Every 2027 budget has a big unknown (on top of the usual Government shenanigans)
Every multi-site operator knows what their booking platform costs per site per month. Same for CRM, POS, feedback, vouchering and WiFi. Those numbers were negotiated and signed off in the last budget, and have an owner against them. Read on to see why the cost of AI in hospitality does not fit the budget.
What will AI cost? No one really knows. I have asked operators of every size, from enterprise to independent, and the answer every time is the same: there is no defined budget and no tracking of it.
Almost every 2026 budget in hospitality was set before AI was being used in anger, and the tools have arrived faster than the planning cycle. The spend is happening, and growing, hidden within other cost lines.
What is ours? AI costs Guestwise £1,000 per employee per month as of August 2026, which hurts!
We sit at the top end of the range, as it’s our core business, so a restaurant group will not expect anything like that number (I hope). That figure covers the AI tools for design, coding, bug fixing and everything our team uses to do their jobs. It excludes the cost of running AI inside the products we sell, which is cost of goods, but that line is also ramping faster than we expected.
AI is becoming too big not to have clarity on the likely cost and expected results in the next budget.
Why it does not fit the budget
Hospitality historically budgets for technology in licences, a fixed price per site per month, on each SaaS contract. A linear number, including new site openings.
AI is mostly priced on usage. The bill goes up when a team uses it more, and the team uses it more as they get better at it. The cost rises with competence, which is the reverse of how every other technology line in the business behaves.
A budget built on per-site licences, allocated to the relevant department, has nowhere sensible to put a cost that spans departments, grows with every new user, and has almost unlimited use cases. So it gets absorbed. It lands on central IT costs, expenses, a personal development line, or quietly inside the marketing budget that was meant to buy media.
What is the total? How to judge ROI? Is AI use a central cost, or for each department to forecast and get budget sign off for?
The pattern is consistent. Marketing buys an AI content tool, ops builds or buys forecasting for stock or rotas. Someone in finance is paying for an AI benchmarking subscription on a personal card. Each purchase is sensible. Added up, they are a line item that will get very big before any of us realise it.
How to forecast the cost of AI
It doesn’t help that nobody knows what AI will cost. Ramp’s AI index, published in June for the American market, puts the median company spending at $11.38 per employee per month. The top 10% spend around $611. The top 1% spend $7,500. PER EMPLOYEE! A 680x spread says that the market has no settled view of the right number, and the answer on ROI becomes even more murky. AI is ‘fast’, but are the outcomes better? Did each employee achieve 2x more, or did they just add a tool, and the bottom line remained the same?
For UK hospitality there is no published figure at all that I can find. That is part of why we have shared the Guestwise numbers, so that we can start to create an industry benchmark.
Operators compare notes on what they pay for everything else in the tech stack, and lots of suppliers publish prices on their website. AI isn’t like that, but sharing numbers in this sector would help everyone make better decisions for 2027.
Whose budget line should it be
It would be easy to be a negative Nancy and say ‘it’s going to be a problem’, so I wanted to have a stab at possible solutions.
AI in aggregate needs a rolled up number, with a named owner who reports the total. It also needs to be called out in each department’s budget.
Other items that serve the whole business (Teams, WiFi, laptops) are typically centralised. But they are a known cost. Every pound of AI fees has to come from somewhere, and needs to be a deliberate decision. The free versions of AI are ‘fine’, but they lack the ability to act as one organisation, make the best decisions, or the protections around data and training and shared knowledge and skills that anyone can use.
I would put it on the tech budget with a per head allowance and a named owner. Marketing can then argue for AI spend on its merits, the way it argues for anything else, and the board can see what the business actually spends overall on AI and to what purpose.
That means changes to the budget, but arguably it is less awkward than trying to hunt around in March 2027 for what the total was and who spent it and why. With the proliferation of models (each of which comes at very different costs), where is AI spending delivering great value, and where is it simply being used with no discernible outcome? That is before the cost of bad decisions is taken into account, which is also very opaque at the moment. As an industry we run the risk of doing things much faster, but not achieving better results for that spend.
Guestwise spends a lot on AI and will keep doing so, but it would be great to have operators share their costs and expected results from using AI in hospitality.
For anyone reviewing this year’s budget or building next year’s, would AI get a cost line of its own? What metrics are being used to evaluate impact? And whose name or names are against it?
AI use is on a huge growth trajectory. Before the costs get out of hand, clarity on how it is being used, how much it costs, and who owns that decision is worth getting right.
