Actually working. Picking up sheet metal, placing it in fixtures, doing the kind of repetitive physical labor that humans have been doing since Henry Ford invented the assembly line.

Why does this matter? Because this is the first time anyone has kept a humanoid robot continuously operating in a real industrial environment for this long. Not a robot arm bolted to the floor doing one specific task. A walking, human-shaped robot that navigates the same spaces we do, uses the same fixtures, works alongside actual humans.

We just crossed a threshold, and most people missed it.

The Company That Went From Zero to $39 Billion in Three Years

Figure AI didn't exist in 2022. Now it's worth more than fucking Lockheed Martin.

Founded by Brett Adcock—a serial entrepreneur who previously built Vettery and Archer Aviation—Figure assembled what can only be described as the Avengers of robotics talent. Engineers from Boston Dynamics, Tesla's Autopilot team, Google DeepMind, Apple. The kind of people who've spent decades making robots that can backflip or navigate complex environments or not kill people.

The valuation trajectory tells you everything about how investors view this space:

• May 2023: $500 million

• February 2024: $2.6 billion

• September 2025: $39 billion

That's a 15x increase in 18 months. For context, that's faster growth than OpenAI, Stripe, or SpaceX at comparable stages. The February 2024 funding round alone pulled in $675 million from Microsoft, OpenAI, NVIDIA, Jeff Bezos, and Amazon's Industrial Innovation Fund. When tech giants who normally compete are lining up to invest in the same Series B, something interesting is happening.

What Actually Makes Figure 02 Special

Let's get technical for a second, because the specs matter.

Figure 02 stands 5'6" and weighs 70 kilograms—roughly the size of an average human. That's not an accident. The entire point of a humanoid form factor is that it can operate in environments designed for humans. No need to redesign your factory floor, reconfigure your equipment, or install special infrastructure.

The hands are where it gets wild: 16 degrees of freedom per hand with human-equivalent strength. That means it can manipulate objects with the same basic dexterity you have, lifting up to 20 kilograms (44 pounds). The custom 2.25 kWh battery provides over five hours of operation—enough for a full shift before needing a recharge.

Here's the kicker: it operates fully autonomously. No remote control. No human operator with a joystick. Six RGB cameras feed into an AI vision system that figures out what to do in real-time. It walks at 1.2 meters per second (about 2.7 mph), which is roughly the speed of a leisurely human stroll.

But none of those specs mean shit if the robot can't actually work day after day without breaking down, getting confused, or accidentally yeeting sheet metal across the factory floor.

The BMW Partnership: Where Theory Meets Reality

When Figure announced their partnership with BMW in January 2024, it was easy to be skeptical. Tech companies announce "partnerships" with major manufacturers all the time. Usually it means someone got a nice PR opportunity and a pilot program that goes nowhere.

This one was different.

BMW put Figure 02 to work on the X3 body shop production line at their Spartanburg, South Carolina facility—one of the largest automotive manufacturing plants in the United States. The robot's job: material handling. Pick up sheet metal components from bins, place them into welding fixtures where other machines (or humans) can work on them.

It's not glamorous work. It's exactly the kind of repetitive, physically demanding task that causes worker fatigue and injury over time. Which is precisely why it's the perfect first use case.

Milan Nedeljković, BMW's Board Member for Production, put it this way: "With an early test operation, we are now determining possible applications for humanoid robots in production. We want to accompany this technology from development to industrialization."

Translation: We're not just testing if this works. We're figuring out how to scale it.

The five-month continuous operation is a dramatic leap from the 20-hour run the company achieved back in May 2025. Twenty hours is impressive for a demo. Five months is what you need to prove that humanoid robots are ready for actual industrial deployment.

Figure 03: The "We're Serious About Manufacturing" Version

Yesterday—literally as I'm writing this—Figure unveiled Figure 03, and holy shit did they redesign everything.

This isn't an incremental improvement. It's a ground-up rebuild engineered specifically for two contexts: homes and mass manufacturing. The new features reveal where Figure thinks this is going:

Tactile sensors that can detect forces as light as 3 grams—the weight of a paperclip. That's the difference between a robot that can pick things up and a robot that can feel what it's doing. Essential for delicate manipulation or operating around humans safely.

Wireless inductive charging through coils in the feet. Just park it on a charging pad. No plugs, no human intervention needed.

Soft textile coverings instead of hard plastic shells. This is the "we're going into homes" tell. Nobody wants a metal exoskeleton bumping into their furniture or, god forbid, their kids.

Enhanced audio systems for voice interaction. Because if this thing is going to work in your house, you need to be able to talk to it like you would a person, not program it like a roomba.

But here's what really matters: Figure 03 is engineered for high-volume manufacturing. Remember, these things are useless if they cost $500,000 each. Figure gets this, which is why they launched BotQ.

BotQ: The Factory Where Robots Build Robots

In March 2025, Figure opened BotQ—a dedicated manufacturing facility designed to pump out 12,000 humanoid robots per year in its first generation. But the long-term vision is even more meta: robots building robots.

Think about what that means. If you can get humanoid robots reliable enough to perform manufacturing tasks, you can deploy them in your own robot factory. They become the material handlers, the assembly workers, the quality inspectors. Each generation of robots helps build the next generation.

This is the recursive manufacturing holy grail that's been theorized for decades. Figure plans to use its own humanoids to eliminate conveyor systems entirely. Just have robots carry components from station to station.

The facility integrates advanced manufacturing execution systems (MES), product lifecycle management (PLM), and enterprise resource planning (ERP) software. This isn't a garage operation with some 3D printers. This is industrial-scale production infrastructure designed for massive volume.

The Market Context: Why This Is Happening Now

Here's the uncomfortable truth that's driving all of this: we're running out of humans who want to do factory work.

The U.S. could face 1.9 million unfilled manufacturing jobs by 2033. Not because the jobs disappeared—because there aren't enough people to fill them. Birth rates in developed economies have been declining for decades. The workforce is aging. Manufacturing jobs, especially repetitive physical labor, aren't exactly attracting Gen Z talent.

China installed 295,000 new industrial robots in 2024 alone—more than half of the world's 542,000 new installations. This isn't about replacing workers to boost profits. It's about maintaining production capacity when the labor supply is fundamentally constrained.

Goldman Sachs projects the humanoid robotics market will hit $38 billion by 2035. The Robotics-as-a-Service (RaaS) market—where companies essentially lease robots instead of buying them—is expected to reach $12.4 billion by 2035, growing at 18% annually.

Those aren't moonshot numbers. That's a mainstream industrial category forming in real-time.

The Competition: Tesla, Boston Dynamics, and the Race for Scale

Figure isn't alone in this space. Tesla's Optimus, Boston Dynamics' Atlas, Agility Robotics' Digit—everyone's building humanoid robots now.

Tesla has set audacious goals: 1 million Optimus robots per year within five years, priced below $20,000. If anyone can hit that kind of manufacturing scale, it's the company that produces 2 million vehicles annually. But here's the thing: Tesla doesn't have a single Optimus deployed in a real production environment yet. Figure does.

Boston Dynamics has decades of experience and the most advanced locomotion technology in the world. Their Atlas robot can do parkour, for Christ's sake. But they've struggled to commercialize. Figure is commercializing now.

The competitive dynamics got spicy in February 2025 when Figure ended its partnership with OpenAI, deciding to develop AI capabilities in-house instead. Their reasoning: general-purpose language models aren't sufficient for embodied robotics. You need AI specifically designed for physical manipulation, spatial reasoning, and real-time decision-making in three-dimensional environments.

That's a bold bet. Either Figure builds better robotics AI than OpenAI, or they've just made an expensive mistake. Based on the five-month BMW deployment, they might be right.

The Deeper Mechanism: Why Humanoid Form Factor Matters

Here's where we need to zoom out and talk about why humanoid robots are the breakthrough, not just better industrial automation.

The world is designed for human bodies. Doorways, stairs, vehicles, tools, workspaces—all optimized for our proportions and capabilities. Traditional industrial robots require custom environments. You build the factory around the robot's capabilities.

Humanoid robots flip that equation. They adapt to environments designed for humans. That means you can deploy them anywhere without massive capital expenditure on retrofitting. A Figure 02 can work at a BMW factory today, and potentially work in a distribution center, a retail store, or your home tomorrow.

This is the same principle that made the iPhone revolutionary. It wasn't the best phone, the best camera, or the best music player. It was good enough at everything in a single form factor. Humanoid robots don't need to be the absolute best at any single task—they need to be capable enough at a wide variety of tasks in human environments.

The alternative is a world where we need specialized robots for every context. A warehouse robot, a manufacturing robot, a home robot, a retail robot. Each one requires unique development, training, and infrastructure. That doesn't scale.

Safety and the "Don't Kill People" Problem

Let's address the elephant in the room: traditional industrial robots are caged for a reason. They're powerful, fast, and dumb. They'll happily crush whatever gets in their path because they can't tell the difference between a metal component and your arm.

Humanoid robots operating in dynamic environments with human workers present a different safety challenge entirely. They need collision avoidance, force limitation, and emergency response protocols. They need to navigate unpredictable situations—a worker stepping into their path, an object falling, equipment malfunction.

Figure's approach layers multiple safety systems: advanced sensors, AI-powered collision detection, comprehensive testing protocols. The five-month BMW deployment is crucial validation that these systems work under real production conditions. Not in a controlled lab. Not in a carefully monitored pilot. In an actual factory with actual humans working nearby.

This is why the continuous operation milestone matters so much. It's not just proof that the robot can work—it's proof that it can work safely for extended periods.

What This Reveals About the Future

Brett Adcock's stated goal: deploy 100,000 autonomous humanoids within four years. That's a 833x increase from current operations. Insane? Maybe. But consider what needs to be true for that to happen:

The technology needs to work reliably. ✓ (Five months of continuous operation suggests they're there)

Manufacturing needs to scale dramatically. ✓ (BotQ facility designed for 12,000 units per year, expanding)

Costs need to come down substantially. ? (TBD, but mass manufacturing typically drives costs down)

Customers need to adopt at scale. ? (BMW deployment provides crucial proof-of-concept)

Three out of four ain't bad.

Here's the thing that gets me: this isn't about replacing all human workers. It's about doing work that humans increasingly don't want to do, in contexts where we're running out of humans to do it. Manufacturing jobs, warehouse operations, repetitive material handling—these roles are already hard to fill.

The question isn't whether humanoid robots will take these jobs. The question is whether there will be enough humans who want these jobs as demographics shift and labor markets tighten.

The Uncomfortable Truth About Progress

Every industrial revolution follows the same pattern. New technology emerges, disrupts existing work, creates anxiety and resistance, and ultimately gets adopted because the economic forces are too powerful to resist.

The Luddites weren't wrong that textile machinery would eliminate jobs. They were wrong about whether that would stop its adoption. Automation has been displacing human labor for 200 years, and yet we have more jobs now than ever. They're just different jobs.

The humanoid robot revolution will be no different. Some jobs will disappear. New jobs will emerge. The workers who adapt will thrive. Those who don't will struggle. That's not a moral judgment—it's just how these transitions work.

What's different this time is the speed and scale. Figure went from not existing to a $39 billion valuation in three years. That's faster than any previous industrial technology transition. The deployment timeline will be faster too.

The Bigger Picture

Five months of continuous humanoid robot operation at BMW isn't just a milestone for Figure AI. It's a signal that we've crossed the threshold from demonstration to deployment. From "cool technology" to "industrial tool."

The robots aren't coming. They're already here. They're already working. And they're getting better at an exponential rate.

Within a decade, humanoid robots in manufacturing facilities will be unremarkable. Within two decades, they'll be in homes, retail stores, hospitals, and anywhere else that humans work. Not because of some sci-fi vision of the future. Because the economics and demographics make it inevitable.

Figure AI's achievement is that they proved it's possible now. Not in some distant future. Not in a carefully controlled demo environment. In a real BMW factory, doing real work, for five straight months.

The future of work just got a lot more complicated. And a lot more interesting.

Links and Observations

• The timing of Figure 03's announcement—literally the day after highlighting the five-month deployment—is peak tech company strategy. Prove it works in the real world, then announce the next-generation product. Well played.

• It's wild that Figure ended their OpenAI partnership to build AI in-house. Either that's visionary or catastrophically expensive. No middle ground.

• The fact that NVIDIA is invested in Figure (and participated in multiple funding rounds) tells you everything about where they think compute demand is going. Not just data centers—embodied AI in physical robots.

• I keep thinking about the BotQ recursive manufacturing vision. If they pull it off, production scaling follows exponential curves instead of linear ones. That's when things get really weird.

• Someone at BMW deserves a promotion for green-lighting a five-month continuous deployment of unproven humanoid robotics technology on an active production line. That's either courage or recklessness, depending on how this plays out.