Booking Holdings keeps AI momentum
Analysts still see Booking Holdings as doubling down on AI for search, recommendation and booking flows—an investment stance that could pressure peers on margin and feature velocity argued. For multi-brand rivals, it’s a reminder to pair product experiments with platform-level guardrails for safe, repeatable rollouts.
Booking announced a transformation program targeting up to $450 million in annual savings by 2027 — with roughly $150 million expected in 2025 — and said part of those savings will be reinvested into AI automation and product development. (cfodive.com) CEO Glenn Fogel described agentic generative AI as a “transformative force” on the Feb. 20, 2025 earnings call and framed GenAI as a way to “bring together different elements of travel into a connected platform.” (phocuswire.com) Booking has run public test programs with OpenAI and signed on as an early partner for the company’s Operator/ChatGPT app efforts, positioning its marketplace to be callable by third‑party agents and in‑chat apps. (openai.com) Analysts are reacting: Morgan Stanley upgraded Booking to Overweight on Feb. 24, 2026 with a $5,500 price target, explicitly citing the company’s scale and AI positioning as reasons the market’s AI fears were premature. (investing.com) Booking reported its merchant booking platform reached about 59% of transactions in the company’s Q4 2024 disclosure, while later quarters showed gross bookings rising 14% year‑over‑year to $49.7 billion — figures executives tie to product and operational changes, including AI features. (fool.com) Executives told investors they are “tokenizing” traveler data for LLM consumption and expect customer‑service automation alone to contribute roughly $150 million of the near‑term savings, signaling emphasis on platform data pipelines for model inputs. (investing.com) Providers and platform teams are already stressing scalable guardrails, centralized policy layers, and observability for agentic flows — guidance echoed across OpenAI’s Agent safety docs and industry playbooks that recommend gateways, evals, and telemetry to avoid prompt‑injection, data leakage and rollout drift. (guardrails.openai.com)
