In capital-intensive sectors, failure almost never comes from technology. It comes from the silent drift between what the capital was committed for, what the board actually decides, and what gets built.
The starting condition: deciding in the dark, with no way back
Industrial capital has two properties financial capital does not. The first is lead time: commitment lands years before revenue, and in between, rates, technology and rules all move. The second is irreversibility: a plant built in the wrong place does not move; an EPC contract signed on the wrong risk structure is renegotiated only by paying.
From this follows something few organisations address explicitly: the quality of a programme is not decided at appraisal but in the chain of intermediate decisions - taken by different people, at different moments, each reasonable in itself. That is where the systems diverge, and where no one looks, because no single decision looks large enough to warrant it.
The three systems, and why alone they explain nothing
Capital intent is not the plan: it is what return, at what risk, over what horizon. It can shift without anyone declaring it - a fund approaching end of life, a rate moving two hundred basis points - while the programme keeps chasing the old one.
Governance is not the board or its committees: it is who decides what, on what information, at what speed. Most industrial governance is designed to report, not to decide.
Execution optimises on metrics that are correct and that do not contain the investment thesis. A programme director measured on schedule has no instrument to notice that the variation saving two months destroys residual value at fifteen years.
Taken separately, all three usually work well. The fault is never inside a pillar: it is always between two.
The three interfaces
I. Capital ↔ Governance - does the board decide, or ratify? It degrades by accumulation: each ratification is reasonable, but after twelve months the board no longer holds its own model of the programme. When the decision requiring independent judgement arrives, the capacity is gone - not removed, atrophied. The question: what is the last capital decision in which this board changed the outcome?
II. Governance ↔ Execution - do decisions arrive intact? Every hand-off translates the decision toward what the next level already knows how to do. The revealing signal is not disagreement, which is healthy and visible, but selective slowness: what is neither implemented nor contested maps exactly where the interface broke. The question: of the decisions taken in the last two quarters, how many are operative today?
III. Execution ↔ Capital - does it still serve the original thesis? The most dangerous of the three, because it produces no negative signal: cost, schedule and quality stay on track while an accumulation of locally correct choices changes the nature of the asset. It surfaces at commissioning or at sale, when the only remaining lever is price. The question: how far has expected return moved since approval, and through which three causes?
Where I have seen this: a rollout, repeated fifteen times
I represent the investment sponsor in a European industrial programme of roughly one billion euro, backed by a Luxembourg-based infrastructure fund, for the rollout of fifteen circular-economy and energy-transition plants. The mandate places me at what the documents call the investor interface: governance, reporting, execution alignment and risk control, in EPC environments and across multiple stakeholders.
That job title is itself the argument. The role exists because the three systems do not govern the space between them: someone has to, and where no one holds that mandate the space is uncovered by definition.
A rollout has a property a single asset does not: the same decision chain repeats. Fifteen times you negotiate the same contractual structure, approve the same class of variation, translate the same board decision into the same site language. This is where drift compounds. A two per cent concession on risk transfer, replicated fifteen times, is no longer a concession: it is the programme's risk profile.
It is also - and this is the useful part - where drift is most correctable. A single asset gives you one chance to set the interface right. A rollout gives you fifteen, provided someone reads the pattern running through them rather than managing each in isolation. The repetition that multiplies the error is the same thing that makes it visible in time.
How to realign, and in what order
The order matters: acting on execution while the governance interface is broken produces only noise, because corrections never arrive. First, write on one page what the capital expects today, not at approval. Second, give the board one real decision back, with alternatives that have not been pre-discarded. Third, give every decision an owner, a date and a report on implementation, not progress. Fourth, require every variation above a threshold to state its effect on expected return.
None of the four requires a reorganisation. They require someone to hold the mandate to watch the spaces between functions - precisely what no org chart assigns.
Three errors that make any method useless
Treating alignment as an event. The three systems diverge by nature, not by defect: the question is not «are we aligned?» but «how often do we verify it?».
Mistaking information for control. A board receiving two hundred pages is less able to decide than one receiving ten and knowing which three assumptions to watch.
Looking for a culprit instead of an interface. The three drifts have no responsible party: they have a location. Replacing people without repairing the interface reproduces the same outcome under different names.
Why this matters now
Capital seeking real assets - energy transition, infrastructure, circular economy, industrial capacity, and now compute infrastructure - is more abundant and more impatient than it has been for a generation. Horizons have shortened while projects have stayed long. In that condition the advantage does not lie in selecting projects better, because everyone selects better: it lies in recognising drift earlier, when correcting it costs a meeting rather than a dispute.