- Published on
Figure CEO on Scaling Humanoids: 'You Can't Code Your Way Out of This Problem'
- Authors
- Name
- Humanoids daily
- @humanoidsdaily
Adcock on Figure's Strategy: "We're Here to Sell Work"
Hours after releasing a new video showcasing the Figure 02's impressive package handling skills, Figure AI founder and CEO Brett Adcock appeared on stage at the Bloomberg Tech summit in San Francisco to provide rare, in-person insight into the company’s ambitious strategy. He painted a picture of a company focused on a single prime directive: to "sell work" by deploying a fully vertically integrated solution of hardware, software, and AI.
"My belief is that the humanoid robotics will not only build the biggest business we've ever seen by many, many multiples, but it's going to be also the biggest product launch we've ever had,"
Adcock told Bloomberg Television's Ed Ludlow.
"You Can't Code Your Way Out of This Problem"
Adcock directly addressed the complexity of the tasks shown in the new logistics demo, using it to explain why Figure is so focused on its end-to-end Helix AI model rather than traditional programming or teleoperation.
"With packages, it's really tough," he explained. "Every single package, in every single instance of packages coming together, is different for the robot every single time. And you cannot solve this with code. You can't code your way out of this problem. It's impossible."
"Can you find me anybody in the world doing this?" Figure CEO @adcock_brett issues a confident challenge while discussing their latest logistics demo at a @Bloomberg event. Watch him explain how their Helix AI autonomously handles diverse, deformable packages at human speeds— a
This, he argued, is the fundamental difference between structured tasks, like some seen in automotive manufacturing, and the chaotic nature of logistics. "They're almost orthogonal," he said of the two types of problems. While one requires high precision and speed, the other demands "generalization from the neural net to figure out how to manipulate packages of varying sizes at human speeds." This AI-first approach is also what allows the robot to handle delicate, unseen objects. When asked if the robot could pick a strawberry without crushing it, Adcock replied confidently, "Oh yeah. We can go one better. We can probably pick up a strawberry and manipulate and do things without ever having seen strawberries before in training sets."
Path to 100,000 Robots
Adcock reaffirmed his ambitious goal of shipping 100,000 robots within a four-year period. He addressed the skepticism around this target by reframing the manufacturing challenge.
"One thing I've realized over the last few years is we're not doing car manufacturing. We're doing consumer electronics manufacturing," he said. "That process is probably several orders of magnitude easier." He revealed that the manufacturing line for the next-generation Figure 03 is already being built out, and that the new model is "literally 90% cheaper" than the Figure 02.
This rapid scaling is central to Figure's strategy, which Adcock described as a potential "winner take all" market. He believes the company that can deploy a large, useful fleet of robots first will gain an insurmountable data advantage. "The robots will learn to get better every day... and they will share that with the collective fleet," he explained. "You're going to want the smartest employee... And that will also be the cheapest." Answering a question about whether he needs more GPUs or more robots, he was unequivocal: "We have a data problem and a robot problem."
Real-World Deployments and Vertical Integration
Adcock confirmed that Figure robots are working "every day in the commercial workforce today," referring to the ongoing deployment at BMW's Spartanburg plant. He emphasized the value of this partnership for learning how to operate reliably on a client's site with "as little or no human interventions per day."
He also confirmed that Figure has a "second customer assigned in the logistics space" and is working on initial deployments there, lending credence to earlier reports about talks with UPS.
To achieve this, Adcock stressed the necessity of a full-stack approach. "We basically have found that there's no other way out but through. We basically have to design all the hardware ourselves, all the electronics, all the software, all the AI models, training, we do manufacturing, we integrate it, we bring it into service, and we make sure it works." This was a key reason for moving away from their earlier AI collaboration with OpenAI. "We were just better at doing it alone," he stated, adding that the partnership created a perception that another company was doing their AI, which "wasn't really true."
Despite his bold claims and rapid progress, Adcock acknowledged the skepticism he and Figure face, attributing his relative lack of public appearances to a "heads-down" focus on execution. "I really feel I need to be at the office, just building the team, building the product, trying to get it out into the world," he said. "And everything else is basically a distraction."