The Steel Bot Manifesto
Three days ago, the FCC banned new foreign-made humanoid robots from the American market. Overnight, the affordable, programmable platforms that American researchers, hobbyists, and startups have been building on were cut off from their future. And there is still no American robot you can simply buy, program, and truly own. Steel Bot exists to change that.
Last year I watched Chinese humanoid companies begin to ship and sell affordable humanoids that were starting to do impressive things. I remember telling colleagues in the American humanoid industry how impressed I was with their approach. The responses I got were dismissive: "Their hardware is not very good"; "They overheat all the time"; "The gears in their actuators often strip"; "They can't do anything useful"; and so on.
None of these answers satisfied me. It seemed obvious that by shipping and selling their robots, and by not being afraid to let them break and fail publicly, these companies would learn quickly and gain the real-world feedback they needed to rapidly iterate and improve their hardware. It was equally clear that selling open and programmable robots was the right strategy, because it was already unleashing the creativity of the hackers, hobbyists, researchers, and entrepreneurs who were finding all kinds of interesting and unique ways to use and control the robots. You could see progress in niche applications like robot boxing, security patrolling, and remote inspection, where the robots were already economically viable and making money. Some said in response that these were tiny markets and that demand for humanoids was still minuscule. But history shows, time and again, that a new technology or platform nearly always starts small in both quantity sold and scope of usefulness.
Once I was convinced that the Chinese playbook of selling affordable, open humanoids was the best approach, I began looking for an American company to join that was following the same model. But I could not find a single one I believed had a realistic shot at displacing the Chinese robots and offering a solid American alternative that was programmable and available to purchase. Every American humanoid company seemed to be aiming straight for the promised land: a fully capable and fully intelligent general-purpose robot that can do anything you ask it to do. I agree that that future will come someday, but I think it is further off than many in the humanoid space would like to admit.
The problem with promising to deliver on that lofty goal (and fundraising based on reaching it) is that it becomes very dangerous to let your robot go out into the real world and potentially fail publicly. You have to maintain the mirage that the promised land is just around the corner, and that you still have a realistic path of getting there before your funding runs out. It becomes almost impossible to actually ship and sell your robot, to let others try it out unsupervised, to field-test your hardware and software in the wild. The chance of serious failure and negative press is too great, so the robot stays in the lab, running pre-tested and pre-scripted demos. To me, this looked like the wrong way to reach general capability and was doomed to fail, much like the self-driving car companies of 2015 and 2016 that kept insisting full autonomy was imminent. Most of them ran out of money and died long before delivering a working self-driving car. The few that survived using the closed, perfectionist approach were bankrolled by massive and profitable parent businesses that could absorb a decade of losses. Tesla sold cars, Waymo had Google, and Zoox had Amazon. No humanoid startup has that luxury.
When I could not find an American company making an affordable and programmable humanoid with a realistic shot at competing with the Chinese robots, it gradually became clear that I would need to take on this problem myself. I founded Steel Bot in December 2025 and set out to design and build an affordable and maximally open humanoid platform that would outcompete the Chinese robots in every meaningful way.
Which brings us back to this week. To be precise about what the ban does: it does not recall the Chinese robots already authorized and sold in the United States, but it closes the American market to every new model going forward. We are accelerating our plans and our timeline to fill this gap. Our first robots will ship early next year. American researchers, startups, and builders deserve an advanced, open, bipedal humanoid platform they can purchase, program, modify, and freely use however they see fit because that is how America leads the global robotics race.
What We Believe
At Steel Bot we believe in openness. Our humanoids are fully programmable: we give our customers root-level access along with completely open-source software. Our customers can freely program their robots and customize them for whatever purpose they see fit, without seeking approval or permission from Steel Bot. Above all, we believe in freedom, and we hold the strong conviction that unhindered creativity will advance the field of robotics as quickly as possible. We want our customers to tinker with our robots, modify them, do research on them, and even build their businesses on top of them — without fear that we will gatekeep or restrict them in any way.
At Steel Bot we believe in true ownership. Our customers have the right to physically modify and repair their robots. Humanoid robots will be used in countless, previously unimaginable ways, and for many applications our customers will need to modify and enhance their machines in ways we expect and in ways we never could. Our robot will have a robust electromechanical connector at the wrist, with published specifications, so that our customers can attach any brand or style of hand, gripper, tool, or other end-effector they need. We will open-source this connector design so that other companies can use it on their robots, and we hope it will become an industry standard, so that anyone can build interchangeable accessories that work across multiple robot platforms. We have also designed our robot to be as modular and easily repairable as possible. We will sell every part of the robot individually, ship the robot with a toolbox with every tool you need for repair, and publish detailed repair instructions and videos, so that fixing your Steel Bot is as easy and affordable as possible.
At Steel Bot we believe in building things that last. We will fight with all of our power against the practice of designing products to fail (planned obsolescence) in which cheap materials and shoddy construction deliberately limit a product's useful life and force customers into regular upgrades. To show that we mean business, we will offer a 100-year warranty on our robots. This may sound excessive, but it will compel us to design for maximum robustness and longevity.
Skeptics will point out that few companies live to be 100, and they are right. That is exactly why everything about our robot is open. Because our designs are published, our parts are for sale, and our repair guides are free, your Steel Bot can be kept running for a century whether or not Steel Bot is around to help. Many great American products built a century ago still work today, and we believe that was no accident. America once held the highest quality standards in the world. What eroded them was not a loss of skill but a shift in incentives: the short-term mindset of Wall Street that rewards cheap, fragile products over things built to last. We will reverse this trend with our robots.
At Steel Bot we believe in the freedom to compute. We will press forward with maximum urgency to improve and refine the built-in software and intelligence that ships with our robots, so that anyone can use them easily straight out of the box. But our customers will always be able to run whatever operating system, AI model, and software they wish onboard our robots. We know this freedom is what our customers want, and we believe it is the best defense against a dangerous centralization of power that could one day put robots to work for mass surveillance or mass physical harm. Because our software is open-source, it can be forked and modified for all kinds of uses, and all kinds of companies can build their businesses on top of Steel Bot as a malleable and open platform. We believe open-source software and intelligence provide the best possible security and minimize the risk of espionage or unauthorized remote control of our robots.
Can we prevent every evil or unintended use of an open robot? No, no more than Apple can prevent a MacBook from being used to do harm. That is the nature of every general-purpose tool, from the personal computer to the pickup truck. Responsibility for misuse lies with those who misuse. And we are convinced that a world of open, transparent, widely distributed robots is far safer than a world where a handful of companies control every machine.
At Steel Bot we strongly believe that affordable, programmable, modifiable, bipedal humanoids made with exemplary quality and care, running open-source software and intelligence are the best way to advance the American robotics ecosystem and to ensure a bright and flourishing future for ourselves and our children.
If you agree with this vision and want to get involved, or would like to join us at Steel Bot, please reach out to me at randall@steel.bot. I look forward to hearing from you.
Randall Briggs
Founder, Steel Bot, @steelbot_