Why Physical AI May Outsmart Digital – The Surprising Edge of Atoms

What we talked about.

This episode of the Sunny Ray Show features Oliver, managing director of ENAM, the Innovation Network for Advanced Materials, discussing why atoms may outpace bits in shaping the future of AI and robotics. Oliver traces his unconventional path from social anthropology and European policy work through consulting and corporate turnarounds into advanced materials. He explains how ENAM connects startups, investors, and corporates globally to commercialize materials innovations, from semiconductor advances like silicon carbide and gallium nitride to novel approaches for data center heat management. Oliver reflects on the chip supply chain crisis during COVID as a turning point that revealed materials as a true bottleneck for compute and AI hardware. He highlights the convergence of AI and materials science, citing Berlin based startup Duna Innovations as an example of self driven labs combining AI with robotics. The conversation also covers financing challenges unique to materials startups and why strong research often fails to translate into successful products without deep market engagement.

A former anthropologist turned materials investor argues atoms, not bits, may hold the real key to AI's future.

The questions, and the answers.

You studied social anthropology, not engineering. How does a kid end up curious about how people and societies actually work?

I started with economics in my bachelor's, which was quantitative, but wanted to explore other approaches to society. I got into a prestigious London based program and was hooked by philosophy and history of science. My master's thesis was actually about the internet's impact on knowledge management, which was unusual for anthropology but showed how flexible the subject could be.

What was it like being a trainee inside the European Commission?

It was a lot of fun and impressive to see how the EU institutions work at their core. There's a lot of legal and regulatory work, but also a lot of space for creativity and international exchange, since Brussels and the EU institutions are inherently international. I'm still in touch with people from that half year traineeship.

At what point did technology stop being a side project and become your main thing?

Probably during my Bundestag days when I worked on tech related legislation like anti-terrorism measures and data privacy reform. But it may have started earlier. As a child visiting relatives in Bavaria, I read old newspapers from the late 50s and 60s obsessed with technological progress, and that planted the first hook for my interest in technology and innovation.

You keep coming back to the line between hardware and software. Why does that matter so much?

Running ENAM, I'm rooted in the hardware space, but software's power can't be ignored. Materials innovations massively support software's evolution, for example moving from silicon to silicon carbide or gallium nitride in chips makes hardware more powerful and energy efficient. Materials also improve data center heat management and enable waste heat reuse, giving huge leverage to improve compute capability.

When did you realize materials and compute, not software, are the real bottlenecks for AI and robotics?

It was during the COVID era supply chain problems, when a lack of chips hit the German automotive sector hard. Looking at chip value chains, everything starts with materials. We began a project to find alternative materials, recycle them from existing products, and explore completely novel solutions, like lab grown semiconducting crystals, that could outperform current setups.

Which material breakthrough excites you most in the next five years?

It's less about a single new material and more about software coming back to materials, meaning AI aided materials development. Ideally you'd say 'material is GPT' and describe the properties you need, and AI would tell you the ingredients, temperatures, and reactions required. Startups like Duna Innovations in Berlin are already combining AI with robots to automate this mixing process.

You say world class research doesn't automatically become world class startups. Where does it break?

Researchers often have a bias toward their own innovation and assume everyone will want it, but a lab innovation isn't automatically a product, and a product isn't automatically sellable. Many break in translating product to market because researchers don't talk enough to industry. Our startups also often work at the input material level, so they must figure out which market and application to address first.

Where is Europe ahead of the US or China on advanced materials?

That's difficult to say. There are really three or four, actually five, major markets doing good work globally: the US, Canada to some extent, Europe, China, which has been massively catching up, and Japan, which is traditionally strong in relevant fields. Because these markets are inherently global, we act globally too, with corporate partners, investors, and startups from around the world.