Ultra raises $62M and will put Physical Intelligence's robot models on its leased industrial fleet
The AMW Read
A mid-size raise in a known space (incremental), but putting a robot foundation model on a commercial RaaS fleet is a segment-level test of whether embodied foundation models work in deployment.
Named counterparties: Physical Intelligence
Ultra raises $62M and will put Physical Intelligence's robot models on its leased industrial fleet
Ultra has raised $62 million, which brings its total funding to more than $100 million. It also announced a partnership with Physical Intelligence, which builds foundation models for robot control. Ultra installs industrial robots for manufacturers and logistics operators and charges a monthly fee instead of selling the hardware. Those robots handle palletizing, machine tending and material transport. Ultra says its fleet grew 10x over the past year and now operates in multiple U.S. states. It did not disclose a valuation. The money will go toward expanding the fleet, hiring engineers and integrating Physical Intelligence's models so robots can adapt to new objects and layouts without manual reprogramming.
The deal links a general-purpose robot model with a commercial fleet that is already deployed. Physical Intelligence has raised over $400 million from backers including Jeff Bezos and OpenAI. It has shown its models folding clothes, packing boxes and assembling components, but most of that work has happened in labs. Ultra's sites are dusty factories with uneven floors, changing light and production schedules that don't pause for software updates. That makes them a live test of whether models that work across many robots and tasks hold up in deployment. The article also notes the known gaps: long-horizon tasks and rare edge cases, such as a damaged box or an unevenly stacked pallet.
Investors should watch Ultra's operating numbers more than the round size. A subscription model only works if robots stay productive, because downtime cuts directly into Ultra's revenue. Builders of embodied AI should track whether Ultra can bring down manual intervention and reconfiguration costs at scale. If it can, fleet operators that own deployment and customer relationships could become the main route by which robot foundation models reach the factory floor, and those operators would capture a meaningful share of the value.
#Robotics #PhysicalAI #RobotsAsAService #PhysicalIntelligence #IndustrialAutomation #Funding
