Case 01: Schneider Electric

AI on a large scale

Schneider Electric, the electronics company, recognised the potential of artificial intelligence early on and decided in favour of a strategic approach. A Chief AI Officer and several AI hubs in different countries pursue an AI strategy with the focus on scalability.

“We wanted to avoid simply introducing AI applications into the company at all costs as part of individual pilot projects or for a single digital factory”, says Philippe Rambach, who became the first Chief AI Officer (CAIO) at Schneider Electric, a French manufacturer of electronic components, when the position was created two years ago. His job is to implement artificial intelligence on a large scale. “Our goal is to specifically combine corporate knowledge from our Energy Management and Industrial Automation business divisions with AI knowledge”, he explains. “We’ve set up AI hubs for this purpose to connect AI experts with our domain experts.”

Philippe Rambach

Chief AI Officer at Schneider Electric

His first official act was consequently to identify people at the company with AI expertise and bring them together at three newly established AI hubs in India, France and the United States. He has systematically expanded the team ever since to its current size of 250.

These specialists possess the necessary knowledge to programme AI applications, develop and teach models and analyse data. They collaborate with all specialist departments as the central point of contact for AI issues.

According to Rambach, they are working towards three goals here. Firstly, “We want to automate our own processes and make them more efficient”.

For example, a team from the AI Hub has developed an application that lets Schneider Electric assign customer enquiries faster to the right domain experts. That saves money across all business divisions.

Secondly, “We’re developing new, AI-powered products and services and upgrading our existing offerings with additional AI applications”. The AI-driven innovation platform known as EcoStruxure Microgrid Advisor, which helps companies automatically forecast and optimise energy consumption, generation and storage, is just one good instance. Last but not least, “A specially established AI consulting team advises customers on how to use AI applications meaningfully”. The team provides consulting services to end users in the energy and chemical industries, amongst others, as well as in utilities.

The investments in AI are already paying off in many ways, as Rambach reports: aside from in-house efficiency gains, Schneider Electric has also been able to increase its innovative strength thanks to the AI strategy. Since launching the “AI at Large” strategy, the company has already filed for 18 AI technology patents, developed 20 internal AI applications and enhanced 15 of its own products with the help of artificial intelligence. The most important lesson learned during the last few years is this: “It’s strategically crucial to always start with the business model and the problem to be solved rather than by asking what a new technology may or may not be able to do”.

Case 02: Neura Robotics

Robots powered by common sense

Neura Robotics, a start-up from southern Germany, develops robots that use sensors to collect sensory impressions and then evaluate them with the help of AI. That opens up new opportunities for manufacturing companies.

Established robot manufacturers and start-ups are working flat out to develop market-ready cognitive robots. One still-young firm from Germany has pulled off an innovative leap: shortly after its foundation in 2019, Neura Robotics of Metzingen, not far from Stuttgart, unveiled the world’s first cognitive robot ready for serial production – attracting considerable international attention. “MAiRA” can navigate autonomously because it is able to perceive its surroundings in real time. The upshot is greater precision and efficiency – as well as better and more flexible interaction with human colleagues.

David Reger

Founder and CEO of Neura Robotics

Sensors for environmental perception are nothing revolutionary in robotics. Yet as David Reger, CEO of Neura Robotics, explains, “The interaction of sensors and artificial intelligence opens up a range of potential new applications”. Robots that, like MAiRA, are developed for interactive collaboration with human beings are commonly referred to as cobots. Cobots built by Neura Robotics are used, amongst other things, in measuring and welding technologies as well as for deburring or in logistics and assembly.

One major difference compared to classic industrial robots, Reger continues, is that, “Normally, a robot plans its movements in advance and then plays them back over and over again largely unchanged”.

His cobots, on the other hand, recalculate their movements from scratch once every millisecond, meaning they can react directly to external influences. Welding work is a good example here: in the quest for the perfect weld, highly qualified employees have traditionally been key – employees who knew instinctively whether they were welding in the right place, feeling the right pressure and hearing the right noise. Now, thanks to builtin microphones, 3D cameras and torque sensors, cognitive robots can do all of this too – easing the burden on firms against the background of skills shortages.

Apart from the hardware, Neura Robotics also provides an open platform – the “Neuraverse”, as Reger calls it. Here, partners can develop their own custom applications and use them to teach the robots new tasks. In other words, the cobots engage in “lifelong learning” at their “workplace”.

Reger is convinced that only through intensive development cooperation between AI specialists and user companies will cognitive robotics make inroads into industry on a large scale. Neura Robotics, for instance, is collaborating with the Japanese robot manufacturer Kawasaki on a joint product series. The aim is to combine industrial-scale automation with the benefits of collaborative robotics.

Interest in these cognitive robots is high: according to Neura Robotics, orders worth more than 450 million euros are in the offing – and Reger is certain that actual demand among customers is much higher.

Case 03:

AI chips for mobile applications, the Californian start-up, has delivered a huge leap in performance with AI chips that surprised many established competitors – and enables new AI applications.

Who builds the best, fastest, most efficient and most performant AI chips? In the search for answers, the US NGO MLCommons invites manufacturers to measure their performance against one another twice a year in various categories with their latest developments.

Krishna Rangasayee

Founder and CEO of

One surprise winner in these benchmarks, which go by the name of MLPerf, was, a start-up from San José, California. Founded five years ago, the company submitted a chip that performed an amazing 50 percent better than the previous record holder, Nvidia’s Orin AGX chip, in terms of speed and power consumption in the “Closed Edge REsNet50” category.

What this means in practice is that data from the operation of mobile, networked devices and machines, such as cameras or robots as well as vehicles, can be processed significantly faster with the Sima chips than with models that were previously industry-leading. What’s more, their power consumption, too, was much lower.

Performant hardware can be crucial in order to get the best out of artificial intelligence. Not all AI chips are the same, though: the design and the performance data must fit the intended application. The decision as to whether the accompanying software should operate in a cloud or on internal servers, for example, likewise plays a role.

The highly specialised chip outperforms its predecessors in one key area, namely when used in mobile devices, which often have to function with comparatively low power and outside the range of a network. In the past, AI chips capable of performing at this level have generally been too expensive and consumed too much power for this purpose. The upshot was that they were often restricted to large data centres and cloud applications.

Krishna Rangasayee, founder and CEO, is proud of what he has accomplished. 53-year-old Rangasayee, a Stanford graduate, has been a fixture in Silicon Valley for 25 years now and holds several high-profile patents in the software and chip industries. Yet above all, he is eager to see his new chips extended to real-world use: serial production kicked off in June 2023.

The company has also built a partner platform giving customers access both to the chips themselves and to the software integrated on them. The CEO is unwilling to reveal who his first customers are and what exactly his chips will be used for. All he will give away is that is already actively collaborating with more than 50 market leaders in the manufacturing, retail, automotive and aerospace sectors as well as with a number of government agencies.

Rangasayee anticipates that established industrial firms will in future cooperate more closely with hardware manufacturers like when developing new products. After all, the realisation of a world in which autonomous vehicles, surgical robots and smart factories are commonplace will depend to a large extent on the hardware. And he firmly believes that AI chips can achieve further leaps in performance. “We’ve only just scratched the surface of capability and performance.”


„The next one to two years will be crucial“

In an interview with SCIO, the leading AI researcher Damian Borth explains which AI experts businesses need right now and how the global AI race will continue.

Mr Borth, what do companies need to bear in mind when using AI?

Prof. Dr. Damian Borth

Director of the Institute of Computer Science at the University of St. Gallen in Switzerland, Professor for Artificial Intelligence

We’re at a similar stage today with artificial intelligence as we were with software development back in the 1990s. In other words, we’ve built the first AI solutions and we’re now testing whether the technology from the lab also works in the real world.

It’s only natural that we’ll run into problems with some aspects, and we’ll be obliged to tackle them and find solutions. However, that’s no reason to shy away from AI – on the contrary! We now need to address the question: How should I test AI models and how do I ensure that they serve my purpose reliably and securely?

How fast do businesses need to act?

The big breakthrough came in 2012 with deep learning. AI with the ability to analyse data has been with us for a decade now. And since generative AI arrived on the scene in 2022, we’ve also got a system that’s capable of generating images, audio and text autonomously from data. As you can see, we’re advancing in leaps and bounds. If they haven’t already done so, businesses would be well advised to restructure now and create new departments to deal with these developments.

What might that look like in practice? What will businesses have to do to become “AI-ready” and use the technology meaningfully?

Many firms are still in the early stages. You can find the odd Chief Digital Officer here and there, of course, as well as various other departments driving the digitalisation of business models and processes. AI falls into that same category because once the processes are digital, that’s when deep learning really gets off the ground. Someone like a Chief Data Officer would be another position that could drive these opportunities forward. Either way, departments must be empowered to try out and use new types of software like open source.

You yourself have conducted research in Taiwan, Germany and at Columbia University and Berkeley in the US. What is your verdict – can Europe keep up in the current race to develop more and better AI applications?

The market in Europe is smaller and there’s less money available than in the US. That’s also reflected in the salary structures. At present, highly qualified experts are actually more likely to head off to the States because they can earn better money there. But things are happening in Europe too. In Germany, for example, we have excellent universities that are at the forefront when it comes to international research. Europe is keen to learn lessons from the last decade and so more and more promising start-ups are emerging.

Will AI continue to develop with such great strides as it has done up to now or can we expect to see the curve flatten off for a while?

It won’t be flattening off for some time yet. Quite the opposite – the pace of development is all set to accelerate even more. Things are going to get crazy during the next year or two. There are some crucial months ahead for researchers and businesses as new applications are discovered for AI.

Which sectors can especially benefit from this?

Manufacturing companies which, depending on the sector they operate in, are already highly automated today will experience a new wave of automation. Above all, the combination of AI and the Industrial IoT – the Industrial Internet of Things – will drive this development rapidly forward in the coming months and years. In mechanical engineering, new possibilities will be opened up for computer aided design. New drugs will be developed much faster in the pharmaceutical industry. In the chemical industry, we can look forward to a new understanding of molecules and their interactions. And in materials research, new synthesis processes will become established. In short, a lot will change in manufacturing firms’ core business. Not only that but a lot will also change in domains outside their core business.

Which areas are you thinking of?

Basically all office jobs – from marketing to human resources. It’s not like the industrial revolution, where steam engines took over physical work, for instance by moving heavy objects. AI can’t build houses for us or take the place of craftsmen. We live in a knowledge society. AI will help us solve intellectual problems and automate intellectual work.


Prof. Dr. Damian Borth is one of the most distinguished European researchers in the field of artificial intelligence. He has conducted research in Europe, Taiwan and the US, including at Berkeley, renowned for its AI research, and Columbia University. He is considered a pioneer where deep learning is concerned. Borth is currently Director of the Institute of Computer Science at the University of St. Gallen in Switzerland, where he holds a Professorship in Artificial Intelligence. He was previously founding director of the Deep Learning Competence Centre at the German Research Centre for Artificial Intelligence (DFKI).