The Robot Business Is Now an AI Business

Boston Dynamics named Rohit Prasad its new CEO on October 7. Prasad spent 12 years at Amazon, where he built Alexa from its early days into a product used by hundreds of millions of people, then led Amazon Nova — the company's family of AI models built for enterprise use. He has never run a robotics hardware company. That is not an oversight. It is a signal about where robots are heading.

Boston Dynamics makes Spot, the dog-like inspection robot, Atlas, a humanoid, and Stretch, a warehouse machine. Its owner, Hyundai, plans to put 25,000 Atlas units to work inside Kia car factories by 2028. Getting from polished demos to thousands of robots handling real, unpredictable factory work is not mainly an engineering challenge. The hard part is making robots smart enough to handle the mess of the real world.

That is exactly the kind of problem Prasad spent his career solving. Building Alexa meant teaching a system to understand millions of different phrases, accents, and contexts without breaking. A factory robot faces the same kind of endless variation every day — new parts, unfamiliar tasks, unexpected obstacles. Those problems require similar tools: training on huge amounts of data, building models that generalize, and shipping systems that hold up in messy conditions. Both fields need the same underlying AI skills — more than most people realize.

This is part of a wider shift. The companies that win in physical robotics over the next few years will probably be the ones that crack the intelligence problem first, not just the mechanical one. Hiring an AI model builder to lead a robot company is a bet on exactly that. If you build automation for clients or plan to deploy robots in any workflow, it is the AI layer — not hardware specs — that you should evaluate. Who trained the model inside the machine now matters as much as how well the machine moves.