The correction windowThe cost of delay that no one budgets for

Fifty-three years later, the window is still open. And the tool expected to close it may well be widening it.

One model, one mechanism, one variable

The Limits to Growth, published in 1972 by the Club of Rome, rests on a simulation model called World3. Five global variables (population, industrial output, food, resources, pollution) interact over two centuries. The report became one of the most cited and most misunderstood texts of the twentieth century, because most people remember a conclusion it does not contain: “the world is going to collapse”.

What the model shows is more precise and more useful. It measures what happens when a signal arrives and the correction is slow to follow. The central scenario does not describe a shortage of resources. It describes a delay in decision. The cascade runs as follows: lower industrial output leads to lower investment, which leads to less food per head, which leads to a decline in well-being. Each variable pulls the next one down.

The variable that separates the scenarios is not the quantity of resources. It is the speed of the response.

Fifty years of data, and the dynamic holds

In 2021, Gaya Herrington, then an analyst at KPMG, published an empirical update of the model in the Journal of Industrial Ecology. She set the real data collected since 1972 against the twelve original scenarios. The data most closely follow two trajectories: BAU2 (resources more abundant than expected, slow response) and CT, Comprehensive Technology (technology widely deployed, decline avoided, growth capped).

In 2024, Nebel and his co-authors fully recalibrated the World3 model using real parameters. The main effect is to raise the peaks of most variables and to move them a few years further into the future. The dynamic of the model holds. The dates move. The mechanism does not.

For a chief executive, this is operational information: the trajectory remains open, and the controlling variable is the delay between signal and decision.

AI and the promise not kept

The CT scenario assumes technological progress of 4% a year. The rate actually observed is closer to 1.5%. The gap between those two figures is the correction window.

The most appealing hypothesis is a simple one: if artificial intelligence accelerates technological progress, it could bring the actual rate closer to the threshold that the CT scenario requires. In 2023, McKinsey estimated that AI could add up to 3.4 points of productivity a year by 2040. Goldman Sachs projected an increase of 1.5 points a year over a decade.

Actual measurements tell a different story. In March 2026, Goldman Sachs published an analysis that finds no significant correlation between AI adoption and productivity at the macroeconomic level, while noting gains of 30% on specific tasks. In 2024, Daron Acemoglu of MIT estimated that the total factor productivity gains attributable to AI will not exceed 0.53 to 0.66% in total over ten years.

The promise stands at 3.4 points a year. The measurement stands at 0.66% in total over ten years. The gap between the two is the correction window itself.

The gap does not rule out the potential of AI. It does, however, rule out assuming it.

The energy paradox

Data centres dedicated to artificial intelligence will more than triple their electricity consumption between 2025 and 2030, according to the International Energy Agency. Total data centre consumption doubles over the same period, from 485 to 950 terawatt hours. These centres connect to the same electricity grids as factories.

According to the Lawrence Berkeley National Laboratory, the median time between a request for grid connection and commissioning has risen from under two years to almost five years over fifteen years. Regional markets are taking on data centre demand faster than they are building the corresponding capacity.

Two feedback loops stand opposite each other. The first is virtuous: AI speeds up the detection of signals, shortens decision times and pushes technological progress towards the required threshold. The second is destructive: AI consumes a growing quantity of electricity, competes with industry for access to the grid, and contributes to the slowdown in industrial output that sets off the Meadows cascade.

The 2025-2035 window is the decade in which one of the two loops will prevail.

The signal that no one read for ten years

The dynamic Meadows described is not confined to planetary models. Every organisation has its own curve.

La Redoute was France’s leading mail-order retailer. The signal from online retail was legible from the early 2000s. The correction was slow in coming. In June 2014, Kering sold the business for a symbolic one euro, having committed around €315 million to a transformation plan and €180 million to measures for employees, together with a plan to cut 1,178 posts out of 3,437 staff. The signal had been public for ten years. It was the correction window that had closed.

The question for a chief executive is not whether Meadows was right. It is how long a signal takes to become a decision within one’s own organisation.

What this changes in practice

I run an organisation augmented by artificial intelligence myself. Analyses that used to take days are produced in a few minutes. Data flows are processed continuously without a permanent team. The shortening of lead times is real, measurable and daily.

That is where the paradox becomes invisible. The only resource I manage consciously is informational: tokens, context memory, speed of production. The electricity that powers those tokens lies entirely outside my field of decision. The individual experience is one of liberation. But when each user perceives only the acceleration, without sensing the energy weight of their own requests, no one feels the signal the model measures. The disconnection between individual experience and systemic consequence is precisely the kind of delay Meadows described.

The variable has not changed

It is not the resource that runs short first. It is not the technology that fails. It is the time between the moment one knows and the moment one acts.

Signals that arrive and that no one reads. Decisions that are deferred because the cost of delay does not appear on this month’s bill. Processes that accumulate because removing them takes a kind of courage the quarterly calendar does not reward.

This reading can be refuted: if measured productivity gains reach the required rate before the electricity constraint begins to bite, the correction window will close of its own accord and this analysis will have been wrong. It is measurable. We shall know within ten years.

The curve does not punish growth. It punishes delay.

The correction window is a concept put forward by Peter Guillon to name the gap between the signal received and the decision taken. It applies to the trajectories of organisations as well as to macroeconomic trajectories. See the definition in the glossary.