Figure’s Humanoid Factory Tour – CEO

Podcast: Sourcery | Participants: Brett Adcock, Molly O'Shea, Moritz

1. Key Themes

The "iPhone Moment" Hardware Strategy: One Body, Infinite Apps

Figure's core thesis is that a single general-purpose humanoid body, like a smartphone, can run any "app" (use case) — from laundry to logistics to healthcare. This amortizes hardware costs across infinite deployments, which is the fundamental economic bet of the company.

"What is the same thing for humanoids? What are the same humanoids to be able to do like dishes and laundry, but also do some like package logistics and healthcare and their stuff. That's what humans can do, right? We're all fairly general purpose. And so we want one set of hardware that we can amortize over like a lot of different use cases." — Brett Adcock 00:17:06

Neural Networks Won: The Death of Code-Based Robotics

Figure made a complete pivot away from traditional code-based control systems to pure neural network AI (Helix). This is not incremental — it's a wholesale architectural replacement. Companies still building on traditional robotics stacks are on a dead-end path.

"You basically can code your way out of this. And I think that's a full dead end or you can run, you can AI for a strategy in the market. And that's what we do here at Figure." — Brett Adcock 00:19:37

"About a year ago, we made a strong pivot away from code into neural networks. And I think the team has probably shown like, I think some of the best neural networks for control in the world on humanoids." — Brett Adcock 00:25:22

Data is the New Moat: The Flywheel That Will Separate Winners from Losers

The biggest bottleneck to scaling humanoid robots is not hardware or software — it's diverse, high-quality real-world data. The companies that deploy robots earliest, at scale, will win because they accumulate training data that nobody else has.

"The biggest blocker for us now of going from where we're at today to like large scale deployment is data. We need like an enormous amount of data. We need to pull a lot of our resources further into pre-training for the Helix team." — Brett Adcock 00:31:33

2. Contrarian Perspectives

AGI Will Be Achieved in Physical Robots Before Disembodied LLMs

Most AI discourse focuses on language models as the path to AGI. Brett argues the opposite — physical embodiment and real-world interaction is the missing piece, and humanoids may get there first.

"It actually might be the case that we get to artificial general intelligence first in these embodiments... Most of human intelligence is built this way. And I think this is like the last missing piece to get the true AGI is this like real world interaction with our environments." — Brett Adcock 00:46:33

Hardware Is Essentially Solved — The Real Bottleneck Is Software/AI

The common perception is that building the robot body is the hard part. Brett explicitly flips this: hardware is now robust and reliable, and virtually all remaining failures are software.

"Hardware has gotten really robust. I mean, the hardware here is like, is great. We just like basically, it's basically a software issue." — Brett Adcock 00:10:53

"Most of our failures today are kind of in software land." — Brett Adcock 00:12:59

Speed Over Cost Is the Right Early Strategy — Even If It Looks Wasteful

Figure deliberately built robots that were hundreds of thousands of dollars each in early generations, prioritizing speed to get hardware into AI developers' hands. Conventional wisdom would say optimize for cost from the start. They argue this was the correct call.

"We optimized these first few generations for speed. We didn't care as much about costs... Even if it's expensive, let's get stuff to the team as fast as possible." — Brett Adcock 00:59:22

Figure 4 Will Be a Bigger Leap Than Anyone Expects — We Are Still in "Flip Phone" Era

The market perception is that humanoid robots are rapidly maturing. Brett argues we haven't even hit iPhone 1 yet, and Figure 4 will be the most dramatic generational leap they've ever made.

"Figure four will be the biggest step up we've ever made by far... We are just so early. We're like almost in like flip phones and now we're entering like iPhone one moment. I think maybe figure four will be our first like iPhone one moment." — Brett Adcock 01:02:26

3. Companies Identified

BMW Leading global automotive manufacturer. First company to deploy humanoid robots on an active factory production line, partnering with Figure for six months in the body shop. The BMW X3 assembled with Figure's help is described as the first car ever built (in part) by a humanoid robot, and Brett personally purchased four of them as collector's items.

"We worked at BMW. And last year we deployed, for six months robots on the basically body shop factory line to build cars... This was the first car in the world built by a humanoid robot that we're aware of." — Brett Adcock 00:35:11

Archer Aviation eVTOL aircraft company founded by Brett Adcock before Figure. Mentioned as the engineering blueprint and talent pipeline for Figure — the systems are analogous (electric motors, battery packs, control software, embedded systems, sensors). The cross-domain experience is what allowed Figure to build so fast.

"My company before this designed like flying robots at Archer and it's got the same properties. We have a battery pack, but instead of like a two kilowatt hours, it's 160 kilowatt hours... It's a robot." — Brett Adcock 00:43:49

4. People Identified

Moritz (Last Name Not Given) Controls team lead at Figure, responsible for the Helix neural network controller. Demonstrated the robot's ability to withstand significant physical pushing forces and maintain balance, and led development of "Project Vulcan" — the capability to lose a knee joint and continue operating without falling.

"Moritz, can you show her what would happen if we lost like a left knee?" — Brett Adcock 00:23:59 "This is kudos to the controls team. I think we have like one of the best controls team in the world." — Brett Adcock 00:25:22

David (Last Name Not Given) Head of Industrial Design at Figure. Leads the team responsible for the aesthetic and human-machine interaction design of the robots — how they look, move, make eye contact, and feel to be around. Not a technical robotics role, but described as critical to Figure's product differentiation.

"We have an industrial design team here internally run by David... We want to create something that is really delightful to be around. And it's not just like the way the robot looks or the size of it. It's how it walks and interacts and how it's body language and how the human machine interaction is." — Brett Adcock 01:09:48

5. Operating Insights

"Trust But Verify" as a Manufacturing Philosophy — Run Failure Cases to Exhaustion Before Shipping

Figure's system integration testing lab operates on a ruthless principle: find every possible failure internally before it reaches a customer. They run robots 24/7 with daily test plans and treat any upstream escape as unacceptable. This prevents customer-site mysteries.

"The banner on here on system integration tests is trust, but verify. We basically have to make sure we run down every potential thing that could go wrong before leaving here." — Brett Adcock 00:09:30

Build Redundancy for Your Most Catastrophic Failure Mode Before You Need It

Rather than accepting that certain failure modes (like losing a leg joint) will cause a full system stop, Figure proactively built Project Vulcan — a dedicated initiative to survive joint failures. This "Never Fall" mentality applies broadly: identify your single catastrophic failure, then engineer specifically to survive it.

"We actually have initiative internally called never fall. It's like, we never ever want to fall... We've been working on a project we call Vulcan here internally that basically allows us to lose a single or even multiple joints in the legs and still not fall." — Brett Adcock 00:23:30

Walk the Factory Floor Every Single Day as CEO

Brett Adcock walks every part of the campus — engineering, manufacturing, testing, design — every day. In a hardware company, physical presence and daily visibility across all functions is a direct operating advantage.

"Every day." — Brett Adcock, when asked how often he walks through campus 00:40:10

6. Overlooked Insights

Sim-to-Real Transfer Is Effectively Solved at Figure — This Is Underappreciated

The ability to train robots entirely in simulation and then "zero-shot" deploy the policy directly onto physical robots with high fidelity transfer is mentioned almost in passing, but it is a massive capability unlock. It means Figure can scale AI training without proportionally scaling physical robot infrastructure — a huge cost and speed advantage.

"We basically can zero shot it onto this robot. Meaning like we can just put it right on, load it to the computer and we can basically get this level of performance in the controller... We have like a basically a very high transfer rate." — Brett Adcock 00:15:08

If this transfer rate is genuinely high, it means the real training compute happens in simulation (cheap, scalable, parallelizable), and physical robots are essentially deployment endpoints — not training infrastructure. This is a fundamentally different and more scalable model than anyone's talking about publicly.

Figure Is Quietly Building a Robots-Building-Robots Factory — Already Started This Year

In a single offhand sentence, Brett reveals that humanoid robots are being deployed into Figure's own manufacturing line this year to help build more humanoid robots. This is the beginning of the recursive manufacturing loop that, if it works, creates an exponential production scaling curve that no competitor can match without having already deployed at scale.

"We'll be shipping robots, humanoid robots into the production lines here this year. And at some point, it'll just be full lights out manufacturing. We'll have like robots only building robots and sending out to the world." — Brett Adcock 00:45:25

This is the most strategically significant statement in the entire episode. The first company to achieve robots building robots at meaningful yield rates creates a self-reinforcing manufacturing advantage that compounds faster than any traditional capital expenditure can match.