Serve Robotics Debuts Physical AI Services for Robot Makers

Serve Robotics announced a new business line called Physical AI Services on Thursday, alongside its third-quarter earnings, selling the connectivity and data tools it built for its own delivery robots to other companies building robots, according to TechCrunch.

What is Physical AI Services?

It's a new unit inside Serve Robotics that packages the company's internal robotics infrastructure into products other robot makers can buy, TechCrunch reported. Serve Robotics spun out of Uber's acquisition of Postmates in 2021 and went public in 2024. It currently builds sidewalk delivery robots and, through its acquisition of Diligent Robotics, hospital robots. Those businesses are staying put, CEO and co-founder Ali Kashani told TechCrunch, but the company sees an opening to license the underlying technology to robot builders in farming, warehousing, and other "uncontrolled environments."

Why does Serve Robotics call this an "AWS moment"?

"I kind of think of it as the AWS moment where we are taking the technology we've built for us, and we're opening it up for more people to build on," Kashani said, per TechCrunch. The comparison echoes Amazon's own path: the retailer built internal computing infrastructure for its e-commerce business, then sold that infrastructure to outside companies as Amazon Web Services. Kashani said he has been thinking about that playbook for years and that it would not have been possible without first running its own robot fleet long enough to work out the hard problems.

What products make up Physical AI Services?

TechCrunch reported three initial offerings. Serve Link connects field robots to AI systems running in data centers, so a robot does not need all of its compute on board. "Any robotic application generally needs to connect the robots to some offline intelligence, whether it's to collect data or do exception handling or whatever," Kashani told TechCrunch. "You need to be able to manage really unreliable networks and still reliably connect, and that's what we solve." A second tool, a data curator, helps companies surface edge cases buried in the real-world data their robots collect, feeding that data back into training models. The third product is a foundation-model data catalog built from Serve's own sidewalk robots, intended to teach AI models how to navigate complex human environments.

Who is already using it?

Bedrock Robotics, a startup building self-driving systems for construction equipment, is a customer, Serve told TechCrunch. The company is also working with T-Mobile to bring AI computing capacity physically closer to robots operating in the field, according to the announcement.

Does this replace Serve's delivery robot business?

No. Kashani emphasized to TechCrunch that Serve's sidewalk delivery and hospital robot operations are continuing. The sidewalk delivery business itself is in transition, moving away from reliance on Uber and toward other partners including DoorDash and Wonder, per TechCrunch's reporting. Physical AI Services is described as an additional revenue line layered on top of, not a substitute for, that existing robot fleet.

Why frame it around "robots in human spaces"?

"We are solving the problem of robots in human spaces," Kashani told TechCrunch, framing the new business as an extension of a problem Serve already had to solve for its own sidewalk robots — operating safely and reliably around pedestrians, traffic, and unpredictable terrain — now packaged for customers facing similar challenges in other industries.