Technology

Serve Robotics Unveils Physical AI Services to Become the AWS of Robotics

Serve Robotics launches Physical AI Services, a cloud‑like platform offering connectivity, data curation, and a foundation‑model catalog to let other firms build robots faster.

S
SciTech Editorial Team
Newsroom
··2 min read
Serve Robotics Unveils Physical AI Services to Become the AWS of Robotics
Serve Robotics, the maker of self‑driving sidewalk delivery robots, announced on Thursday the launch of a new business line called Physical AI Services. The venture is positioned as the "AWS moment" for robotics, offering a suite of connectivity and data tools that enable third‑party developers to build and scale robot applications without constructing their own infrastructure. CEO and co‑founder Ali Kashani told TechCrunch that the company is "solving the problem of robots in human spaces" by providing high‑performance, reliable links between field robots and powerful AI models hosted in data centers. Physical AI Services will initially roll out three products. Serve Link is a low‑latency networking layer that keeps robots tethered to off‑site intelligence, allowing them to tap into large language models and computer‑vision algorithms without on‑board compute overload. A data curator tool automatically extracts edge‑case scenarios from the massive streams of sensor data robots collect, feeding those examples back into training pipelines. Finally, the company will publish a foundation‑model data catalog built from its own sidewalk fleet, offering a real‑world reference for AI systems learning to navigate complex human environments. Early adopters include Bedrock Robotics, a startup developing autonomous construction equipment, and a partnership with T‑Mobile to bring edge AI compute closer to robots. The move builds on Serve Robotics' rapid evolution since its spin‑out from Uber’s acquisition of Postmates in 2021 and its 2024 IPO. The firm already operates two distinct robot families: the iconic googly‑eyed sidewalk delivery unit and a hospital service robot acquired through Diligent Robotics. While the delivery arm expands its partnerships beyond Uber to include DoorDash and Wonder, Kashani emphasized that the underlying technology platform—refined through years of fleet operation—can now be licensed to firms targeting farms, warehouses, and other uncontrolled environments. Industry analysts see the announcement as a potential catalyst for a broader robotics ecosystem. By abstracting the heavy lifting of connectivity, data management, and model training, Serve Robotics lowers the barrier to entry for niche players and could accelerate the deployment of autonomous solutions across logistics, construction, and healthcare. However, challenges remain: maintaining reliable links over unreliable networks, ensuring data privacy, and scaling the catalog to cover the full diversity of real‑world settings. Looking ahead, Serve Robotics plans to iterate on its Physical AI Services platform throughout 2027, adding more edge‑compute partners and expanding the catalog with multimodal datasets. If successful, the company could redefine how robotic startups source AI capabilities, echoing the transformative impact Amazon Web Services had on cloud computing a decade ago.

Share

More from this section: Technology