Android of Robotics: Feather Builds a Platform for Developers
The Android of robotics is the vision behind Feather Robotics, a humanoid startup that wants developers to build real-world applications on affordable hardware today, not years from now.
- Android of Robotics: Feather Builds a Platform for Developers
- Why Feather Chose a Developer-First Approach
- Traction: Revenue, Customers, and Field Testing
- The $30,000 Price Point and Its Market
- Learning From Unitree and the U.S. Market
- The Big Bet: An Ecosystem Around Hardware
- What Developers Should Watch Next
Feather was founded in 2025 by Hoa Mai and Parsa Bakhtiari. Mai previously sold a humanoid startup to 1X. Bakhtiari is a former Tesla Model 3 engineer who at one point reported directly to Elon Musk. Together, they are building what they hope becomes the Android of robotics: a modular platform that others can customize and extend.
Their timing reflects a split in the industry. Robotics founders and investors agree that a “ChatGPT moment” for general-purpose robots has not arrived. They disagree on whether it is close or a decade away. Feather is designing for either timeline.
Why Feather Chose a Developer-First Approach
Companies like Tesla and Figure are pursuing a demanding strategy. They are building both the robot body and a foundational “brain” that is not yet widely deployed. That approach leaves developers with little to work with right now.
Mai explained the problem simply in an interview. “You can’t buy a Tesla robot today and develop on top of it,” he said. He pointed to how successful hardware companies grew. Nvidia and Apple started with products that worked and could be deployed. They added complexity over time.
Feather applies that lesson to humanoid robots. Instead of chasing a fully general machine, it sells a toolkit. Developers get hardware and software they can use for actual tasks while the broader technology matures.
A Modular Robot You Can Customize
Feather’s system is modular by design. Developers can adjust the hardware for different use cases, including changing arm lengths. That flexibility matters for teams working on specialized jobs rather than one universal robot.
The software side is equally open. The hardware can run models from any leading robotics AI provider, including Generalist, Skild, or Physical Intelligence. Developers are not locked into a single AI stack.
Traction: Revenue, Customers, and Field Testing
Feather has already moved past the concept stage. The startup has begun selling to customers and surpassed $1 million in revenue. It is not disclosing its customers, but it shared some examples of deployments.
Its robots are working as cooks in restaurants in Japan. They are also cleaning up science labs. These are practical, repetitive tasks that fit the current capabilities of the hardware.
Mai said the company has been selling small quantities of these robots. Over the past year of field testing, it resolved almost all the issues that surfaced. “We’re getting ready for a big product launch,” he said.
That field-testing period matters for developers. A platform is only as useful as its reliability in real environments. Feather’s early customers are effectively proving the hardware under real conditions.
The $30,000 Price Point and Its Market
Feather’s robot costs $30,000. According to the company, that is about half the price of Unitree’s H2 Edu. Price is central to Feather’s pitch because it changes the math for employers.
Darian Shirazi, general partner at Gradient, framed the comparison in labor terms. “You would hire a laborer for $50,000 to $60,000 a year, you’d have to train [them],” he said. “They’d have to have HR. They’d have to have all these different support mechanisms.” A Feather robot, in his view, offers an alternative at a comparable upfront cost.
Shirazi believes the market size at that price could be very large. He also noted that Feather is operating efficiently, having spent only a fraction of its pre-seed funding.
Backing From Gradient Ventures
Gradient Ventures backed Feather at its inception. The firm led the startup’s previously announced $7.6 million pre-seed round. The vision of an Android of robotics was a key part of the appeal.
Shirazi said Feather does not have direct competitors in the U.S. He divides the hardware robotics landscape into three groups. Startups like Sunday are building domestic household robots. Heavyweights like Figure and Tesla are aiming for general-purpose machines. Feather, he believes, is the only U.S. startup developing a modular humanoid platform.
Learning From Unitree and the U.S. Market
Mai admits that Feather took a page from Chinese robotics companies like Unitree. That influence shows in the focus on affordable, deployable hardware rather than distant grand ambitions.
The competitive picture has shifted, though. New foreign-made models are restricted from entering the U.S. market. That leaves Feather uniquely positioned as a homegrown option of its kind.
For U.S. developers, that matters. They get a domestic platform with local support and no import complications. It also gives Feather room to grow without direct competition from the low-cost Chinese models that inspired parts of its approach.
The Big Bet: An Ecosystem Around Hardware
Feather’s core bet is that value in robotics will come from an ecosystem built around the hardware. That is the Android model: a common platform that thousands of developers extend in ways the original creator never anticipated.
Mai is explicit about the scale he expects. “If we think about the market size today for physical AI companies, it’s very small,” he said. “But if you think about how many physical AI application companies will probably exist in five years, we expect it to be in the thousands.”
Feather hopes to be the platform that powers all of them. If that happens, the Android of robotics would not be a single robot. It would be the base layer that countless robots are built on.
What Developers Should Watch Next
Feather has signaled that a big product launch is coming. The company has spent a year fixing issues found in field testing, and it now has paying customers in restaurants and labs.
For developers evaluating the platform, several factors stand out:
Modular hardware that can be adapted to different tasks and form factors
Open software that runs models from multiple robotics AI providers
A $30,000 price point that lowers the cost of entry
U.S.-based manufacturing and support amid restrictions on foreign models
Early real-world deployments that validate reliability
The “ChatGPT moment” for general-purpose robots may still be years away. Feather’s answer is to let developers build useful things in the meantime. If the Android of robotics succeeds, the applications will come from the ecosystem, not from Feather alone.
