Waiting is a legitimate strategy. Plenty of companies that ignored the first wave of every technology cycle did perfectly well by adopting later, when the tools were cheaper and the mistakes had been made by someone else.
But waiting deliberately and waiting by default are different things, and only one of them is a strategy. If you are going to wait, it is worth knowing what the position costs.
Your staff have already started
This is the cost that is almost always underestimated. In most companies, some employees are already using AI tools for work — usually free consumer accounts, on personal devices, without telling anyone.
They are not being reckless. They have a deadline and a tool that helps. But it means company information is going into services under terms nobody in your company has read, with no processing agreement, no record of what was shared, and no ability to answer a customer who asks where their data went.
Deciding not to adopt AI does not prevent this. It only prevents you from governing it. In practice, "we haven't decided yet" is usually a decision to have an ungoverned version of exactly what you were avoiding.
The knowledge gap compounds
Companies that have run one or two AI projects have something more valuable than the projects themselves: they know how their own organisation reacts. They know which processes were messier than documented, where review actually happens, which people adopt new tools and which quietly do not.
That knowledge cannot be bought later. A company starting in two years will not start where today's adopters are — it will start where today's adopters started, while its competitors have moved on. The gap is not in tooling, which will be cheaper and better. It is in organisational experience, which accumulates only by doing.
The obligations arrive regardless
The EU AI Act applies to companies that deploy AI systems, not only to those that build them. Its obligations — including requirements around AI literacy among staff who use these systems — do not wait for you to have a strategy.
If your employees are using AI tools informally, you may already hold duties you have not assessed. Adopting deliberately at least lets you meet them deliberately.
What waiting genuinely does buy you
To be fair to the position: costs are falling, tools are improving, and the failed experiments of early adopters are informative. A company that waits and then moves decisively can skip a lot of expensive learning.
That is a real advantage — but only if the waiting is active. Watching what works, maintaining a shortlist of your own candidate processes, and setting a trigger for when you will act are all compatible with not building anything yet.
A minimum position
If you are not ready to adopt, three things are still worth doing now, and none of them require a budget.
Find out what your staff are actually using. Not to punish anyone — to know. Write a one-page rule on what may and may not be put into external AI tools, so people have an answer rather than a guess. And note the two or three processes you would automate first if you decided to move, so that the decision, when it comes, starts from something.
That is not an AI strategy. It is the thing that stops the absence of one from costing you quietly.
