The BSC has started the computational design of candidate proteins that will form the basis of a new, sustainable nutrition for animals. To achieve this, BSC is applying protein language models (PLMs), the same type of AI models behind chatbots like ChatGPT, which rely on large language models (LLMs). PLMs, however, work with the language of proteins rather than human language. In this context, amino acids act as the vocabulary, the protein sequence serves as the grammar, and together they determine meaning in the form of protein structure and function. This work is powered by Marenostrum 5, one of the largest supercomputers in Europe, enabling the extensive exploration of candidate proteins. Through the integration of AI and supercomputing, SYNFEED is developing a data-driven approach for the future of animal nutrition.