AI is moving beyond the data centers where models are trained and onto factory floors where machines must handle constant change.

But traditional industrial robots work behind a cage, repeat a single task and were never designed for the complexities of today’s market. Manufacturers now want general-purpose machines that can switch tasks and improve as processes evolve. The hard part is deploying it, according to Adrien Gaidon (pictured, right), co-founder and chief strategy officer of Walden Robotics Inc.

“If you’re changing things and you have machines that cannot change and continuously learn in deployment, then you become a burden,” Gaidon said. “What we do is we do machines that continuously learn from deployment and adapt so customers can change their mind and improve the process and give them the superpower to change even more and faster.”

Gaidon and John Mancuso (left), vice president of field engineering at CoreWeave Inc., spoke with theCUBE’s John Furrier and Dave Vellante at Fully Connected, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how cloud infrastructure, simulation and human assistance help robots learn and adapt on factory floors. (* Disclosure below.)

Physical AI moves from data centers to the factory floor

Walden spun out of Toyota Research Institute in January 2026 and launched with $300 million in July to put general-purpose robots to work in manufacturing. Its robots have worked full production shifts alongside people since May. They handle machine tending, parts kitting, and subassembly, tasks manufacturers had either failed to automate or never tried, Gaidon explained.

“There’s half a million open positions for machinists in the U.S. because it’s a hard job and it requires skills,” Gaidon said. “But part of that job does not require the craft or the human creativity.”

Getting a robot to that point takes serious compute long before it reaches a production line. CoreWeave, which has been betting on physical AI, also brings mechanical engineers and robotics experts into customer engagements alongside its infrastructure, Mancuso explained.

“In the end, before a model is ever going to do something in the physical world, it’s going to have a tremendous amount of simulation that happens in the virtual world,” Mancuso said. “We’re providing the infrastructure.”

But while simulation in data centers can make a robot useful, but it can’t make it perfect. Walden pairs autonomous robots with remote human assistants. The setup is comparable to automated looms of 120 years ago, when one person oversaw 40 machines, Gaidon noted.

“Our robots are autonomous, but they know how to call for help,” Gaidon said. “They know when they need some assistance – and that assistance feeds the most valuable data to improve over time.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Fully Connected:

(* Disclosure: TheCUBE is a paid media partner for the Fully Connected 2026 event. Neither CoreWeave, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE