Your Guest Data Is the Moat. The AI Model Was Never Going to Be.

As AI models become commodities, they stop being a source of competitive advantage — a Hospitality Net opinion piece argues the real differentiator for hospitality companies is the ability to unify and act on their own guest data. Its central line: “The AI model itself is not your moat. The capability to turn your own guest interactions into connected, trusted and continuously improving intelligence is.” Most hospitality businesses still keep guest information fragmented across PMS, CRM, and loyalty platforms — fragmentation that blocks the coherent, trustworthy guest picture better AI decisions actually require. The piece proposes treating data as a strategic product built on four capabilities: connecting data across systems, reliably identifying individual guests across touchpoints, activating intelligence at the point of decision, and continuously learning from each interaction — together forming what the author calls a “learning flywheel” that improves with every guest interaction and can’t simply be bought by licensing the same AI model a competitor uses.

That thesis echoes an argument Terence Ronson made a month earlier about where value actually accrues in the hotel technology stack. Mapping a four-layer architecture — system of record, application/semantic layer, frontier model, and compute — Ronson concluded that durable value sits in the application layer wrapped around the PMS, not the underlying model, warning that “whoever owns the semantic layer effectively owns the guest” as AI lock-in migrates one layer up from traditional PMS lock-in.

The stakes get concrete once agentic booking enters the picture. Markus Busch, editor of hospitality.today, has argued loyalty databases are the one asset AI agents can’t easily replicate: Marriott Bonvoy alone counted 283 million members in 2025 (43 million added that year), with Hilton Honors above 200 million and World of Hyatt at 63 million and growing 19% year over year. Those numbers are exactly the kind of proprietary, connected guest intelligence the “data is your product” argument is about — and Busch’s warning that overexposing that data to external agents could let those agents absorb it and hollow out the moat is the flip side of the same coin.

For hotel technology leaders, the strategic question shifts from which model to license to whether the data underneath it is worth licensing anything against at all.