Amadeus and Nvidia have signed a memorandum of understanding to explore faster, more scalable and more efficient flight-search technology as AI assistants generate greater volumes of increasingly complex travel queries. The collaboration covers GPU- and CPU-based performance improvements, travel-specific artificial-intelligence models and more efficient cloud infrastructure, while retaining the accuracy and bookability of displayed flight options.
The companies have completed initial technical assessments and benchmarking on Nvidia’s Hopper and Blackwell platforms. Using Nvidia’s CUDA computing platform, Amadeus adapted part of its flight-search process so that multiple calculations could run in parallel on graphics-processing units. Early tests indicated potential improvements in response times and scalability, with broader evaluation planned across additional travel-search workloads.
The partners are also examining the potential for travel-specific foundation models, potentially developed with other industry participants, to address requirements that general-purpose systems may not handle effectively. In parallel, they are exploring cloud-resource planning and workload optimisation to improve infrastructure utilisation while maintaining service performance and reliability.
The MoU is an exploratory technology-development agreement. No commercial launch date, quantified performance improvement, named customer, production deployment or financial terms were disclosed.