RobCo crosses a $1 billion valuation in nine months
Modular arms sold as a service get a unicorn tag, while MicroAGI names Animesh Garg as Chief Research Officer and bets on post-training.
Munich robotics startup RobCo crossed a $1 billion valuation just nine months after its last round, according to Lukas Ziegler, who posted the news. Existing backers Sequoia Capital and Lightspeed doubled down. Cherry Ventures and the European Tech Collective, a group of European tech founders, came in as new investors.
RobCo was founded in 2020 by Roman Hölzl, Paul Maroldt and Constantin Dresel. The product is modular robot arms that, as Ziegler puts it, snap together like building blocks, sold to mid-size manufacturers on a Robotics-as-a-Service model. That is the unglamorous end of the field, and it is where the money landed.
The US is now RobCo's fastest-growing market, with production in Austin and a lab in San Francisco.
The US is now RobCo's fastest-growing market with production in Austin and a lab in San Francisco.
What is not shipping yet
The thing people will want to see is Alfie, described as an autonomous industrial robot designed to perceive its surroundings, reason about what needs to happen, and act independently, aimed at unstructured and safety-critical tasks. Nothing of it has been shown. The official unveiling is set for March 2027 in Munich, so treat the capability description as a company pitch rather than a demo.
Ziegler frames the round as evidence that European robotics is picking its own lane. He notes China still leads in humanoids with 90% of the global market, and argues Europe is building on the factory floor and in the mid-market instead.
MicroAGI hires Animesh Garg and picks a bottleneck
Separately, MicroAGI announced Animesh Garg as its new Chief Research Officer, with the team from Usable Robots joining him. The company's argument is that frontier AI works, hardware is getting cheap, and the industry is learning to collect robot data at scale, so the missing piece is post-training, meaning taking a general-purpose model and teaching it one specific job in one specific environment until a company can depend on it.
MicroAGI says it is building both halves. With joinshift it is assembling what it calls one of the largest egocentric datasets in the world, egocentric meaning first-person video shot from the wearer's point of view. With a system called Atlas it post-trains and harnesses models directly on enterprise deployments. The stated goal is to deploy 1 million robots.
That number is an ambition, not a deployment. No customer count, no timeline and no results have been shared. What the post does offer is a clear read on where one well-staffed team thinks the gap sits, and it is not in the base model.
In LLMs, post-training is what turned impressive demos into products. In robotics, that layer barely exists yet.