Compular: Shaping the Future of Computational Material Analysis

Designing advanced materials is at the heart of human progress, making our concrete stronger, fuel cells more efficient, and batteries more energy-dense. Yet, a lot of this progress relies on trial and error and performing a vast amount of lab experiments. 

What if you could simulate a material’s properties on your computer and analyze the results quickly and efficiently? Scientists have been developing such simulation methods for decades, but they haven’t been very accessible outside of academia. 

The startup Compular is here to change this by developing a platform to make computational material analysis straightforward. Founded by Johannes HenrikssonRasmus AnderssonEmil Krutmeijer, and Fabian Årén, it raised a pre-seed round from Icebreaker VC and Chalmers Ventures in 2022.

Learn more about the future of computational material analysis from our interview with the CEO, Johannes Henriksson

Why Did You Start Compular?

My co-founder Rasmus did his Ph.D. in engineering physics, researching electrolytes to make batteries for electric vehicles more efficient. But there wasn’t any good tool to analyze material properties from the ground up – and he went on to create one himself. His patent-pending algorithms now form the backbone of our analysis tool, which helps researchers evaluate data from experiments or simulations like density functional theory or molecular dynamics. 

He wanted his research and this analysis tool to have some real-world impact instead of getting stuck in some drawer. So we got introduced to each other as a founding team through the innovation department of Chalmers University, and it was a great match: We had similar, high levels of ambition, and we all really enjoyed working on a startup whose technology has big potential to improve material development, battery storage, and ultimately many other domains. 

How Does Computational Material Analysis Work?

We simulate material properties mainly for batteries, such as electrodes, electrolytes, and their interfaces, to improve their energy and power density, charge capacity, fast charging and more. We are very flexible in terms of methods, using whatever suits the problem setup best, and our main innovation is around analyzing the simulation results. There we look at the trajectory of a simulation, e.g. tracking the distances between molecules and which bonds are forming, to elucidate the resulting material properties. 

We built a very user-friendly interface, which RnD scientists can use without requiring a Ph.D. in computational physics. Working in the cloud allows us to scale these simulations and analysis based on our customer needs and eventually perform thousands of simulations, saving lots of costly lab experiments. Ultimately, lab experiments are still our benchmark, and we have shown in the past that we can closely resemble their results – but that we’re a lot faster and more cost-efficient. 

How Did You Evaluate Your Startup Idea?

We got the first projects going with battery manufacturers in the Nordics. It’s still mainly consulting, but it helps us develop our platform further to address more complex simulation problems and learn about our customers’ needs. We continuously implement the learnings from our customers in our product development, and as we are now starting to hear the same customer requests repeatedly, we will shift to focus more on scaling our offering. 

Before raising our pre-seed round with Icebreaker VC and Chalmers Ventures, we have been bootstrapping Compular for two years, determining problem-solution fit and developing our technology. Now, working closely with our customers, we’re working on product-market fit: Figuring out how much value we can deliver and how much of that we’ll be able to capture. 

We got a long waiting list of interested clients and a clear roadmap to expand also to fuel cell materials, alloys, or specialty chemicals. Our analyses are currently based on computational chemistry and statistical physics methods, but as we’re collecting lots of data, this will become valuable on its own e.g. for training machine learning models. There is also still more trust in the industry in machine learning models evaluating experimental data instead of just relying on computer simulations – and we’ll be combining the best of both worlds and establish confidence in first-principle simulation approaches. 

What Advice Would You Give Fellow Deep Tech Founders?

Deep tech is really different from building a consumer brand. Product development and testing take a lot longer. But one thing is absolutely the same: You need to listen a lot to your customers! You need to be curious but patient and build a complementary team to address all the various challenges that will come up. 

Comments are closed.