This will undoubtedly be the flavour of the month across the internet, particularly on platforms such as LinkedIn. People will suddenly be saying, “Aha! AI isn’t saving the world. CEOs are seeing the cost of all that token usage and realising that they shouldn’t have replaced people.”
Yes, some CEOs will be receiving unexpectedly large bills, but there is nothing new about the underlying problem. It is the same challenge every corporation faces when assessing value for money and trying to understand the total cost of a decision before committing to it.
My favourite recurring example is the comparison between contractors and permanent employees. Taken as a whole, a good contractor does not cost more than a permanent employee, provided you pay a sane market rate and you get what you asked for delivery-wise. Permanent employees come with many additional and often less visible costs, including benefits, training, leave and potential employment liabilities.
This does not mean that one option is automatically better than the other. It simply means that avoiding consultants or contractors does not always produce the savings an organisation expects. A good consultant who delivers genuine value is as good a deal as a dedicated employee. {This is a shameless plug}.
The same principle has appeared throughout the history of business outsourcing. It may look inexpensive to outsource work to the cheapest available supplier in another country, but is it really cheaper if the result is increasing technical debt and creating remediation costs further down the line?
Going back even further, buying cheap tools may appear to be a cost-saving exercise. However, if they need to be replaced repeatedly, they may ultimately cost far more than buying the right tools in the first place.
AI is no different. Every generation falls for some version of the same promise. When something appears too good to be true, it usually is. Everything has a cost, and there is no such thing as a free lunch.
What we need to avoid now is another knee jerk reaction. AI is brilliant and is a major step up in what corporations do day to day. However, it is not a universal solution, and its costs go beyond simply doing the work and providing the necessary computing power.
Vast amounts of investment have been poured into AI companies, and those investors will eventually expect a return. One way or another, that is likely to mean that prices will rise and costs will increase.
So, we return to the same advice that finance departments have been giving organisations since time immemorial: look at the complete cost of a decision.
Do not ask only how much something will cost this year. Ask what it could cost over the next ten years. Consider the supporting systems, the people, the remediation work, the supplier risks and the potential cost of changing direction later.
Before making large-scale changes, take the time to understand what you are buying, what it will replace and how much the service is likely to cost over its entire lifetime.