Jay Li, founder of robotics startup Proception, has learned a hard lesson: avoid getting sued by Tesla while trying to launch a company. Yet he believes the experience ultimately made his venture stronger.
“I think it’s kind of like a resilience test, or pressure test,” Li told TechCrunch in an exclusive interview. “People say that what doesn’t kill you makes you stronger, right?”
Li, a former technical lead on Tesla’s Optimus humanoid robot program, was accused by his former employer last year of misappropriating trade secrets to start Proception. After months of legal battles, he reached a settlement with Tesla, which dismissed the lawsuit earlier this month. (Tesla did not respond to a request for comment.)
From Litigation to Innovation
With the legal matter behind him, Li is now focused on what he considers an even greater engineering challenge: making robot hands function with human-like dexterity.
To accelerate that mission, Proception announced Monday that it has raised an $11 million seed round led by First Round Capital, with participation from Y Combinator and early-stage fund BoxGroup.
The company also announced it is shipping the first batch of its “high-dexterity robotic hand” to researchers and robotics firms, while opening up to broader orders. Li’s goal is for Proception to become the leading hand supplier for companies that lack the time or resources to develop “dexterous manipulation” in-house.
As of 2026, the humanoid robotics market continues to attract massive investment, with global funding projected to exceed $8 billion annually. Yet Li believes too little of that capital is directed toward perfecting the end-effector—the hand.
The Industry Consensus: Hands Are Hard
One of the most prominent voices highlighting this challenge is Li’s former boss, Tesla CEO Elon Musk, who has repeatedly stated that robot hands represent one of the hardest engineering problems still unsolved in robotics.
While Musk has predicted that Optimus robots could begin working in Tesla factories within a few years, most experts believe we remain years away from robotic hands matching human capability. Kevin Lynch, director of Northwestern University’s Center for Robotics and Biosystems, told The Wall Street Journal last year that his team estimates it will take a decade before robotic hands are “functional and useful and able to do some of the things that humans do.”
Li believes Proception can achieve this far faster, largely due to its data-collection strategy.
A Data-Driven Approach to Dexterity
Most companies training humanoid robots rely on teleoperation: a human wearing a VR headset sees what a robot sees and remotely manipulates objects in its environment. The robot then learns from these human commands.
Li identifies two major drawbacks to this method. First, the teleoperator receives no tactile feedback from the objects the robot touches. Second, the approach is limited by the number of robots a company can deploy at any time.
Proception’s alternative is a sensor-laden glove worn by human testers. As they manipulate objects naturally, the glove captures rich haptic and motion data. This data is then used to train Proception’s robotic hands directly, bypassing the teleoperation bottleneck.
By decoupling data collection from robot hardware availability, Proception aims to scale its training pipeline rapidly. The company’s first customers include academic research labs and early-stage robotics firms, with broader availability expected in late 2026.
Looking Ahead
With the Tesla lawsuit resolved and fresh capital in hand, Proception is positioned to tackle one of the last frontiers in humanoid robotics. If Li’s bet on data-driven dexterity pays off, the startup—and not its deep-pocketed former employer—could be the one to crack the code on truly human-like robot hands.
via TechCrunch AI