Robotics is ready to cross the gap from simulation to reality. Once confined to controlled pilots and tightly scripted tasks, robots are now operating in dynamic, unpredictable settings where adaptability and autonomy are essential. McKinsey Partner Christian Jansen recently spoke with David Reger, founder and CEO of Neura Robotics, about what it will take to move robots from promise to widespread deployment. Drawing on his experience in industrial and robotics companies, as well as his public service as a member of Germany’s Senate of Economy, Reger explains how intelligent machines are evolving from stand-alone devices into learning systems, enabled by platforms that make robotic capabilities accessible to all. The following is an edited transcript of their conversation.
The world’s next productivity engine
Christian Jansen: Let’s start by envisioning the future. How do you see the next phase of robotics developing? What will be the most important achievements by 2040, both for Neura as a company and the industry as a whole?
David Reger: At Neura, we’re focusing on the industrial sector first, because we see it as important for humanity. If you look at Europe, China, and Japan combined, there will be 100 million fewer human workers in just the next five years. This is where I see the biggest need and the biggest pressure. Robotics players only have a short time to solve that problem.
In the industrial space, you don’t need a humanoid robot that can do everything. You can just solve the little things, like pressing a button, turning on a switch, or picking up a part and putting it into a machine. Humanoids today can already do these things, which means they could handle about 50 to 60 percent of workforce tasks worldwide while operating within the same physical footprint as humans.
In 2040, I would be proud if by then we are using robots as daily tools, as our helpers, to make our lives easier, safer, and happier. I believe that by 2040, robots will be a new standard, where it will be hard to think back on a time when we had to clean our houses ourselves or do dangerous or monotonous tasks, including in factories, just to survive.
Christian Jansen: That’s a good point. Some videos show humanoids doing ballet or kicking melons or whatever; but there are already robots, including Neura’s, that can do real tasks. Neura has a full portfolio of form factors—from mobile actuators to full-scale industrial robots to humanoids. How do you think robotic form factors will evolve over time, and does that shape your vision of the future?
David Reger: You might think, “We have one form factor that looks like a human. That’s going to be enough, because it will be able to do all these tasks.” But it takes time to build a fully functional humanoid, and during that transition time, we need to develop other useful devices. I think it’s a German mindset. We can’t just have a big vision and work toward it; we have to prove that we’re on the right track and show that we can generate revenue along the way.
So the first goal of Neura is not about the form factor; it’s to enable any form factor to make smart decisions. If you take a robot arm and implement the senses of hearing, seeing, and feeling—enabled, for instance, by cameras and microphones connected to a strong GPU [graphics processing unit] with a neural network—that’s the first step in gaining the right data and giving a device the ability to have a fully autonomous reaction in different situations. That could give you an advantage in developing humanoids. You already know it works; you’re just putting it into a different form.
From simulation to scale
Christian Jansen: There’s a phrase that’s often used by tech people in robotics: the sim-to-real gap. Can you describe what that gap is and how Neura intends to cross it?
David Reger: Let’s say there’s a digital twin in a virtual environment where you’re training a robot to do a task. Then you try to deploy a real robot in the real world, in a situation where everything is moving and isn’t as precise as it was with the digital twin. To close this gap, you need robotic intelligence.
We’ve started building Neura Gyms, physical environments where you can find a wide range of robot form factors, including arms, mobile robots, and humanoids. You can remotely control these robots by wearing a sensor suit—you’re essentially “going into the body” of the robot to solve a task. This is the only way to close the sim-to-real gap.
Neura Gyms also address what I refer to as the human–robot disability gap. A human has more sensors and more degrees of freedom than a robot. So we have more than 1,000 workers worldwide—not Neura employees but workers in other companies—who wear data suits every day while doing their tasks. Their data suits have a camera, tactile sensors on the gloves, and acoustic sensors. The results of this experience, mixed with other data, allow us to create the best model.
Everyone asks, “What is this robot capable of doing?” Dentists are asking if robots can be their assistants; hairdressers want to know if robots can cut hair. Manufacturing companies are asking if robots can perform tasks that humans cannot be trained to do. In Neura Gyms, you can try the robot out. There aren’t a lot of physical tasks that can’t eventually be taught to a humanoid robot given the right training environment, data, and safety conditions. You just need the space and a place where everyone can go to train it.
An app store for physical AI
Christian Jansen: You mentioned the use of tactile sensors, which sets Neura apart from most other robotics companies. Say more about why you’re taking that approach.
David Reger: Right now, everyone is hyping vision-language-action [VLA] models simply because they are trained on video data. I think it’s completely the wrong way to go because most tasks are not solvable by just having vision. Take swimming as an example: You can’t learn how to swim by watching a video.
It’s also not possible to use only simulation. With physical AI, the only way to train the model and solve tasks like a human would is to involve hearing, seeing, and feeling. You need tactile information, and sometimes you even need to hear a click to solve the task correctly. This is what we’re doing in Neura Gyms.
Christian Jansen: So the gyms are largely about sensory inputs—the “spine,” so to speak. What is Neura doing on the “brain” side?
David Reger: Neuraverse is our platform that allows anyone to build an end-to-end model for a task—meaning a robotic skill that goes from perception to action and is ready to do the job where it is needed. Instead of being a company that tries to solve all the world’s issues, we give everyone the tools and platform to solve them faster and more easily themselves.
One part of the platform enables simulation, so it’s like the imagination part of the brain. Another part is a developer environment where everyone can bring their own AI models and get add-on devices. The third part of the platform is the marketplace, which is like an app store for AI tools, solutions, software, and complete apps that you can simply click on and deploy directly.
Why is this so important? Because we believe that the future won’t be just a very smart robot that solves all the world’s problems; it will be a layer above that. A smart robot shouldn’t have to open a fridge or oven to see what’s inside. It should be able to connect its “eyes” to the fridge or oven and use that information to get work done most efficiently. With the Neuraverse platform, you can combine, say, a robot with sensors or sensors with AI models in a very simple way. What we built is basically an interface that lowers the barrier to robotics development, so manufacturers and developers can combine models, sensors, and robotics hardware without having to start from scratch. Everyone can just go there, click on one of the models, and combine it with the sensors so that, for example, a microwave can have a small brain to “think” about the state of the food it’s cooking before turning itself off.
We can’t solve every problem in the industrial sector either, but we want to give every company a platform and the tools to solve its own problems. On Neuraverse, companies can own their own know-how and IP [intellectual property]. They can decide which capabilities they make publicly available and which ones they’d like to keep for themselves to gain a market advantage.
Apple could have just built a better Snake game and tried to compete with Nokia on that. They didn’t. And later, they opened the gate for the gaming and entertainment industries to get access to the iPhone and turn it into a multipurpose device. That’s the analogy we’re using for the platform we’re building. We want to enable the world to contribute its knowledge, and that’s why I don’t think we need to wait until 2040 for robotics to scale. It will go much faster if the tools are provided that allow the whole world to help make it happen.
Europe’s opportunity to become a robotics powerhouse
Christian Jansen: Cognitive robotics is still in its early days. Companies are inventing and reinventing many new technologies, and often they’re partnering with other companies to do so. What are your thoughts on the robotics ecosystem? How do you see it evolving?
David Reger: Germany—and Europe in general—is the right place to build a robotics company, and also for the whole ecosystem for robotics. We are in an automotive country. Everything here is about very large quantities of cars going out into the world. So the ecosystem for that is already built. For sensors, including the process of developing and scaling them, there’s almost no country in the world with more knowledge than Germany.
To win a global race, a lot is required, including motivation. Our country and our continent should have the greatest motivation to win in robotics because we need a new economic pillar alongside automotive. There is so much more competition worldwide in the automotive industry. The only way we can win is by getting more efficient. And how do you get more efficient? With robots, of course. In my opinion, robotics could be the largest future market, and we could take part in it and maybe even win in the long term.
Why waiting is losing
Christian Jansen: What advice would you give to CEOs and other decision makers on winning in robotics, both in the short term and through 2040?
David Reger: We are living in a very special time. If you’re thinking about using robots to get more efficient or to be part of the future, my biggest recommendation is not to wait until those products are ready.
When I’m having discussions with potential partners, the first thing they ask is, “Can you show me a robot?” Then they’ll say, “Oh, it’s awesome. I’ll wait until it’s working 24/7 for a year, and that is when I’ll apply it.” I have a simple response to that: If you wait until it’s done, others will have it before you, and your company may no longer exist.
That is how the automotive sector was defined. There was a time when cars were built without robots. Then the first robots came to market. There were a lot of players, and everyone did it their own way, but automotive took ownership of robotics. The proof is at companies like Volkswagen; they developed the whole control systems behind robotics. General Motors helped Fanuc; they had joint ventures for some time. Today, Fanuc is the largest supplier of industrial robots.
Now is the time to take ownership, rather than waiting until things are ready. Form a real partnership and solve problems together before others do. That’s the only way to win.
