Siemens Senior Director for Battery Industry: Industrial AI Is Coming

March 19, 2025

Sebastian Blanco

In late 2024, Siemens joined the Global Battery Alliance. (Siemens)

Siemens is working to make industrial AI a reality. But, to integrate generative AI into industry, the AI has to be trained on the language of industry — engineering and manufacturing languages — not just the modalities of text and speech and images. We spoke with Puneet Sinha, senior director, battery industry, for Siemens Digital Industries Software.

You said you train AI on these industrial languages. How do you do that? What is the information you’re feeding into the model?

Our focus is to train large language models, not only on the current modalities. So, for example, how will AI train on CAD data? How can it understand the manufacturing processes that are happening or the business processes that are happening.

Puneet Sinha, senior director, battery industry, Siemens Digital Industries Software. (Siemens)

In the industrial world, there are rightful questions about who can access certain data within a company or at a supplier. And if you are going to use AI to enhance digital twin technology creation or the execution of it, there should be minimal hallucinations. They should be trained on the business processes and engineering processes and language of that company.

So some of the things that we are heavily investing is how we can bring retrieval-augmented generation in our portfolio. We are implementing that with Teamcenter, which is our PLM platform, so that not only can companies train and leverage the generative AI, but they can also do it on their business processes, and they can control who can access that data. Those are the elements of industrial AI.

AI retrieves data sort of how we create memories, through re-creation. Is that reliable enough in an industrial setting? You said it has to have minimal hallucinations, but wouldn’t you need that to be zero?

Yes, absolutely. Siemens is looking at how we can bring the power of AI so that companies can actually make decisions with it, being confident that they are not having hallucinations. We are getting a lot of customers using comprehensive digital twins for battery. In simulation, you can account for all these domains — mechanical, electrical, electronic, material, software, system — and the tradeoffs, what is the implication of the chemistry or how it needs to be cooled or the safety implications. There are simulation suites that we have that account for all of those things.

When designing, especially during validation in terms of how it needs to be cooled, what the thermal management strategy needs to be, how the battery pack needs to be charged, and fast charging strategy, all of that can be now done in a really short time, leveraging the power of generative AI that is trained with the not only with the testing data, but also synthetic data coming from simulations, and that is helping engineers to explore new designs and validate designs against various requirements in a really short time.

Do you see automotive engineers using AI more to reduce development time or to explore more options?

It is absolutely a combination of two. We now have embedded these, you can think of them chat bots, in our simulations. If companies have done simulations, let’s say at the system level, rather than recreating a lot of those results or extracting new analysis, engineers can easily interact with the software, asking,‘ If that battery is used, what is the range going to be for these different driving styles?” It’s a lot more interaction, so you can generate a lot more analysis more naturally.