Exhibitors & Products
Events
Speakers

The company is developing a platform that serves as a kind of “search engine for materials”. It uses machine learning to design new materials for various applications. The first areas of application include.

  • Gas separation and carbon capture: New materials could significantly improve the efficiency of CO₂ capture.
  • Semiconductor materials: As chip miniaturization progresses, materials research is becoming increasingly important to find new, more efficient alternatives to silicon.
    •

    The platform is based on generative AI models that generate and simulate new molecules. In addition, CuspAI uses physics-based models to predict the properties of these molecules.

    Technological foundations and AI models

    The platform relies on various methods:

  • Generative models that design new molecules and condition their properties.
  • Force fields that enable physical simulations by calculating the energies and forces of atoms.
  • Property predictors that bypass simulations by directly predicting material properties.
  • AI-powered search agents that efficiently filter out the best materials from a large set of potential combinations.
    •

    Welling emphasizes that AI enables enormous progress in materials research. Simulations that used to take weeks can now be carried out in seconds thanks to AI-supported models.

    Data sources and collaboration with companies

    CuspAI draws on various data sources, including:

  • Public databases such as Materials Project and Nomad.
  • Company specific data provided through collaborations with companies.
  • In-house simulations that generate synthetic data to improve the AI models.
    •

    Combining domain-specific knowledge with AI is a particular challenge. Welling describes how many companies have data, but not the necessary machine learning experts to use it efficiently. CuspAI offers a bridge between material science and AI here.

    For the next phase, Welling is still looking for excellent researchers in the fields of machine learning and chemistry. The company's locations - including Cambridge, Amsterdam and Berlin - offer an attractive environment for European talent.