Mobility Insights
ROI, not hype: how carmakers are really adopting smart robotics
At our latest Mobility Mixer in Paris, Shift4Good Co-Founder and Managing Partner Matthieu de Chanville moderated a panel with Arnaud Deboeuf, former EVP Manufacturing at Stellantis, and Louis Dumas, CTO of Inbolt. Arnaud brought the carmaker’s perspective, from a career spanning Renault, the Renault-Nissan Alliance and Stellantis. Louis brought the view of a startup building the technology: Inbolt uses AI vision to let robots see, train and adapt their movements in real time, and is deployed in more than 100 factories worldwide. Here are five takeaways from their conversation.
1. Vision finally brings value to robots
Industrial robots have been around for about 40 years, and machine vision for more than 20. Until recently, the two didn’t work well together. Arnaud recalled that 20 years ago, vision-equipped cells were more of a constraint than a benefit: they slowed the line, struggled with changing light and flagged false defects.
Without vision, robots simply follow programmed paths. If a part is slightly out of place, they stop. Arnaud said this is why the body shop, where a car’s metal frame is welded together, typically runs at around 80 to 85% efficiency, compared with about 98% in paint and final assembly. Robots that can see and adjust in real time, without slowing the line, could close much of that gap.
2. ROI drives everything
For Louis, the purchase decision always comes down to return on investment.
“ROI drives everything. Every car with a defect is a car that won’t be sold. Every time a robot stops, that’s a car that isn’t getting built.”
What counts as value depends on the workshop. In the paint shop, it’s quality, because a paint defect means an unsellable car. In the body shop, it’s uptime. In stamping, it’s output: more parts mean more cars. Inbolt’s payback ranges from one month to one year, with a target average of three months. Arnaud added that payback also depends on location, and is faster in high-wage countries like the US than in lower-wage markets.
3. Upgrade, don’t replace
Both speakers stressed that smart robotics doesn’t have to mean new robots. Carmakers have invested too much in existing equipment to scrap it, Arnaud said, and the opportunity lies in bringing intelligence to robots already on the line, even ones that are decades old. Louis confirmed that about half of Inbolt’s deployments use existing robots, either ones running with faults or ones being repurposed for new tasks.
This is also where ROI and sustainability meet. Reusing robots avoids the footprint of building new ones, better quality means less scrap, and more reliable robots waste less energy.
4. The real divide is legacy vs newcomers
Asked whether adoption differs between Europe, the US and China, Arnaud argued that geography matters less than company strategy.
“I don’t see a regional split, because carmakers have a global strategy. There isn’t a Chinese approach or a European approach. There’s a company approach. On one side you have the legacy players, and on the other the newcomers. Even in China, you have the legacy state-owned companies and then the newcomers, which are very agile.”
Louis gave an example: an automation engineer at a Chinese newcomer told him all of their body-shop robots are equipped with vision, while a legacy plant his team had visited had none.
Geography still matters for startups, though. Inbolt keeps its engineering team in Paris, where Louis estimates one US engineer costs as much as five in France. Yet the company quickly needed an office in Detroit, because US customers buy faster and the US is where it can scale.
5. Less data, more value, and the long game on EVs
Carmakers have always been data-driven, Arnaud said, but the challenge today is having too much data. He called for frugality: select what matters, process it locally on the robot, and avoid storing data nobody uses, which wastes energy. A real-time dashboard in an executive’s office adds little. A more useful example is a Stellantis maintenance tool: every night, plants report the technical problems they faced and how they solved them, and AI combines this with supplier documentation to support diagnosis and problem solving across all plants.
On the shift to electric vehicles, the speakers saw a long-term opportunity with short-term friction. For Arnaud, electrification creates new workshops, such as battery shops, where new technology is easier to introduce. Inbolt’s deployment at Stellantis’s Sochaux plant came with a new battery shop. Cost pressure from Chinese competition also makes fast-payback automation more attractive. Louis added a caveat: EV volumes at legacy carmakers are still low, which weakens the ROI of automation for now.
See Inbolt’s robots at work: Bolt tightening · Robot programming
