Generative Artificial Intelligence and Supercomputing: The New Frontier of Pharmaceutical Research
Technological innovation in the pharmaceutical sector is undergoing an unprecedented acceleration. At the heart of this transformation, Generative Artificial Intelligence (Generative AI) and supercomputing are progressively redefining research paradigms, enabling increasingly predictive, personalized and efficient approaches to drug development.
Andrea Beccari, Head of Discovery Platform and Vice President at Exscalate, addressed this topic at the conference “Generative AI in Life Sciences: Navigating Wisely in a Changing World”, organized by AboutPharma. In his remarks, Beccari highlighted the central role of AI in reshaping modern medicine, offering a clear and insightful overview of both the challenges and the opportunities now emerging for pharmaceutical research.
“Generative AI represents the promise of a new evolutionary phase for medicine and pharmacology,” Beccari noted. “Thanks to supercomputing and new algorithms, we can finally tackle the complexity of biological systems in a more systematic and predictive manner.”
Exscalate: A Pioneering Platform for Drug Discovery
Within this landscape, Dompé’s Exscalate platform stands out as one of Europe’s most advanced initiatives. Created to accelerate the drug-discovery phase, Exscalate leverages supercomputing and AI models to rapidly analyze the world’s largest structured molecular library, identifying promising candidates for preclinical and clinical development.
By adopting Generative AI early in the preclinical arena, Exscalate has been able to evolve its algorithms at pace. Its technologies now cover strategic domains including:
computational molecular design (generative chemistry);
prediction of drug–receptor interactions;
assessment of safety profiles;
identification of the most relevant pathological targets.
A New Conception of Scientific Data
One of the most revolutionary aspects of this approach is the shift in how scientific data are conceived and managed. As Beccari explained, it is no longer merely a matter of collecting experimental data; the challenge now is to actively generate data—real or synthetic—that can effectively train AI models.
The use of synthetic data—computer-generated information with characteristics like real datasets—helps overcome ethical constraints and dataset bias, enabling, for example, the creation of synthetic cohorts or digital twins to enhance the statistical robustness of clinical studies.
At the same time, the move towards graph-based databases (knowledge graphs) has made it possible to analyze not only isolated data points but also the relationships between them, improving our understanding of pathological mechanisms and therapeutic responses.
The European Context: Data and Infrastructure as Strategic Levers
Despite these promising developments, Beccari stressed that fully harnessing AI in medicine requires accessible data and advanced technological infrastructure. Although the European Union mandated data sharing as early as 2004, it is only since 2016 that a real framework for making such data usable has begun to take shape.
Countries such as the United States and the United Kingdom are already frontrunners in this field. To ensure Europe does not fall behind, it is essential to foster an ecosystem that capitalizes on existing information assets and strengthens the global competitiveness of European scientific research.
Watch the full interview