Where are we on the robotics and automation journey, and what’s next? Daphne Luchtenberg welcomes guests from some of the world’s leading robotics companies to explore the role robotics and automation can play in rewiring the front line and pushing the boundaries of productivity. In this episode of McKinsey Talks Operations, Daphne is joined by McKinsey Partner Ani Kelkar, Ujjwal Kumar of Teradyne, Marc Theermann of Boston Dynamics, and Etienne Lacroix of Vention. The panel shines a light on some recent successes and goes deeper into the adoption challenges they’re seeing.
Daphne Luchtenberg: Ani, let me come to you first. Please set us up a little bit to talk about the current landscape and what you’re seeing in your discussions with clients. How are they thinking about robotics today? And why is it a much more realistic prospect than ever before to make it an ingredient in the operations front line?
Ani Kelkar: Thank you, Daphne. I think the past 12 to 18 months have really been transformational in the robotics landscape. Interest in the topic, whether you analyze company reports or press publications, has increased five- to sixfold, and that’s also reflected in conversations we’ve had with clients. Historically, robotics was confined to certain use cases, and there were concerns about ROI, implementation costs, and the complexity of integrating it into operations. A lot of those challenges are being resolved through technology that has revitalized interest and agendas.
That said, every time we’ve surveyed executives, they remain nervous about scaling beyond pilots. For the pilots that they’ve deployed, around 40 percent of executives came back and said, “Those were exciting. There was a lot of appetite in my operations, but it was unclear what the business value actually was.” It’s not an insurmountable challenge, but we still need to navigate legacy business systems and think about this not as a purchase of a tool but as a build-out of a capability.
Daphne Luchtenberg: Thanks so much, Ani. Let me bring in our guests. I’d love for each of you to give me a quick, short value proposition about the products and services you’re bringing into this ecosystem and how you are serving customers today. Ujjwal, let me start with you.
Ujjwal Kumar: Teradyne Robotics and its companies, Universal Robots and MiR, offer an AI-powered, open, and scalable robotics platform designed for manufacturers and industrial operators to enhance automation, improve flexibility, and future-proof their operations.
Daphne Luchtenberg: Thank you. And Etienne, let me come to you. How are you part of this new ecosystem and marketplace?
Etienne Lacroix: The best way to think about Vention is LEGO Mindstorms meeting 3D CAD software meeting Amazon Prime. So Ujjwal and I are actually both believers in the concept of platforming and productization to bring simplicity and adoption to the industry in a completely different way, versus what I think we both call traditional automation.
Daphne Luchtenberg: Marc, I’d really love for you to tell me a little bit about what is burgeoning in this marketplace and who the customers are that you are serving today.
Marc Theermann: It’s an extremely dynamic market at the moment. As you know, Boston Dynamics has been a leader in the field for about 30 years now. And we’ve been developing autonomous mobile robots for much of our history. In the last four and a half years, we moved from fundamental research for governments and institutions into the productization and commercialization of our inventions. So today we have three robotic lines. We have Spot, our quadruped inspection robot that looks a little bit like a yellow dog. We have Stretch—a truck-unloading robot for the logistics industry. And we have Atlas, the most advanced humanoid robot in the world for manufacturing.
Daphne Luchtenberg: Ani, building on the opening gambit there, what made robotics adoption so difficult in the past, what’s changing, and what will it take to scale?
Ani Kelkar: If we reflect on the present moment, I think the need for robotics and automation has never been stronger. If you look at the Western world, industrial productivity over the past decade or so has stagnated from upward of 1.5 percentage points to now below 1.0 percentage points. A lot of that is due to secular investment decline, the end of Moore’s law productivity and traditional production systems, and the outsourcing of a lot of low-cost labor. In addition to the productivity challenge for the manufacturing base, we have a demographic challenge. We see significant shrinkage in workforce participation and the available workforce in many modern developed economies. There’s labor turnover and a labor shortage from a skills perspective as well.
Given all this, the need for robotics has become more pressing and a reason why robotics will be looked at more extensively. All the guests we have here are building robots that are so much more capable than robots were five or ten years ago—not just in terms of the types of tasks these robots are able to take on that they weren’t able to take on before but also in terms of the ease of use of robots and the ability to program them and change them to be able to tackle a wider variety of applications and operating contexts than in the fixed robotic mindset of several years ago.
Finally, the era of operations digitization has brought to the forefront of organizations the need to invest in people and capabilities rather than just buying solutions. That becomes a playbook for thinking about robotics as well. This is a reason to continue to invest in capability and talent and, again, not think of it as a capital-expense activity.
Daphne Luchtenberg: And Marc, robotics has taken such a long time to reach broad-scale adoption. What will it take to beat the scaling slump?
Marc Theermann: You’re right, it has taken quite some time, and there are some fundamental things left to be invented to really reach scale. Our vision is to create general-purpose robots that can go anywhere a human can go—robots that understand and manipulate their surroundings. And only when they can do all three of those things do you really have a general-purpose robot. For the past 30 years, we’ve been working on the “go anywhere” portion, and we’ve become pretty good at it. Now, our robots can go almost anywhere that a human can. The next two challenges we’re trying to tackle are semantic understanding and manipulation challenges. That’s where we’re spending most of our time. Those are the fundamental building blocks for the tremendous scale that people are forecasting for these types of robots.
Daphne Luchtenberg: Ujjwal, can you give us some examples of where this is really manifesting on the front line?
Ujjwal Kumar: Yes. We know that, historically, some automation projects have been undermined by a lack of digital skills for configuring and maintaining tools. Companies have also struggled to integrate automation technologies into existing workflows.
Now, if we want both productivity and flexibility, we need to look at a newer type of robotics. Collaborative robots can be moved around the factory floor as required. They can be integrated into existing processes. This is where, in my view, we are also going to see huge gains from AI-enabled advanced robotics in the coming year, as tasks that were previously very difficult to automate, like navigating in a complex dynamic environment, become doable.
Etienne Lacroix: I’d like to say there is nothing more manual than industrial automation. There’s a tremendous amount of manual integration that needs to take place. To give you a sense today, if you are a system integrator putting those robot cells in factories, you will need to learn the software to program a robot. You will also need to learn the software to program a PLC [programmable logic controller], and you will need to learn software to 3D design the actual robot cell. All that software is currently in disjointed environments. It’s extremely hard to learn all of this. And that expertise has a price, which makes industrial automation more costly.
Ani Kelkar: Adding to what Etienne just said in terms of the capability around software and integration being expensive, it’s also not a capability that many companies have. When we surveyed executives around barriers, 61 percent of them reported that one of the main barriers was that, even if they found a good business case, they just didn’t have the internal capability to pull it off. And so, when you are a CFO risk weighting the portfolio of initiatives, you apply a very high-risk range to anything to do with industrial automation because you’ve seen it fail so many times.
Etienne Lacroix: We often forget that the only way to know if a robot cell or automated equipment will work is to design it, purchase it, assemble it, deploy it, and then test it. And sometimes it does, and sometimes it doesn’t right away, so more manual investment is required to make the system work. That’s why CFOs put such a discount rate on the success of industrial automation.
Ujjwal Kumar: And I’ll just add, Etienne, that with digital twins, which we see more and more, some of the risks you’re talking about go away. Now you can deploy a new automated system in the virtual world, test it, perfect it, and then—from the digital twin itself—you can download the code into the production environment.
Daphne Luchtenberg: Ujjwal, can you talk a little bit more about the versatility of these robots now? I heard you speak in the past about how you don’t just buy these robots to do one thing. They can now be taught a range of different tasks and be deployed in different parts of the value chain. Can you talk a little bit more about that?
Ujjwal Kumar: The big difference is that traditional automation was a custom-made, perfect solution for one application. The new age of AI-integrated robotics has standard products serving multiple applications. You go into multiple applications through software and some end-of-arm tooling differences. This development removes the biggest barrier to drastic reduction in total cost of ownership for this new generation of automation. So, say for universal robots, we now have the world’s largest number of collaborative robot installations. We just crossed 100,000 units, but we have just six configurations. You would have needed 100,000 configurations with the previous technology!
Subscribe to the McKinsey Talks Operations podcast
Daphne Luchtenberg: Ani, the business case has completely changed, hasn’t it? Can you say a little bit more about that?
Ani Kelkar: Historically, the business case for robotics and automation was framed around five- to seven-year paybacks, often justified by longer-term production planning. That’s changing now that we have much more flexible robotic solutions. Moreover, we can simulate and optimize production systems using digital twins. This allows companies to have business cases that pay off in one to three years and reduce the integration risks with their legacy IT and OT systems, making the adoption of robotics more frictionless.
Shorter payback periods shift how companies think about risk. If a project shows an ROI within a year, you can start small, prove value, and then scale. In contrast, a five-year payback timeline carries risks of loss of knowledge, employee turnover, and shifting priorities, which makes it harder to attribute success or failure of an automation deployment and often distorts institutional memory around automation.
The other major factor is labor availability and shortages in particular. In many industries, the turnover rates are upward of 40 percent, meaning companies are constantly retraining. That makes manual, expertise-driven processes brittle. So, from a business continuity and resilience standpoint, robotics is essential and not just a cost play.
Daphne Luchtenberg: Etienne, more to say there?
Etienne Lacroix: Let’s start by setting a benchmark for industry. If you really want to move the needle for manufacturers—and there are so many manufacturers that have barely started their automation journey—payback needs to go below one year. If you’re below one year, you fly under the fiscal year budget cycle. Ani, I’m sure you have those discussions with your clients all the time, but below one year companies get agnostic between software investment or hardware investment. At the same time, you can try to eliminate that discount rate that we’ve talked about because automation projects tend to be more risky. A way to do this is to fully digital twin ahead of capex investment. Then you’re in a true comparison state.
So if that’s the benchmark, let’s talk about where the industry is today. At Vention, being an online platform where people can design their systems, program them, and purchase them, we have a tremendous amount of data that we look at every single day.
Today, if we look at a project that got deployed in 2024—we just finished it two or three months ago—roughly, we stand at 1.3 years’ payback. That’s for the robot and the surrounding devices for the robot as well. Our clients, which include Fortune 500 but also small and medium manufacturers, on average, ask us for 2.4 years to payback. If we had that conversation four years ago, I would have told you that my average payback across all the projects realized would be closer to 1.7 to 2.0 years. We’re now at 1.3 years. The path to one year is not far off, and it’s mostly driven by what Ujjwal talked about—bringing more physical AI to those machines. When machines become more autonomous, they are more agile because of AI-driven behaviors. You can apply them to more use cases, they’re easier to train, and the total cost of ownership goes down.
Daphne Luchtenberg: But of course, all of that is going to rely on an increase in knowledge and education of the folks who are making those decisions today, and growth and pivots in capabilities and skills. What are some of the critical success factors to really make these applications work?
Marc Theermann: On the application-specific level, I think you need to be customer-obsessed. More importantly, when I zoom out a little bit, we believe that the four pillars of this upcoming robotics revolution are dependability, safety, customer value, and AI. And only if those four come together will you be able to roll out autonomous mobile robots at scale and get the adoption curve and change management under control. It’s one thing to walk into a factory with a Spot, a walking robot, because it looks a little bit like a dog and a dog is man’s best friend. People are usually fairly accepting of that robot. We think that change management will be dramatically different if you walk into a factory with a humanoid robot. So, if you can’t prove that this is a safe robot that provides value to the associates on the ground, the acceptance will be particularly hard.
Daphne Luchtenberg: So what should teams do first? Ujjwal, let me come to you. If you’re thinking of starting on this journey—you’ve been exploring, you’ve been doing some background reading, you believe in the promise—what should you do first?
Ujjwal Kumar: Thinking about that integration and how you get up and running in a way in which you get quick ROI is critical. I would have two pieces of advice in that direction. Number one, leaders need to rethink automation through the lens of flexibility, not just efficiency. Traditional automation tools were built for a high-volume, low-variability environment. But today’s market demands agility. Too many manufacturers try to solve new problems using the old tools that have been used for the past 50 years. Taking a modern industry 5.0 approach requires prioritization of adaptability, empowering line workers with robots that can be reprogrammed and redeployed as demand shifts, which is the biggest benefit of having these very flexible systems coming online quickly.
Number two is to treat robotics integration as leadership and workforce transformation. Let me clarify that. Successful automation is about the people. Our technology is designed to be incredibly easy to use, but organizations should still embrace the opportunities available for workforce upskilling and look to find technology champions to help drive the cultural shift.
Daphne Luchtenberg: Wonderful. Etienne, anything you would add to that?
Etienne Lacroix: I can complement Ujjwal’s perspective with how you put that into action in the factory and what a first project looks like versus a subsequent project. What do we see at our clients who want to embark on that journey of getting more autonomous at driving their own automation road map and building capabilities? What we very often say is: First, you need a technical champion. They don’t need to be an expert. The technology is getting so simple and accessible versus just a few years ago, that if you have somebody who’s familiar with the general principle of technology, and obviously understands the craft of what’s being manufactured, that person can be an effective champion.
I always advise clients that the first project they should tackle is not necessarily the one that will drive the best ROI or the best payback. It’s the one that will deliver a success that will be visible within the organization. Usually, it’s not the most complex challenge but the one that the team, supported with the right partners—whether that’s universal robots or an invention—will make a success. Management will perceive this success. We then use that project to build the first layer of capabilities and skills within that company.
Marc Theermann: You’re right. This is a dramatic shift because it’s a little bit like introducing a new species into your facility. The best advice is that you need to start today because there are autonomous mobile robots available right now that you can purchase both from us and from our competitors. If you don’t start, you’re going to be behind because there’s no way that you can instantly go from zero to humanoid. You can’t just deploy humanoid robots tomorrow without having practiced with some sort of other autonomous mobile robot. Maybe that’s an autonomous mobile robot, or a drone, or a walking robot from Boston Dynamics. For the change management needed, you can practice with the robots that are commercially available today because you need to change your IT infrastructure, you need to create Wi-Fi in your facility, and you need to work with your workers’ council to create acceptance and appropriate behavior guidelines for both the robots and the associates. So there’s a lot of work to be done, and that work can be done today with robots that are readily available.
Daphne Luchtenberg: And Ani, this is not about replacing workers, right? This is about enhancing workers and the workforce in the front line and making everything more productive and just more operationally excellent. What would you say to that?
Ani Kelkar: One of the big shifts required to truly harness robotics and automation is moving beyond viewing it as a local continuous improvement effort. We’re increasingly seeing robotics and automation elevated to the CEO agenda. Automation is becoming a core enterprise capability that is critical for competitiveness, business continuity, and resilience, and when done right, the business case pays for itself. But to unlock that value, companies need to start with a clear North Star. What does automation enable across the enterprise and not just in pockets? That strategic vision then gives local leaders the air cover to run experiments and iterate without losing momentum when priorities shift. Looking through a long-term lens is key.
From there, as Etienne described, it’s about identifying lighthouse projects—places where automation can drive real impact and create visible wins. And just as important, it’s not about replacing people. It’s about making jobs safer, more flexible, and more meaningful—and freeing up workers to focus on higher-value tasks. There are plenty of repetitive tasks that don’t fully leverage our workforce skills. Automating those while investing in capability building helps elevate the labor force and future-proof operations.
Daphne Luchtenberg: Marc, I loved when you were talking about it taking a decade from prototype to full-scale adoption of robotics in the workplace. And we’re already starting to see robots or robotic applications showing up in the consumer environment, for example, in restaurants and in other places. How long do you think it will be before I can take a robot home to help me with the dishes?
Marc Theermann: We often get asked when we’ll start building consumer robots. And we think that it’s a super exciting vision. We want to get there, but the way is probably through service robots. Today, we’re in this industrial era where there are strong safety standards, repeatable and scalable facilities, and the robots work in very structured environments. And to make the jump from that to “I’m going to bring a robot home with me,” as an industry, we need to get through this service era where end consumers will get into closer contact with these types of robots in restaurants, retail environments, hotels, and theme parks. This way, they will get closer to the robots, maybe with a professional operator nearby. Only once we’ve gone through that era will we be able to bring these robots home.
Daphne Luchtenberg: I think that’s a great place to finish. So we’re on a journey that feels like it’s going to accelerate over the coming months and years. Thank you so much for joining us for this conversation.
