Ex-Tesla Optimus Scientist Rémi Cadène Unveils UMA's Northstar, a Paris-Built Humanoid Robot Betting on Europe First
UMA, the Paris startup led by former Tesla Optimus and Hugging Face engineer Rémi Cadène, has emerged from stealth to unveil Northstar, a lightweight humanoid robot aimed at European factories and warehouses.
Overview
Rémi Cadène, an engineer who helped build the AI behind Tesla’s Optimus humanoid and later led Hugging Face’s open-source robotics toolkit, has surfaced with a Paris startup called UMA and unveiled its first humanoid robot, Northstar. According to The Next Web, the company — short for Universal Mechanical Assistant — is pitching a lightweight, general-purpose machine for manufacturing plants, logistics warehouses, and eventually homes. Electrek reports the company is based in Paris and emerged from stealth in December 2025.
What We Know
Cadène is co-founder and chief executive of UMA, The Next Web reports. He spent roughly three years at Tesla, from 2021 to 2024, working on the AI behind Autopilot and building the first neural networks for Optimus, according to the same outlet. He left in early 2024 to join Hugging Face, where he led the development of LeRobot, an open-source robotics library, Electrek reports.
The founding team draws on that lineage. The Next Web names Pierre Sermanet, a veteran of Google DeepMind and NYU, as chief science officer; Simon Alibert, a LeRobot co-founder, as chief technology officer; and Robert Knight, the designer behind the widely used open-source SO-100 arm, as chief robot officer.
The robot itself is described as a lightweight, general-purpose, AI-powered humanoid, and Electrek reports that UMA completed a prototype it calls Version 0 in nine months with a small team, designed and assembled in Paris. There is no shipping product yet, according to the same outlet.
UMA’s strategy is explicitly regional: Europe comes first, before any push into the United States or Asia, both The Next Web and Electrek report. Electrek reports the company already counts about 50 potential customers in its evaluation pipeline, while The Next Web says UMA plans several pilot programmes in logistics, manufacturing, and healthcare during 2026.
The company has assembled a notable set of backers and advisers. Both outlets name Greycroft among its venture investors; The Next Web additionally lists Red River West, Kima Ventures, and Factorial, while Electrek names Relentless and Unity Growth. Yann LeCun, Meta’s chief AI scientist and Turing Award winner, and Hugging Face co-founder Thomas Wolf are both attached to the company, per The Next Web; Electrek also lists Datadog chief executive Olivier Pomel as an angel investor.
Cadène framed the opportunity around European labour economics. “Labor costs are very high and, given the demographic trends, there will be significant demand,” he said, according to Electrek.
What We Don’t Know
The size of UMA’s funding is unresolved. Cadène was reported to be seeking around $40 million in seed funding, but the company has not confirmed a final figure and the size of any closed round remains unclear, The Next Web reports. Neither cited outlet published verified performance specifications for Northstar — such as payload, runtime, or price — or a firm date for when the robot will move from pilots to commercial deployment.
Analysis
UMA joins a growing cohort of European players trying to keep the continent competitive in humanoid robotics against heavily funded US and Chinese rivals. The Machine Herald has previously reported on Renault’s plan to deploy hundreds of Wandercraft humanoid robots across its factories, one signal of European industrial appetite for the technology. UMA’s pitch leans on the same demographic argument — aging workforces and high labour costs — that Cadène cited, and on a founding team pulled from the open-source robotics community he helped build at Hugging Face. Whether that pedigree translates into hardware that survives real factory floors, rather than staged demonstrations, is the question its 2026 pilots are meant to answer.