If there’s one factor robotics founders and traders agree on, it’s that the “ChatGPT second” the place a general-purpose robotic can seamlessly adapt to any atmosphere has not arrived. Opinions fluctuate on whether or not the breakthrough is simply across the nook or a decade away.
Feather Robotics, a humanoid startup based final yr, is creating a robotic for both timeline. In contrast to Tesla or Determine, that are tackling the lofty mission of constructing each the physique and a foundational “mind” that isn’t but extensively deployed, Feather offers the {hardware} and software program toolkit for builders to sort out real-world duties now.
“You’ll be able to’t purchase a Tesla robotic in the present day and develop on prime of it,” Feather co-founder Hoa Mai informed TechCrunch. “We realized that this isn’t how most firms turned profitable. {Hardware} gamers like Nvidia or Apple began with a product that labored and was deployable, after which they added complexity over time.”
In 2025, after his earlier humanoid startup was bought to 1X, Mai (pictured above, left) teamed up with Parsa Bakhtiari (pictured above, proper), a former Tesla Mannequin 3 engineer who at one level reported on to Elon Musk, to construct what the duo hopes will turn out to be the Android of robotics.
That imaginative and prescient was engaging to Gradient Ventures, which backed Feather at its inception, main the startup’s beforehand introduced $7.6 million pre-seed spherical.
Since then, Feather has rolled out a modular robotic system that enables builders to customise the {hardware} for various use instances, corresponding to adjusting arm lengths. The startup has already begun promoting to clients, surpassing $1 million in income.
Though Feather just isn’t disclosing its clients, it stated that its robots are working as cooks in eating places and cleansing up science labs. On the software program aspect, the {hardware} can run fashions from any main robotics AI supplier, together with Nvidia, Skild, or Bodily Intelligence.
“We’ve been promoting small portions of those robots, and now that we resolved like virtually all the problems during the last yr of area testing, we’re preparing for a giant product launch,” Mai stated.
Based on Darian Shirazi, normal accomplice at Gradient, Feather doesn’t have direct rivals within the U.S. He breaks down the present {hardware} robotics panorama into three classes: startups like Sunday constructing home family robots, heavyweights like Determine and Tesla aiming for general-purpose machines, and Feather, which, he believes, is the one U.S. startup creating a modular humanoid platform.
Mai admits that Feather took a web page from Chinese language robotics firms like Unitree. However with new foreign-made fashions restricted from getting into the U.S. market, Feather is now uniquely positioned as a homegrown possibility of its sort.
Feather’s different benefit is its value. The robotic prices $30,000, which is about half the value Unitree’s H2 Edu.
Shirazi believes the market measurement at that value could possibly be very giant. “You’d rent a laborer for $50,000 to $60,000 a yr, you’d have to coach [them]. They’d should have HR. They’d should have all these completely different help mechanisms,” he stated. “Now you can purchase a Feather robotic to do this.”
Feather can be working extraordinarily effectively, having spent solely a fraction of its pre-seed funding, based on Shirazi.
The startup’s large wager, in fact, is that the worth in robotics will come from an ecosystem developed across the {hardware}.
“If we take into consideration the market measurement in the present day for bodily AI firms, it’s very small,” Mai stated. “But when you concentrate on what number of bodily AI utility firms will most likely exist in 5 years, we count on it to be within the 1000’s.”
Feather hopes to be the platform that powers all of them.
If you buy by hyperlinks in our articles, we could earn a small fee. This doesn’t have an effect on our editorial independence.