Serve Robotics Opens Its Stack: Sidewalk Bot Maker Launches Physical AI Services

Serve Robotics is turning the technology it built for its own sidewalk delivery robots into a product for everyone else. On Thursday, alongside third-quarter earnings, the company announced a new business line called Physical AI Services, aimed at selling connectivity and data tools to other firms building robots.

The company knows this playbook well. Serve spun out of Uber's acquisition of Postmates in 2021, went public in 2024, and now wants to be the plumbing underneath other people's robots — not just an operator of its own fleet.

Its first announced customer is real. Bedrock Robotics, a startup developing self-driving systems for construction equipment, has signed on. Serve is also working with T-Mobile to bring AI computing power closer to robots in the field, the company said.

"We are solving the problem of robots in human spaces," Serve Robotics CEO and co-founder Ali Kashani told TechCrunch.

What does the new unit actually sell?

Physical AI Services launches with three products, and each one targets a pain point that any robot developer already knows:

• Serve Link connects robots in the field to AI systems running in data centers. That lets companies tap more powerful AI models without stuffing all that compute onboard the machine.
• A data curator tool helps companies find edge cases buried in the reams of real-world data a robot collects. Those edge cases can then be fed back into a training model to improve performance.
• A real-world foundation-model data catalog built from Serve's own sidewalk robots, designed to teach AI models how to understand complex human environments.

The connectivity piece is harder than it sounds, Kashani argued. "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," he said. "It's the connectivity piece that has to be really high performance. You need to be able to manage really unreliable networks and still reliably connect, and that's what we solve."

Why an operator is becoming a supplier

Serve already makes two kinds of robots: the googly-eyed sidewalk delivery bot that made its name, and a hospital robot it gained through its acquisition of Diligent Robotics. Those businesses aren't going away, Kashani emphasized in a recent interview.

The opportunity, he said, is selling the technology Serve developed for itself to other companies building robots for farms, warehouses, and other uncontrolled environments.

"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. He added that this wouldn't have been possible without first developing his company's own robot fleet.

The strategy borrows directly from Amazon's history. "When the technology is so nascent, you have got to build the end-to-end application because that's the only way you know what to build, but then over time, as the technology gets more advanced, a lot more people are going to be wanting to build similar things," Kashani said. "And what you want to do is to open your technology so others build with you and on your platform."

A shift already underway in the delivery business

The launch comes as Serve's core sidewalk delivery operation keeps expanding — but its partner mix is changing. The company is shifting away from Uber, its original parent's ecosystem, toward other partners including DoorDash and Wonder.

For robotics practitioners, the practical question is straightforward: does this lower the barrier to deploying autonomous machines in messy, real-world settings? If Serve Link and the data catalog deliver, smaller robot makers could get data-center-grade AI and hard-won real-world training data without building a fleet first. Much now depends on whether Bedwood-type customers beyond construction follow, and whether the T-Mobile edge-computing tie-up can scale.