UMA Unveils Northstar Humanoid Robot, Targets Europe First

Europe has been the quiet slot on the humanoid robotics map, so a Paris team stepping into it with credible Tesla and DeepMind pedigree changes the shape of the field a little. Bloomberg reported that Rémi Cadene, previously a research scientist on Tesla's Autopilot and Optimus programs and most recently a principal researcher at Hugging Face, has unveiled plans at his new startup UMA to build a lightweight humanoid called Northstar aimed at manufacturing plants, logistics warehouses and homes.

The commercial pitch is that UMA is already in talks with 50 potential customers about use cases, with several pilot programs in logistics and manufacturing planned for 2026, and that the robots will target Europe first. UMA, short for Universal Mechanical Assistant, came out of stealth in December 2025 with a leadership team that reads like a supergroup: Pierre Sermanet, a founding member of the robotics team at Google Brain and DeepMind, as chief science officer; Simon Alibert, a LeRobot co-founder, as CTO; and Robert Knight as chief robot officer. Yann LeCun, Meta's chief AI scientist and a Turing Award winner, sits on the advisor bench.

Why this matters even if you do not care about robots specifically: Europe has been visibly absent from the front row of the humanoid race, which has been led by U.S. and Chinese contenders. A team with Optimus and DeepMind CVs, choosing Paris rather than the Bay Area, is the kind of signal European industrial buyers and policymakers have been waiting for on data-sovereignty and supply-chain grounds. UMA also says the hardware will be lightweight and repairable, which is a different bet from the heavy industrial humanoids being pitched at automotive lines today.

The honest caveats are the usual ones. 'In talks with 50 customers' is a soft number, not booked pilots, and the reporting does not give hard specs for Northstar (payload, price, autonomy), a headline funding figure, or which named European manufacturers are in the pipeline. Humanoid timelines have slipped for better-funded competitors, and European labor and safety rules around factory-floor robots are not trivial.

The forward-looking bit is who benefits if UMA delivers something close to what it is promising. European logistics and manufacturing groups get a homegrown supplier as an alternative to importing hardware from the U.S. or China, and the LeRobot lineage on the founding team hints at a more open, dataset-driven development loop than closed-shop rivals are running.