Artificial intelligence could reduce drug development costs and help create more targeted clinical trials and personalized care. Participants in a healthcare industry panel at Portfolio Deep Tech 2026 cautioned, however, that the technology alone cannot resolve inequalities in access to modern treatments.
Fewer dead ends in research
Kristóf Szalay, founder of Turbine, said developing a drug can cost around $2 billion. Successful products must recoup not only their own production costs but also spending on research and development and on drug candidates that fail along the way. Without more efficient methods, that burden could grow.
Turbine began with the premise that AI could help researchers understand complex biological systems. If fewer development projects reach dead ends and require a return to the drawing board, it may become economically viable to develop drugs for smaller patient groups that previously did not justify the investment.
Szalay stressed, however, that simulation does not automatically eliminate the need for real-world tests. He used Formula 1 to illustrate the limits of models: even calculations based on well-understood laws of physics can miss effects that only emerge on the track. Biological models are subject to similar limitations.
More targeted clinical trials
Erik Bogsch, director of the biotechnology business at Richter Gedeon, said there is currently no registered drug invented by AI, although such developments have reached phase 3 clinical trials. For now, the industry sees the technology primarily as a tool that augments human intelligence.
Bogsch expects the greatest change in clinical trials. AI could help design shorter, more focused studies and reduce failures caused by flawed research design. He estimated that clinical costs could fall by roughly 35–45% in the coming period.
Applications extend beyond the search for new molecules: the technology could narrow the range of preclinical tests and help design studies involving volunteers. Some pharmaceutical companies are also experimenting with digital twins, for example to model which combination therapy might suit a particular patient.
Faster and cheaper development, however, does not guarantee that every patient can access modern treatments. Panelists said generic and biosimilar medicines will continue to play an important role in reducing global disparities.
Europe and personalized care
Zalán Szántó, associate professor at the University of Pécs, said Europe should not compete with US and Chinese healthcare systems on the volume of procedures. As an example, he cited a thoracic surgery center in Shanghai that he said performs six times as many operations each year as are performed across Hungary.
He sees Europe's advantage instead in individualized care: consistently tracking the patient's journey, the procedure and the postoperative period. The high-quality data generated in this way could also be valuable for AI development.
The thoracic surgeon also addressed robotic surgery. He stressed that robots do not currently operate independently and that the surgeon remains responsible for every movement. Precise control and detailed visualization can nevertheless help in complex cases, particularly after immunotherapy, when changes in the tissue environment can make surgery more difficult.
The discussion also covered early diagnosis and the combination of imaging and blood-based tests, research into autoimmune and neurological diseases, and persistent gaps in gynecological care. Participants discussed AI's role in supply chains and communication with patients. The panel was moderated by Márton Baranyi, project manager at Portfolio Csoport.



