When a factor of production becomes abundant, value migrates to the factor that remains scarce. It is one of the few economic laws without known exceptions. AI is making execution abundant — analysis, drafts, code, summaries, variants — at unprecedented speed. The factor that remains scarce is the capacity to decide well: what to ask for, what to discard, where to stop, whom to trust. We call that capacity decision capital, and the thesis of this article is that it has become the principal constraint on organisational growth.
The abundance that produces no results
The data describe a precise scene. Organisational AI adoption has reached 88% according to Stanford University’s AI Index (Stanford HAI, 2026). At the same time, more than 80% of organisations using generative AI report no tangible impact on enterprise-level results, according to McKinsey’s State of AI survey (McKinsey, 2025).
Nearly everyone has the tool; almost no one has the result. The lazy explanation is that more “technological maturity” is needed. The correct explanation is that the bottleneck has moved: it is no longer the capacity to produce, it is the capacity to choose. An organisation that generates a hundred options at the cost of one is not ten times faster — it is ten times more exposed to the quality of its own decisions.
AI multiplies options. It does not multiply judgement. The difference between the two is the bill organisations are starting to pay.
What decision capital is made of
Decision capital is not the talent of individuals: it is a property of the organisation. Four components define it.
Clarity of decision rights. Who decides what, with what mandate, within what thresholds. In organisations that leave this implicit, AI makes matters worse: it multiplies inputs and turns every meeting into a tribunal of options. The boundary of algorithmic delegation is the first decision right to put in writing.
Quality of the case file. A decision is worth the preparation that precedes it. AI can enrich the file — more scenarios, more counterarguments — or contaminate it with unverified plausibility. The difference is made by the method, not the tool.
Speed of learning. Organisations with high decision capital record their decisions, compare them with outcomes and correct the process. Those with low capital remember only the successes and rewrite the history of the failures.
Tolerance of dissent. The cheapest signal about the quality of a decision is someone contesting it with arguments. Where dissent costs careers, the uncomfortable information always arrives after the signature.
Why now
This is not a timeless reflection: it is an urgency with a date. The redesign of workflows — the factor McKinsey’s survey identifies as most correlated with economic impact — has been undertaken by only 21% of organisations. Those doing it now are deciding, for the next ten years, where the line runs between what the machine executes and what people choose.
Doing it without decision capital means inheriting the implicit decisions of vendors, defaults and individual departments. Doing it with decision capital means designing the organisation around the decisions that matter. It is the difference between enduring the post-automation era and directing it.
The acceleration that multiplies decisions
There is a dynamic that makes the subject urgent even for those who consider themselves behind on adoption. In the United Kingdom, a quarter of businesses reported using some form of AI technology in late December 2025 — up 15 percentage points since the question was first asked in September 2023 — and among businesses with 250 or more employees the proportion reaches 44%, with a further 15% of all businesses planning adoption within three months (ONS, 2026).
Every adoption brings with it a cluster of decisions that did not previously exist: which system, with what data, for which processes, with what limits, under whose responsibility. The paradox of automation is all here — automating execution increases the number and weight of the decisions to be taken. An organisation whose adoption grows on that curve without strengthening its decision-making capacity is widening the base of the pyramid while thinning its apex.
How it accumulates, how it disperses
Decision capital behaves like all capital: it accumulates slowly and disperses quickly.
It accumulates through documented practice. Organisations that write down their important decisions — the context, the alternatives considered, the reasons for the choice — build an archive that turns individual experience into collective property. No sophisticated tools are needed: what is needed is the discipline of writing before and verifying after.
It accumulates through outcome review. The most profitable meeting an executive committee can institute lasts one hour a quarter: three past decisions, expected against actual, a single question — what does this tell us about how we decide? Not about the merit of each choice: about the process.
It disperses through unguarded turnover, when those who saw the hard cases leave without anyone extracting what they learned. It disperses through ceremonial haste, when case files become attachments nobody reads and meetings ratify choices already made elsewhere. And it disperses — this is the new risk — through unwitting delegation, when systems absorb decisions one default at a time and the muscle of judgement, no longer exercised, atrophies exactly where it would be needed most.
Measure it, do not celebrate it
An asset that is not measured is not managed. Three practical indicators, within reach of any leadership team: the cycle time of critical decisions (from question to choice, not from choice to execution); the real review rate on decisions delegated to systems — if it is zero, that is not trust, it is abandonment; traceability: how many decisions above a given threshold have a case file reconstructable six months later.
One clarification, because the misunderstanding is common: high decision capital does not mean deciding slowly. It means the opposite — real speed comes from clarity. Organisations that know who decides what, with what minimum preparation and what accepted risk thresholds, decide in days what elsewhere takes months of defensive meetings. Pathological slowness is not born of excess rigour: it is born of ambiguity, where everyone protects themselves and nobody signs. Rigour, once its set-up cost is paid, is an accelerator.
None of these indicators requires new technology. They require what has always been scarce: leadership discipline. It is the ground on which our institute’s method works — because the right lens is not “how to adopt AI”, but “how to decide amid the abundance AI produces”.
What this means for decision-makers
Three operational implications. First: treat decision capital as you treat financial capital — with a balance sheet, an owner and a periodic review. Second: invest in the case file before the tools; every pound spent on judgement capacity pays out across all future decisions, every pound spent on technology alone pays out only if the judgement already exists. Third: protect structured dissent — it is the cheapest quality control there is.
There is, finally, good news for those starting now: decision capital does not require scale. A mid-sized company with clear decision rights and honest case files decides better than a giant where choices dissolve into nameless committees. It is one of the few terrains where size is no automatic advantage — and where the discipline of those who lead is worth more than the budget of those who buy.
The post-automation era will not reward those who execute fastest. Fast execution will belong to everyone. It will reward those who choose best — and choosing well, unlike execution, cannot be bought by the unit.