Predictive maintenance, logistics, automobile factories: Boston Dynamics robots are leaving labs for industrial sites. Interview with Marc Theermann, chief strategy officer, on the future of humanoids.
JDN. Founded in 1992, Boston Dynamics caused a sensation for years with its acrobatic robot videos. How did the shift towards industrial marketing come about? ?
Marc Theermann. Boston Dynamics is what I like to call a 30-something-year-old startup. We have existed for three decades and have been a bit like patient zero of legged robotics. Today, we are 80% owned by Hyundai and 20% by Softbank. Under their ownership, we began to commercialize our robots and turn them into products. We started with Spot about five years ago, looking for our product-market fit for this quadruped robot. Boston Dynamics employs more than 1,000 people, having quadrupled its workforce over the past five years. Among our main clients are Chevron, Nestlé Purina, Lidl, and Gap.
Spot is your four-legged robot and also your flagship product. What are its main use cases today ?
After about a year of sales to companies, mainly to innovation teams, two use cases emerged. The first consists of teleoperating the robot in very dangerous situations or extreme environments, whether they are very cold or very hot, or whether they involve nuclear disasters like Fukushima, or hostage situations. The robot replaces humans where it is too risky. The second use case is autonomous inspection. Spot can operate alone in a factory to collect a lot of data and thus becomes a mobile connected sensor thanks to its infrared cameras and acoustic vibration sensors.
Concretely, what is the benefit of this data collection for manufacturers?
The robot moves around and collects a considerable amount of data which is then analyzed via our Orbit cloud solution. This detects anomalies and tells factory owners what is different from one day to the next. For example, if a machine is 10 degrees hotter than yesterday, we can predict that it will break down within two weeks. It can also identify possible leaks in the ceiling, such as in a CO2 pipe. In summary, Spot allows you to do predictive and preventive maintenance.
“AI has given our robots a semantic understanding of the world”
What led you to develop the Stretch robot for logistics rather than using your Atlas humanoid robot ?
The idea for Stretch was born about seven years ago, following the publication of a video showing our humanoid robot carrying a package. Many companies have contacted us to purchase this type of robot. We told them that the humanoid was not yet suitable for this, but that we could develop a tailor-made robot for logistics. We worked with several clients, including DHL, to design Stretch, unveiled in 2021 and marketed in 2023. It is a simple but effective robot. Just press a button and it scans its surroundings, enters a container and unloads it autonomously. With software updates, we hope it will eventually be able to perform other tasks like palletizing or loading trucks.
Atlas, created in 2013, went viral thanks to its acrobatics videos. Why did you decide to switch from a hydraulic to electric version, and when will it be marketed? ?
Atlas, our humanoid robot, aims to help companies assemble and manufacture their products, including cars. With Boston Dynamics now owned by Hyundai, an automobile manufacturer, we will start by adapting Atlas for this industry, but we believe it could be applied to any manufacturing sector in the future. The initial Atlas was entirely hydraulic, the only way to achieve the power density needed at the time. Today, advances in electric actuators and motors make it possible to replicate and exceed this density. We have been developing an electric Atlas for two years and we will launch a commercial version in early 2026.
How recent advances in AI have transformed the capabilities of your robots like Spot or Atlas ?
The advancements in AI over the past 24 months have been dramatic for robotics. They give our robots a semantic understanding of the world. Two years ago, Spot treated a trash can, a forklift or a human in the same way, as simple obstacles. Today, thanks to AI, the robot recognizes a human and behaves differently. He can recognize a ladder and go around it instead of going under it. If there are cables on the ground, the robot understands that its foot could get tangled. If you tell the robot “go to the kitchen”, it can now understand where it should go. Historically, robotics companies used control software to operate their robots. But with reinforcement learning, the robot learns on its own how to behave based on the policies we create.
What is your business model? Do you only sell the hardware or do you also offer associated services? ?
We are quite flexible from a business point of view. Today, the majority of our customers purchase the hardware and then take out a subscription for our software. Many also use our deployment services. These three elements constitute our main sources of income. We have a dedicated team that supports our customers in deploying their robots all over the world. Last year, we launched two certification programs to allow partners to deploy and repair our robots. For example, the Korean company Clobot is both our certified reseller and integrator in Korea. Intel, also a major customer, became our first certified repair shop.
With Hyundai as the majority shareholder, how are you adapting your roadmap? Do you plan to develop robots just for this business ?
No, we do not plan to develop robots exclusively for Hyundai. However, the advantages are twofold. On the one hand, we have a client ready to deploy thousands of humanoid robots in its factories, on the other we benefit from their valuable help to develop these robots and ensure that they are as useful as possible once deployed to our clients. Hyundai is a company known for its ability to produce high-quality products at low cost. So it’s a great partnership.
The robotics market is intensifying with the arrival of Tesla, Figure AI, but also many American and Chinese start-ups. How Boston Dynamics plans to maintain its technological edge ?
We are going through a period of considerable hype around robotics. Our advantage is that we have been in this industry for many years and there are four pillars necessary to succeed. First, hardware reliability, which takes time and cannot be rushed, despite what some competitors think. We have thousands of Spot robots deployed, which gives us significant feedback. Then, the value for customers, who expect a real return on investment. A nice demo video posted on YouTube is not enough and we have been practicing this with Spot and Stretch for five years.
Third, security. Some competitors, particularly in China, seem to me to be rather lax in this area, which is dangerous. Finally, you need expertise, talent and a solid go-to-market capability. In conclusion, even if competition is intensifying, our experience in marketing and deployment is a real asset, and we have the ambition to remain the leader in this market.
Do you think there is an opportunity for these autonomous mobile robots to contribute to the reindustrialization of Europe, which has been largely outsourced to China for several decades? ?
I do think that there is a real opportunity to relocate and bring manufacturing back to Europe thanks to autonomous mobile robots. These robots could help revitalize old factories. Starting at a price of less than $300,000 per humanoid robot, an industrialist could buy several of them, place them in an old factory and begin to automate part of the production. This is one of our fundamental missions. We believe that our humanoid Atlas could provide a positive return on investment in less than three years, taking into account an environment where the robot would work the equivalent of approximately two and a half shifts.
Purchase prices for these humanoid robots remain high. Do you expect a significant cost reduction in the medium term ?
Manufacturing costs are indeed falling dramatically now. The cost of a humanoid roughly consists of one third for the motors, one third for the computing power and one third for the rest of the components, such as batteries, sensors and other systems. The good news is that the costs of all three are coming down significantly.
Robert Playter, CEO of Boston Dynamics, said we could see domestic robots in our homes within 10 years. What do you think ?
I am very excited about these pet robots. This is also the reason why I joined Boston Dynamics. I think it will be even faster than Robert predicted. However, those who claim that we will have a life-size humanoid in our homes next year are talking nonsense. This would be far too dangerous; technological capabilities do not yet allow it. Today we have an epidemic of loneliness in the world, especially in megacities like Paris, Seoul or Shanghai.
Many people would like to have a pet but cannot. This is where a pet robot could provide real companionship through an emotional connection, no matter what it looks like. Research shows that embodiing AI in a body allows lonely or sick people to heal faster and respond better to therapy. Regarding the vision of the humanoid robot that cleans your house and cooks your meals, we will probably have to wait another decade, or even more.
Can we imagine Boston Dynamics marketing its robots in B2C in the years to come? ?
Today, our focus is 100% on industry and B2B. We are in the midst of the industrial era of robotics, where robots are designed for specific needs. I think that we will then enter an era of services, with robots which will undoubtedly be remotely operated by local operators who will be responsible for an entire fleet. Once we have a good handle on this, and understand how to get robots to interact with humans, and how to provide value to them, whether in restaurants, hospitals or theme parks, then we can think about how to bring these robots into our homes. If I had to make a prediction about Boston Dynamics, I think we’ll probably go through this era of service before we consider any B2C robot marketing.
Marc Theermann is chief strategy officer at Boston Dynamics, where he oversees the development of the company’s go-to-market strategy and is responsible for partnerships, sales, marketing and customer success. Previously, he worked at Google, where he was responsible for cross-product strategy and key account management for the gaming and startup industries. he holds an MBA from Northeastern University and a master’s degree in International Management from Thunderbird.
