AI tilts capital toward what can be measured. The long bet loses on the business case.

Jesper E. Thomsen · 16 October 2026

AI makes measurable improvements easier to find and to document. In a funding contest, the ten-year decision therefore meets a stronger opponent than before. Nothing in the model has to prefer the short term; the bias comes from what can be shown. The counterweight is a different kind of scrutiny, run on a slower clock.

McKinsey has described how an AI operating system could change the way a CEO runs a company. Information no longer travels through layers. Decisions are made continuously rather than at the quarterly review. Executives spend less time finding out what is happening and more time deciding. Most of that seems right to me, and most of it is about the present. What happens when an organisation gets very good at optimising the present while the future stays as uncertain as ever? My answer is that capital tilts toward what can be measured. The long bet loses on the business case unless it is judged on a clock of its own.

Two clocks

An organisation makes two kinds of decisions. One kind can be checked against information that already exists, e.g. margin, utilisation, working capital and cost to serve. AI can move these onto a continuous clock. Opportunities are found, modelled and executed as they appear, each with a case attached.

The other kind rests on assumptions that take years to validate, e.g. which capabilities to build now, which markets will matter and which technology to commit to before its value is visible. These need a clock on which the assumptions can be examined without reopening the allocation every time a short-term alternative turns up. In most companies that is the strategy round.

The company then runs two cadences for capital, a fast clock producing a stream of documented proposals and a slow clock producing a few judgment calls.

The faster and more precise the quarterly numbers become, the more attention and capital they attract. Nobody decides to favour the short term. It happens because that is where the evidence is. And much of what this quarter shows is still the consequence of decisions taken years ago, which no amount of precision in the numbers reveals.

The burden of proof moves

The early AI cases I see show value on cost before growth. A procurement team that re-tenders two hundred contracts a year instead of forty shows the saving in the first quarter. An account team using the same tools to open a new customer segment has nothing to show for two years. Released capacity is quickly absorbed into more activity, so the visible gains come from stopping work and removing cost. That is the lesson the organisation learns first.

The same logic influences allocation. An agent for invoice matching will likely beat building the data platform that would let the company sell monitoring to its installed base in five years. Boards do fund the latter, but it will likely need stronger sponsorship than before.

The fast clock also does not reliably use its own history. A finance partner who has taken cost out of customer service three times will say so when the fourth proposal arrives. An agent optimising against a cost target may find the fourth cut and present it as cleanly as the first, even with the previous three in its context. The remedy is to require a proposal to show the cumulative cut to date, so the fourth cannot be written without the first three.

The slow-clock review

The CEO still answers for this year and for the business that has to compete in ten. This year's case now arrives with better arguments every week. The ten-year case needs scrutiny of its own kind to stay in the contest.

That scrutiny runs on the slow clock, where the long decisions are made. It asks whether the thesis still holds, rather than whether the bet has paid back, which it cannot have yet.

A thesis review has three possible outcomes. The assumptions hold and execution is slower or dearer than planned. The environment has moved and the thesis needs rewriting. Or the original reasoning was wrong and continuing compounds it. A moved environment or wrong reasoning are reasons to stop or change course. Slow execution is a reason to ask whether the route is still right. A capability can stay necessary while building it in-house stops making sense and buying or partnering takes over. Slow execution is not a reason to abandon the objective, although that is how the fast clock would read it.

A board reading an AI-enabled plan could ask where the decision that cannot be measured sits, who signed for it, and when its thesis is next reviewed. That requires the original reasoning to exist somewhere retrievable. That is the subject of a separate piece.

When short-term focus should take priority

Where a business is in distress, the measurable is survival and the tilt is correct. Where the industry's own clock is short, there may be no ten-year bet worth protecting. And if AI makes scenario work and capability valuation good enough that a thesis arrives with something resembling a model, the contest gets fairer on its own. I do not expect that soon. What is missing is information about the future. Better analysis does not supply it.