Every AI investment case shows you the cost of acting: licences, training, setup, maintenance. Neatly listed, down to the franc. Almost none shows you the cost of waiting. Yet doing nothing is not a neutral position. It is a decision - one with a price tag that nobody prints, because no invoice arrives for it.
This is not a rhetorical trick but a calculation you can set up just as rigorously as the cost side. It simply has two parts, both of which tend to get overlooked. Let us look at them.
Part one: the money you do not save
The first part is simple and uncomfortable at once. If your company could save a given sum per year with AI - say CHF 240,000 through automated document processing - then every month without scaling is a month in which around CHF 20,000 of that saving expires. Not deferred. Expired. It does not arrive later, because the time in which it would have arisen is gone.
After six months of hesitation you are not short "a bit of efficiency" but a concrete CHF 120,000 you will never recover. After twelve months, a quarter of a million. The insidious part: that number appears nowhere. It shows up in no balance sheet, nobody books it, and precisely for that reason it does not feel real - even though it is. A licence invoice for CHF 20,000 would trigger discussion immediately. A forgone saving of the same size gets discussed by nobody, because it stays invisible.
Part two: the head start of the others
The second half is the less pleasant one. While you wait, your competitors do not necessarily wait with you. And whoever starts earlier builds something that cannot simply be bought later: experience. Established processes, teams that know which prompt works and which does not, an organisation that has learned to handle the tools.
This productivity lead grows cumulatively, not linearly. After roughly six months we are talking about a single-digit percentage lead. After twelve months, about 15 percent. After 24 months, a third and more. Which means: the cost of waiting is not only your own forgone saving. It is the forgone saving plus the lead your competition builds in the same period - the one that later costs you market share, margin or simply speed.
A worked example
Stay with the company at CHF 240,000 in possible annual savings. A competitor starts today, you wait twelve months. By the end of that year you have left CHF 240,000 in savings on the table - that is the visible part, once you actually work it out.
Now the invisible part: over those twelve months the competitor has not only saved the same sum but built roughly 15 percent of productivity lead. If you play in the same league, that means: they can offer the same service 15 percent cheaper, or deliver 15 percent more in the same time, or deploy their people on higher-value work while yours are still stuck in the manual process. That gap does not disappear when you start a year later - you begin at zero while they keep building on their lead. This is why the competitive gap behaves differently from a simple lag: it is compound interest, not a fixed amount.
Why this calculation is almost never made
Waiting costs have a perception problem, and it is cleanly explicable in psychological terms. Visible spending hurts, invisible spending does not. An investment of CHF 50,000 in an AI project sits in the budget as a line item, everyone sees it, everyone can question it. A forgone saving of CHF 200,000 sits nowhere - it is the absence of a gain, not the presence of a loss.
People, and therefore companies, weight a visible possible loss more heavily than an invisible forgone gain. That is human, but expensive in business terms. It leads to decisions that come out systematically too cautious - not because the cautious option is better, but because its price stays hidden. "We will wait until the technology matures" sounds like commercial prudence. In truth it is a bet with a real stake - and the stake is placed against your own company.
"But acting rashly is expensive too"
The legitimate objection: does this mean you should charge blindly into every AI project? No. Rash action carries its own costs - badly chosen tools, missing preparation, change resistance. But that is not the counterargument it is taken for. It is an argument for clean selection and preparation, not for unlimited waiting.
The point is not "fast at any price" but "the cost of waiting belongs on the table". A company that invests three months in a good selection makes a different decision from one that waits three years because the technology never seems mature enough. The first case is diligence, the second is deferral with a justification. You can only tell them apart if you know the costs of both sides.
What you do with this
You do not have to act rashly, and you do not have to believe every number to the last rappen. You only have to know both sides of the calculation before you decide. The cost of acting on one side - you know that anyway, it sits in every proposal. The cost of waiting on the other: forgone saving plus the competitive lead of the others. Only once both columns are on the table do you make an honest decision instead of a gut call dressed up as caution.
That is exactly what the opportunity-cost view of our AI scaling simulator is built for. It works out the invisible side: what does every month of waiting cost you, cumulated over time, and how large does the gap grow to a competitor who starts today? You see it for your industry and your headcount, in francs, in black and white - so the expensive half of the decision stops being the invisible one.
→ Work out your own cost of waiting
