ChinaTravelNews, Ritesh Gupta – Tech giants and platform ecosystems may invest heavily in training massive foundational models in China, but Trip.com Group is taking a pragmatic approach to the artificial intelligence (AI) revolution.
The company is funnelling resources into application-level refinement (counting on existing, pre-trained AI models), banking on long-term efficiency gains where automation and hyper-personalisation streamline operational costs, boost conversion rates, and offset initial technology outlays.
Addressing questions regarding the financial weight of technological upgrades, Trip.com’s Chief Financial Officer Cindy Wang outlined a disciplined capital allocation strategy during the recent Q2 earnings call. When asked whether shareholders should anticipate significant capital expenditure spikes driven by AI initiatives, Wang explained that while near-term spending will see a moderate increase to expand computing infrastructure and capabilities, “most of our work is application-oriented development and post-training refinement rather than building large-scale foundational models.”
By intentionally avoiding the massive expenditures required to train foundational models from scratch, Trip.com is ensuring that its incremental investment remains controlled and manageable.
As competing ecosystem players race to focus on aspects like conversational tools and superapp assistants, Trip.com’s leadership is betting that competitive advantage will come less from building the underlying foundation model and more from applying AI across the realities of travel search, transactions and fulfillment.
AI Playbook
Executive Chairman James Liang framed this strategy around a clear premise, noting: “We believe AI will meaningfully reshape how travellers discover, search for, and ultimately book travel. It changes the user interface but does not eliminate the underlying need for high-quality travel supply, real-time availability, transactions, and fulfillment.” Liang mapped out the modern traveller journey across four distinct stages: inspiration, search, transaction, and fulfillment, detailing how the platform is embedding intelligence into each phase.
At the inspiration stage, where travellers typically present broad or unstructured desires, Trip.com expects AI agents to become increasingly important discovery and inspiration channels. Recognising this shift across the broader digital commerce and transactional platform landscape (where ecosystem competitors are deploying their own conversational discovery interfaces), Trip.com is actively expanding partnerships with AI platforms and exploring emerging discovery mechanisms to keep pace with evolving consumer behaviour. Simultaneously, the company is deploying AI internally to scale and enrich its travel content creation.
However, the real economic value crystallises once intent sharpens into search and transaction. As Liang noted, once a traveller knows roughly what they want, success hinges on timely, complete, and accurate data covering inventory, real-time pricing, availability, and strict cancellation policies. To capture this, Trip.com has rolled out fully AI-powered search algorithms designed to elevate user intent matching and filtering. Crucially, the company is optimising externally via generative engine optimisation (GEO) and agentic engine optimisation (AEO), while building internal proprietary capabilities that fuse large language models with deep domain expertise and structured travel data.
Transaction and Fulfillment Moat
The most critical nuance of Trip.com’s strategy lies in its view of the transaction and fulfillment phases. As James Liang emphasised during the call, “Discovery only creates value when it converts into a completed booking.” By engineering a seamless closed loop that binds real-time inventory, payment systems, and instant confirmation with minimal friction, the platform protects its core economic engine.
Liang shared that AI-assisted orders “through TripGenie on Trip.com increased by approximately 400% year-over-year, and nearly 60% of TripGenie interactions are now booking related, spanning hotels, flights, and attractions. These trends demonstrate that travellers are increasingly relying on AI not only for inspiration, but also to make and fulfill travel decisions”.
Moreover, travel as a buy does not terminate at the booking confirmation. Disruptions, cancellations, rebooking requests, and on-the-ground support require real-world execution. Highlighting the necessity of keeping operational heft at the core, Liang noted that “we are using AI to improve service where it adds value while keeping human support and operational execution at the center.” By doing so, Trip.com is reinforcing a competitive barrier that pure-play software agents simply cannot replicate alone.
Economics
By avoiding the heavy financial drain of training large-scale models, Trip.com ensures its financial model remains resilient. Highlighting the long-term compounding effects of these targeted deployments, Chief Financial Officer Cindy Wang noted during the earnings call that “over the longer term period, our goal is to translate these investments into improvements across the group.”
As AI scales across the entire user journey, greater automation and hyper-personalisation are projected to streamline operational costs, while superior matching and targeting drive higher conversion rates and stronger returns on marketing spend.
