Former Tesla Optimus chief Ashish Kumar launches new industrial robot startup
Ashish Kumar ran Tesla's Optimus humanoid program. Now he's building his own robot company, and he's not starting small.
According to reporting by The Information, Kumar has launched Intelligent Machines in San Francisco and is looking to raise approximately $100 million in a seed round. Several former members of Tesla's humanoid team have reported joined him.
The target: industrial tasks that require the kind of skill most robots don't have, like high-tech manufacturing and aerospace ... the jobs where a mistake costs more than a productivity metric.
Why "intelligent" is the operative word
In a way, Kumar's pitch seems almost a little retro, a little old-school automation. But the difference is in mass production versus skilled work.
Most humanoid startups pitch general-purpose robots: give us a warehouse, a factory floor, a distribution center, and we'll figure out the tasks. Intelligent Machines is targeting specifically skilled work, the kind where the premium is on precision and judgment, not just repetitive throughput.
Aerospace manufacturing is the canonical example of this gap.
Assembling aircraft components requires consistent force application, spatial awareness, and the ability to work in tight tolerances across irregular surfaces. These aren't tasks you can solve with warehouse-style pick-and-place training data. They need a different approach.
Kumar's approach is reinforcement learning: rewarding robots for successful task completion, penalizing specific failure modes. The training paradigm is designed around the outcome, not around mimicking human motion ... which means the robot can find solutions that don't look human, as long as they work.
A Tesla Optimus talent dispersal?
Kumar is not the only Optimus veteran striking out on his own.
Rémi Cadène, who also worked on Optimus and built Hugging Face's LeRobot framework, unveiled UMA's Northstar humanoid in December 2025, a European industrial humanoid backed by Greycroft, Relentless, and Unity Growth, with Yann LeCun among the angel investors.
This pattern has been building for a while.
The Optimus program has been, among other things, a training ground: a place where engineers worked on the hardest robotics problems at scale, with real production pressure. The people who came through that program know what it takes to actually ship a robot, not just demo one.
If they leave, they take that knowledge with them.
What $100 million buys in industrial robotics
A $100 million seed round would be large even by 2026 standards. It signals an expectation that industrial robotics for skilled tasks requires substantial compute, hardware iteration, and customer development before a product can be proven.
The comparison point is companies like Apptronik, which has spent years and hundreds of millions developing Apollo for industrial use, and Persona AI, which raised $42 million for its shipyard welding humanoid. Intelligent Machines' seed target puts it in a different league, suggesting either a more hardware-intensive path or a faster timeline to commercial deployment than either of those.
The skilled manufacturing and aerospace markets are also unusually willing to pay for quality. A robot that reliably assembles aircraft components at human-level precision could command a premium that no warehouse automation robot approaches. The total addressable market is smaller than logistics, but the price-per-unit economics are different.
The question the raise will answer
This is very early.
Intelligent Machines hasn't publicly shown a robot and the raise is still in progress. What we know: the founding team has credible engineering credentials, the target market is high-value and underserved, and the reinforcement learning approach differs from the imitation-learning path most humanoid companies are pursuing.
Whether that differentiation translates to a real product — and whether skilled manufacturing turns out to be a better early market than logistics — is the $100 million question. The Tesla Optimus program is where some of the best robot engineers in the world spent the last several years. It's worth paying attention to where they go next.
Sources
Written by
HMH