Amadeus and NVIDIA have signed a memorandum of understanding to deepen their collaboration on flight search.
The agreement focuses on accelerated computing and travel-focused artificial intelligence.
It is intended to address changes in travel search as AI assistants increasingly act on behalf of travellers. That shift is increasing both the volume of requests and the complexity of the computing needed to return flight options that are available and bookable.
The collaboration is organised around three areas: improving flight-search performance, assessing travel-specific foundation models, and examining cloud and infrastructure efficiency.
In the first area, Amadeus has been testing the use of Graphics Processing Unit and Central Processing Unit technologies in parts of the flight-search process. It used NVIDIA CUDA to adapt one element so multiple calculations could run simultaneously on GPUs.
Initial technical assessments and benchmarking have already taken place on NVIDIA Hopper and Blackwell systems. Those tests were designed to assess whether accelerated computing could support travel shopping workloads at large scale.
Early tests showed potential to improve response times and scalability, and Amadeus plans to assess the approach across a wider range of travel-search workloads.
Search demand
The second part of the collaboration focuses on travel-specific foundation models. These models could help address travel-related problems that general-purpose models may not handle as effectively, while also improving existing travel-specific models.
That work is being assessed alongside other participants, although no additional partners were named. The aim is to respond to emerging travel use cases as digital assistants become more involved in trip planning and booking.
The third area concerns cloud and infrastructure planning. Amadeus and NVIDIA are examining ways to improve the use of cloud resources and optimise workloads while maintaining service performance and reliability.
For travel sellers, the work is intended to support more complex requests as AI-generated search traffic rises. For travellers, the focus is on returning options quickly while ensuring displayed fares and itineraries remain available to book.
Technical focus
The collaboration reflects a broader shift in how search systems are being designed for the travel sector. Traditional consumer search typically involves a person entering a limited set of criteria, but AI assistants can generate more queries and request more complex combinations of routes, pricing and conditions.
That places pressure on systems used by travel technology providers, airlines and sellers, particularly where search results must also be accurate and directly transactable. In airline retailing, speed alone is not enough if an itinerary cannot be booked at the price shown.
Amadeus occupies a central position in that process as a technology intermediary connecting travel providers, sellers and other digital channels. It described its role as a neutral execution layer linking those parties.
Lutz Vorneweg, Senior Vice President, Strategic Partners at Amadeus, described the challenge in operational terms.
"As AI transforms the travel ecosystem, maintaining fast, accurate and bookable flight search at global scale requires new approaches. As the neutral execution layer that orchestrates connections between travel providers, sellers and digital assistants, we are excited to work with NVIDIA to explore how accelerated computing and AI technologies can help power the next generation of travel search," said Vorneweg.
NVIDIA framed the agreement as part of a broader rise in computing demand driven by the spread of AI in industry-specific applications.
"AI is unlocking new possibilities across travel, from more intelligent search to emerging uses of AI assistants, and creating new demands on computing infrastructure. Our collaboration with Amadeus combines NVIDIA accelerated computing and AI with deep travel expertise to explore faster, more efficient flight search at global scale," said Hogan.