Serve Robotics Launches New Initiative Targeting the AWS of Robotics
Serve Robotics, recognized for its autonomous sidewalk delivery robots, is expanding its horizons to serve as a technology supplier for other robotics firms.
The company, which originated from Uber’s acquisition of Postmates in 2021 and went public in 2024, unveiled a new division called Physical AI Services on Thursday.
Serve Robotics will focus on offering its connectivity and data solutions to various clients. Bedrock Robotics, a startup specializing in self-driving technology for construction equipment, is one of its customers. Furthermore, Serve is working with T-Mobile to bring AI computing capabilities closer to robotic systems, a partnership revealed alongside its third-quarter earnings on Thursday.
“We are tackling the issues of deploying robots in human-populated areas,” said Serve Robotics CEO and co-founder Ali Kashani in an interview with TechCrunch.
The company has already rolled out two varieties of robots: the endearing googly-eyed sidewalk delivery robot and a hospital robot obtained through its acquisition of Diligent Robotics.
Kashani clarified in a recent conversation that these operations will persist. However, he envisions opportunities to market the technologies developed for their robots to other companies working on robotic solutions in diverse environments such as farms and warehouses.
“I see this as our AWS moment, where we share the technology we’ve built for ourselves to empower others to innovate,” he noted, emphasizing that this progress was only possible after establishing a fleet of its own robots.
This initiative aligns with the growth of Serve’s sidewalk delivery service, which is evolving from a dependency on Uber to collaborations with partners like DoorDash and Wonder.
The new Physical AI Services division will feature three primary products. Serve Link will enable field robots to connect to AI systems located in data centers, allowing companies to leverage more advanced AI models without the need for extensive computing capabilities installed directly on the robots.
“Every robotic application generally needs a link to some external intelligence, either for data gathering or addressing exceptions,” Kashani elaborated. “The connectivity must be exceptionally resilient, ensuring reliable connections even in suboptimal network conditions, and that is exactly what we focus on.”
In addition, the company is introducing a data curator tool aimed at helping businesses pinpoint edge cases within the massive datasets collected by robots, which can enhance training models for better outcomes. The third offering is a comprehensive catalog of real-world foundational model data gathered from Serve’s sidewalk robots, which can inform AI models on navigating complex human environments.
Kashani shared that he has been mulling over this idea—a strategy inspired by Amazon—for several years.
“In the early phases of technology development, you must create a complete application, as that is the only way to identify what is needed. However, as technology matures, more companies will want to build comparable solutions,” he explained. “Therefore, it’s crucial to make your technology available for others to develop alongside you and on your platform.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.


