One consensus among robotics entrepreneurs and investors is that the “ChatGPT moment,” where a versatile robot can effortlessly adjust to any setting, has not yet occurred. Perspectives differ on whether this significant advancement is imminent or still a decade away.

Feather Robotics, a humanoid-focused startup established last year, is creating a robot suitable for either scenario. In contrast to Tesla or Figure, which pursue the ambitious goal of creating both the physical structure and an underlying “brain” that isn’t universally available yet, Feather offers a toolkit for hardware and software that enables developers to address real-world challenges immediately.

“You can’t purchase a Tesla robot today and build on it,” Feather co-founder Hoa Mai shared with TechCrunch. “We recognized that this isn’t how most firms achieve success. Hardware companies like Nvidia or Apple began with a functional, deployable product and then gradually built up complexity.”

In 2025, after his previous humanoid enterprise was acquired by 1X, Mai (depicted above, left) collaborated with Parsa Bakhtiari (shown above, right), a former engineer from Tesla’s Model 3 team who previously reported directly to Elon Musk, to create what they envision could become the Android of robotics.

This concept was appealing to Gradient Ventures, who supported Feather at its founding, leading the startup’s earlier disclosed $7.6 million pre-seed financing round.

Since then, Feather has introduced a modular robotic system that enables developers to tailor the hardware for various applications, such as varying arm lengths. The startup has already initiated sales to clients, exceeding $1 million in revenue.

While Feather is not revealing its client list, it mentioned that its robots are functioning as chefs in eateries and tidying science laboratories. On the software front, the hardware is compatible with models from any top robotics AI developer, including Nvidia, Skild, or Physical Intelligence.

“We have been selling limited quantities of these robots, and now that we’ve addressed nearly all issues from the past year of field testing, we’re preparing for a major product launch,” Mai stated.

As per Darian Shirazi, general partner at Gradient, Feather faces no direct competition in the U.S. He categorizes the existing hardware robotics market into three segments: startups like Sunday creating domestic robots, major companies like Figure and Tesla pursuing general-use machines, and Feather, which he believes is the sole U.S. startup focusing on a modular humanoid platform.

Mai acknowledges that Feather drew inspiration from Chinese robotics firms like Unitree. However, with new foreign models restricted from entering the U.S. market, Feather is now distinctly situated as a domestic contender of its type.

Another advantage for Feather is its cost. The robot is priced at $30,000, approximately half of what Unitree’s H2 Edu costs.

Shirazi perceives that the market potential at this price point could be significant. “You would employ a worker for $50,000 to $60,000 annually, you’d need to train [them]. They would require HR. They would need a variety of support systems,” he noted. “You can now acquire a Feather robot to fulfill that role.”

Feather is also functioning with great efficiency, having utilized only a small portion of its pre-seed funding, as per Shirazi.

The startup’s major hope, of course, is that the future value in robotics will stem from an ecosystem surrounding the hardware.

“If we consider the current market for physical AI companies, it’s rather limited,” Mai remarked. “But if we contemplate the number of physical AI application firms that could emerge in five years, we anticipate it to be in the thousands.”

Feather aspires to be the platform that powers all of them.