Hotels Can't Cut Their Way to Better Margins Anymore

Hotel margins are shrinking, and all eyes in the space are on that gap in 2026. In 2024, CBRE reported expenses through gross operating profit rose by 4.1%, compared to a 2.3% rise in total revenues. 2025 not only saw a 0.3% drop in RevPAR, it saw the United States' first ever RevPAR decline in a non-recession year, according to Costar. Occupancy dropped during that time, and while the 2026 World Cup is expected to spike reservations, experts are guiding a tempered excitement on that front. Instability around the world and within the US are driving an increased hesitancy to travel and increased costs along the supply chain.

For investors and managers, this timing couldn't be worse. A refinancing wave has been building in the difficult rate environment of the last few years. If an asset is underperforming at 7% interest rates, that is a much higher pressure environment than that same asset at 4%. Every basis point matters, and a 100 basis point improvement can be the difference between an asset finding a successful exit or falling deeper into a debt-servicing hole.

The pressure to improve NOI is real and immediate, and the question becomes where can that improvement come from?

The Staffing Ceiling

Labor is the first target for cost reduction or line item scrutiny, a reasonable decision when its typically the largest expense on the P&L. Six years on from COVID, however, the industry has already made and maintained deep cuts. Few properties have returned to pre-pandemic staffing levels, leaving most operators hitting a floor on labor reduction. Further reductions degrade the guest experience, and at mid-to-luxury service levels, the guest experience is the product.

Reinforcing this floor is the continued turnover problem facing hospitality. Quit rates remain sky-high, reaching 4% of staff leaving in a single month. Every departure triggers a domino effect of spending: recruiting, training, and the inevitable mistakes made by new hires as part of the learning process. Experienced team members can be multiples more effective than a new hire at the same task. Removing their payroll costs more than it saves once you factor in service recovery, lower reviews, and the reduced guest retention rate.

What this means for ownership and operators alike is: labor cuts and headcount reductions have largely run out of utility. The next lever has to be in making the existing operations more effective.

Where AI Fits, and Where It Falters

AI has begun to enter the hospitality industry, but adoption and use cases have both been narrow. Implementations have been predominantly chat bots and voice systems designed to automate routing and deflect routine questions. These tools do have a labor impact, but they largely don't move the needle substantively enough to impact the most important metrics that drive profitability over time: guest satisfaction, GOP, and ADR.

The bigger opportunity for portfolios is AI that actually improves the underlying operations and guest experience of the property. Catching service failures before they lead to recovery costs, identifying underlying patterns that erode ADR, enhancing the staff that creates the experience and not replacing them.

Truly impactful, hospitality native AI has to meet a higher bar than most vendors are able to deliver today.

What Hospitality AI Actually Needs

It has to be reliable. Baked into their functionality is the fact that LLMs can produce different outputs from the same input. For a 24-hour operation where a mistake at 2AM can become a brand-damaging argument by 11AM, inconsistency is a liability. Any system plugged into the operational workflows needs to be dependable enough for the property team to trust it without constant oversight.

It has to be secure. Hotels hold personal and financial data at scale, and the industry has learned the cost of mishandling it the hard way. That sensitivity is part of why hospitality was slow to adopt cloud and mobile, a transition many properties are still working through. Breaches cost both the property and the brand by spawning fines, litigation, and lasting damage to brand trust that directly suppresses rates. AI introduces new surface area for data exposure, and any system that touches guest data needs to treat security as the foundation, not as an afterthought.

It has to be cost-effective at scale. AI processing costs scale with usage. For a single property or a simple task, the economics look manageable. For a portfolio handling millions of reservations and billions of interaction data points across hundreds of assets, the costs spiral out of control. The ongoing tech industry debate about whether token costs will rise as subsidies from major AI providers fade makes this both critical and timely. Any AI deployment needs to be architected for cost control from the start, and has to prove its value against that cost.

Built by People who Understand Hospitality

There's one more factor that matters as much if not more than the technical requirements: the people building the AI need to understand the business. Making a hotel more effective isn't purely a technology problem. It's knowing that resolving a guest issue tactfully saves not just the cost of the comp, it protects the ADR and rate health of the property. It's knowing that when a sudden arrival wave hits and 30 rooms still need to be cleaned, the most expensive labor hours of the property are spent having the GM or Director of Rooms jump in and turn rooms.

That is why I started Caleta. We're building AI for hospitality. By identifying underlying issues and risks impacting the guest experience and property operations, we reduce overtime and service costs, protect the rate setting power of your assets, and directly drive improvements in your NOI. We're doing it with a team who has been in the industry, and we're building for reliability, data security, and cost-effective scalability from day 0.