Akzo Background

• Mostly focused on formulating science. also do molecule design, often polymeres (large scale)
• Have 1k FTE; relatively unsophisticated techniques to-date
  • Conducting thousands of experiments per month
  • Hard to link the information between experiments in a human form
    • Not sure if it is a big data issue, or an AI / ML / GenAI issue
    • have experimented with it before, but data was too big, too many variables; tried with Quantum Black (McK)
  • Talking AI with various companies; also working hard on robotics, to break the traditional approach
• AI will play a role. Scaling up and processability is important; currently we have 20k raw materials that we use, from natural to highly synthetic, and there’s a huge variability in performance
• Not focused on ultra high value molecules - very different from pharma

R&D Process at AkzoNobel

• We are designing polymers, but very empirically
• We will only know if it’s successful after applying it in paint with years of exposure. Not as sophisticated as pharma
• We work extensively with the materials science companies like BASF that do molecule design
• They know more about the application, so they can focus on designing polymers as well as the scale-up process

How does BASF work?

• They are iterating a lot on their process
  • Their feedstocks are changing - no more gas from Russia - moving to bio renewables, carbon capture etc.
  • Changing to bio-renewable polymers. Not just new monomers but also unnatural variations will be different
• Mostly an incremental industry, but there is much pressure to move to safer source
• Reducing carbon footprint is about changing production processes, changing catalysts
• They did a pilot project with Microsoft to try to work out the orientation of molecule vs catalyst - but was just exploratory

Johnson Matter is creating the catalysts (BP and Shell have been working on it for a long time, to get more efficiency)

• JM would come with new catalysts; we wouldn’t them to come up with something new because AN isn’t big enough, and it would get too expensive