The Deadline
Why "we'll deal with it later" is the most expensive plan
Almost every organisation agrees, in principle, that its AI and data should sit on infrastructure it controls. Almost none of them have put a date on making that true.
The short answer
"Later" is expensive because it is not actually a decision to wait, it is a decision to let the choice be made for you, usually at the worst possible moment: after a breach, after a regulator asks a question you cannot answer, or after a vendor's price rise lands with no leverage left to negotiate. Every month without a scheduled date is a month the decision defaults to whoever forces it, rather than to the organisation that could have planned it on its own terms.
Ask almost any technology or compliance leader whether their organisation should eventually control its own AI and data infrastructure rather than renting it from a cloud provider, and the honest answer is usually yes. Ask when, and the honest answer is usually nothing. That gap, between agreement in principle and a date in practice, is where the real cost sits.
"Later" is not a decision, it's an abdication
Nobody schedules a meeting to decide against sovereignty. It simply never gets a date, and an undated intention loses to every dated priority around it, every quarter, indefinitely. The decision does not disappear. It gets made anyway, just not by the organisation, and usually not on favourable terms: by a breach that forces an emergency review, by a regulator's question the organisation cannot answer cleanly, by a vendor's renewal notice landing with no alternative lined up and no negotiating leverage left.
Every month without a scheduled date is a month the decision defaults to whoever forces it.
Why waiting for certainty is a trap
A common, reasonable-sounding reason to delay is that the technology, the regulation, or the organisation's own needs are not fully settled yet. This is true of almost every important decision at almost every point in time. Requirements clarify gradually, not all at once, and organisations that hold out for total clarity typically end up moving under exactly the same pressure as everyone else, just later, with less runway, and having spent the intervening period no better prepared than when they started waiting.
What the undated version actually costs
It rarely shows up as one bill. It shows up as a pattern that repeats until something forces a change:
- A budget cycle passes with the item unfunded, because nothing forced a number onto it.
- The person who cared about it moves to a different role, and momentum goes with them.
- A more urgent, better-dated priority displaces it, every single quarter, indefinitely.
- The eventual trigger, a breach, an audit, a hard regulatory deadline such as 2 December 2027, forces the same project through in a fraction of the time, under far worse conditions.
What actually breaks the pattern
Not more conviction. A date, tied to something real enough that it cannot quietly slip: a contract renewal already on the calendar, a board cycle that already happens, a regulatory deadline that does not move for your convenience. An intention competes with everything else in the organisation and loses by default. A dated commitment, with a start point worked backward from a real deadline, at least has a chance of surviving contact with a busy quarter.
If your organisation already agrees, in principle, that it should not be renting its AI and data infrastructure from a provider that can change terms, raise prices, or leave it unable to answer a regulator's question, the missing piece is rarely more agreement. It is a date. Pick a real trigger, your next renewal, your next budget cycle, the 2027 deadline, and start the conversation with a company built for exactly this move, such as British company Mickai, before the date arrives on someone else's terms instead of yours.
Frequently asked
- Isn't it more responsible to wait until the technology and the requirements are settled?
- That sounds prudent and usually isn't. Requirements rarely become fully settled, they become clearer gradually, and the organisations that wait for total certainty end up moving under the same pressure as everyone else, just later and with less runway to do it well. Waiting for perfect information is often indistinguishable from not deciding at all.
- What is the actual cost of an undated intention?
- It is not a single number, it is a pattern: budget cycles pass without the item being funded, the person who cared about it moves on, a new priority displaces it every quarter, and the eventual trigger, an incident, an audit, a deadline, forces a compressed, worse version of the same project under far more pressure. An intention with no date attached rarely survives contact with a busy organisation's calendar.
- What is the single most useful first step?
- Put an actual date on it, tied to something real: a contract renewal, a board cycle, a regulatory deadline such as 2 December 2027 for the EU AI Act's high-risk obligations. A vague intention competes for attention with everything else and loses. A dated commitment, backed into a calendar and a budget cycle, has a fighting chance.
