Rohit Prasad, Ex-Amazon AI Chief, Takes Boston Dynamics CEO Role to Sell Robots at Scale

Rohit Prasad will become Boston Dynamics CEO on October 7, the company said. He joins from Amazon, where he ran Alexa and the company's artificial general intelligence research. Boston Dynamics is owned by Hyundai Motor Group. The company said the appointment is meant to speed its Physical AI work toward commercial deployment and narrow the distance between robotics research and products industrial customers buy.

Prasad spent 12 years at Amazon and 14 years before that at Raytheon BBN Technologies, working on machine learning. At Amazon he was a senior vice president and the head scientist for Alexa and artificial general intelligence, and he led work on the Nova family of foundation models. Boston Dynamics expects him to take a board seat once standard approval steps finish, according to the company.

The mandate is unusually specific for a leadership change. Boston Dynamics wants to stop being a research lab that produces admired demonstrations and become a company that ships robots to industrial buyers at scale. That transition defines its current phase under Hyundai Motor Group, and the board handed the problem to an executive whose record is in AI products rather than mechanical engineering.

What the Boston Dynamics CEO Job Actually Involves

Three programs sit under the new chief executive, and they are at very different points of commercial readiness.

Spot and Stretch address industrial categories that already exist: repeated inspection rounds and trailer unloading, which is punishing physical work. Atlas runs on a different schedule. Boston Dynamics presents it as the platform for general-purpose work, and it is still in development. The first units go to Hyundai Motor Group and Google DeepMind.

Neither recipient is a typical industrial buyer. Hyundai Motor Group owns Boston Dynamics, which makes it a captive testbed for factory tasks. Google DeepMind is a frontier AI laboratory, which makes it a research collaborator rather than a customer ordering in volume. Read together, the early Atlas placements look like a validation program. They generate operating data from real sites and credibility from outside the company, but they are not the industrial order book Boston Dynamics says it is building.

Near-term revenue comes from inspection and logistics, not humanoids. Spot and Stretch fit budgets that already exist for inspection, uptime and maintenance, and a new chief executive can grow that line with the products on hand. Atlas sells a claim about future labor capacity. Claims about the future do not pay for a manufacturing ramp.

The Bet Behind the Appointment

For most of the past decade the binding constraint on commercial robotics was physical. Actuators, battery density, unit cost and reliability in unstructured settings decided which machines left the laboratory. Putting a foundation-model executive in the Boston Dynamics CEO seat implies the company now judges cognition, integration and go-to-market to be the harder problem.

Boston Dynamics says its aim is to connect advanced robotics with large-scale AI productization, which is a software and services problem as much as a hardware one. A humanoid fleet in a plant or a warehouse needs perception models that hold up in dust and glare, autonomy stacks that fail safely, remote supervision tools, and update pipelines that push improved policies to machines already in the field. Prasad worked on those disciplines at Amazon, applied to Nova models and to consumer devices shipped worldwide.

The technical logic is similar. Nova models are multimodal, handling images, video and text together, which is close to the perception problem a mobile robot faces on a factory floor. Spotting a damaged pallet or reading an analog gauge draws on the same class of model that describes a photograph. Boston Dynamics has strong locomotion and control stacks. What it has had less of than software-first rivals is a production AI platform underneath them, and Prasad spent his Amazon years building that layer.

Timing adds a second signal. The appointment lands as the humanoid category fills with entrants chasing the same industrial customers, which raises the cost of a research-led identity. Boston Dynamics has been famous for years. Fame does not book revenue, and it does not shorten a deployment cycle.

Where the Appointment Could Fall Short

The strongest objection is that AI leadership is the wrong résumé for a hardware business. Industrial robotics is won on unit economics, service networks, spare parts availability, safety certification and manufacturing yield. None of those appear on a foundation-model CV. Amazon's consumer devices shipped at price points and durability tolerances far below what a factory floor imposes, and running an AI at consumer scale differs from supporting machinery a customer expects to operate for years.

That objection has force, but it assumes Boston Dynamics still needs someone to solve the robot. It does not. Spot and Stretch are established products, and Atlas comes from a company that has spent decades on locomotion and control. The remaining gap is deployment: getting machines onto customer sites, wiring them into existing systems, keeping them running, and proving that a robot working near people is safe and worth its capital cost. That work is software-heavy and service-heavy, and an executive who has run large-scale AI products can plausibly change outcomes there.

The other uncertainty is cultural. Boston Dynamics built its reputation on mechanical ingenuity and long research horizons, and its engineers are among the most sought after in the field. A chief executive whose instinct is to ship, iterate and measure adoption can speed commercialization and still lose the researchers who make the next generation of Atlas possible. Holding both halves of the company together may matter more than any single product decision.

The larger risk is structural. Prasad inherits three product lines at three maturity levels, a parent company that is also a customer, and a research partner in Google DeepMind that competes with nearly everyone in AI. Managing those relationships is a political job as much as an operating one. The Boston Dynamics CEO role will be judged on bookings rather than on model benchmarks.

What to Watch

Three signals will show whether the strategy is working. The first is whether Boston Dynamics names paying industrial customers for Atlas beyond its own parent. The second is whether Spot and Stretch grow as a share of the business while humanoid spending continues. The third is whether Prasad's board seat is confirmed on the expected timeline, which would give him direct influence over the Hyundai relationship he already manages as a subsidiary head.

Why this matters

The hire says where Boston Dynamics thinks its bottleneck now sits, and the bottleneck is not the hardware. For companies evaluating industrial robots, the useful questions shift toward software support, integration effort and service life instead of raw capability demonstrations. For anyone tracking the humanoid category, the Atlas customer list is the telling detail: two early recipients, one a parent manufacturer and one a research laboratory, with no volume industrial order book yet. That is the gap Prasad was hired to close, and it will show up in bookings long before it shows up in any robot video.

Related Articles

• How the Hyundai Boston Dynamics Acquisition Reshapes Physical AI Strategy
• Boston Dynamics IPO Pushed Past 2027 as Hyundai Bets on Factory Scale
• Bezos Secures $10B for Project Prometheus Physical AI

✔Human Verified

Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.