Ex-Tesla Optimus Scientist Unveils European Humanoid Robot Startup
A former Tesla Optimus engineer is building a humanoid robot startup in Paris, betting Europe's labor shortage and industrial base will outpace Silicon Valley's speed.
A former Tesla Optimus engineer is building a humanoid robot startup in Paris, betting Europe's labor shortage and industrial base will outpace Silicon Valley's speed.
Rémi Cadène, who spent three years developing AI systems for Tesla's Optimus robot and driver-assistance software, has launched UMA, a Paris-based humanoid robotics startup. The company unveiled its Northstar prototype and claims to be in talks with 50 potential customers, marking a significant shift in the humanoid robot race: a credible Tesla insider is now competing directly, but from Europe rather than the US.
What is UMA building and who backs it?
UMA, short for Universal Mechanical Assistant, emerged from stealth in December 2025 with Cadène as CEO and three co-founders including former Hugging Face engineer Simon Alibert and robot designer Rob Knight. The startup is backed by venture firms Greycroft, Relentless, and Unity Growth, plus angel investors including Yann LeCun, Datadog CEO Olivier Pomel, and Hugging Face co-founder Thomas Wolf.
Cadène's path to UMA runs through Hugging Face, where he led development of LeRobot, an open-source robotics library that grew to over 12,000 GitHub stars in roughly a year. That software expertise appears to be UMA's core strength. The company is targeting general-purpose, AI-powered humanoids for manufacturing plants, logistics warehouses, and eventually homes.
Why is Europe the beachhead market?
UMA is deliberately avoiding the US and China, where most humanoid competitors are racing. Instead, Cadène is betting on Europe's aging workforce, high labor costs, and dense industrial base. The continent has both the economic pressure to adopt automation and a deep talent pool of engineers from DeepMind, Tesla, Nvidia, and Hugging Face.
The company plans to launch industrial pilot programs as early as 2026. This is a contrarian move in a field dominated by American players like Tesla and Figure, and Chinese manufacturers like Unitree that are competing on cost and speed. UMA is wagering that Europe's manufacturing depth and automation appetite create an opening for a homegrown competitor.
The biggest bottleneck with humanoid robots appears to be software: having the robots understand and interact intelligently and autonomously with their physical environment.
How does UMA compare to Tesla's Optimus?
Tesla is targeting low-volume Optimus production at Fremont this summer, with CEO Elon Musk claiming it will become the company's most valuable product. But execution lags rhetoric. Musk admitted in January that no Optimus robots are doing useful work at Tesla in any material way, and the company has announced no external customers.
Figure AI is further ahead on actual deployment. Its robots have run shifts at BMW's Spartanburg plant, helping build tens of thousands of vehicles. The Hyundai and Boston Dynamics partnership has also demonstrated real commercial work.
UMA has no shipping product yet. What it has is a credible team with deep robotics and AI experience, significant investor backing, and a claimed pipeline of 50 customers evaluating use cases. In a field where demos routinely outrun deployments, whether that pipeline converts into paid pilots this year is the metric that matters most. Based on images shared so far, the hardware appears early stage, but software is where UMA claims its advantage lies.
Why would a Tesla Optimus engineer leave to start a competing humanoid robot company?
Cadène spent three years at Tesla working on Optimus AI systems, then moved to Hugging Face where he built LeRobot, an open-source robotics library that became core infrastructure across the field. That experience gave him both deep knowledge of what Tesla was doing and exposure to the broader robotics ecosystem. Starting UMA in Europe lets him pursue a different market strategy (targeting labor-constrained Europe rather than racing on cost in the US or China) with a team of proven engineers and significant backing from top-tier investors.
Is UMA actually ahead of Tesla on humanoid robots?
No. Tesla is further along on hardware production, with plans to start low-volume manufacturing this summer. UMA has no shipping product yet. However, UMA's advantage appears to be in software and AI, which is actually the bigger bottleneck in humanoid robotics right now. Figure AI is currently the furthest ahead on real-world deployment, with robots running shifts at BMW's plant. UMA's credibility rests on its team's track record and whether its claimed 50-customer pipeline converts into paid pilots.
Why would customers choose a European startup over Tesla or Figure?
UMA is targeting Europe specifically, where labor costs are high and the workforce is aging. European manufacturers have deep automation expertise and are under real pressure to adopt robotics. UMA's team includes engineers from Tesla, Hugging Face, and DeepMind, giving it credibility on the software side. The startup is also betting that European customers prefer a European vendor over American competitors, and that the continent's industrial base and talent pool give it a structural advantage that US-focused competitors may overlook.