There’s a moment I keep seeing in boardrooms this year. Someone slides a piece of research across the table, the room leans in, and within about ninety seconds the conversation lands on the same question it always lands on.

“How do we move faster?”

I was in one of those rooms a few weeks ago when this happened with the new PwC data. The study measured 1,217 executives across 25 sectors and found that 74% of AI’s economic value is going to 20% of organisations — organisations generating 7.2 times more AI-driven revenue than their peers. The headline was stark. The room absorbed it quickly. And within two minutes, someone had already reframed it as a speed problem.

I want to be honest about something. I don’t think it’s a speed problem. I don’t think moving faster will close the gap. I think the organisations asking “how do we move faster?” are running in the right direction but playing the wrong game. And I think the PwC data, read carefully — not just read for the headline — tells you exactly why.

Let me show you what I see when I apply the Ripple Effects framework to this data. Not the first-order story, which is the one in the headlines. The second and third-order consequences that are already arriving, mostly unread.

What the Headline Is Actually Counting

The PwC data captures what most organisations want to see measured: AI investment producing economic return. By that measure, progress is real. Some organisations are seeing significant, documented, measurable advantage from their AI investment.

That’s the first-order effect. It’s true. It’s in the numbers. And it’s the story every boardroom presentation on AI will be built around for the next 18 months.

But first-order consequences are the ones visible in the room when a decision is made. They show up on dashboards. They validate investment cases. They close the loop on the choices that seemed reasonable at the time.

Second and third-order consequences are rarely in the room when the decision is made. That’s what makes them consequential.

The PwC data, if you read past the headline, reveals something specific about why value is concentrating. It isn’t that the 20% are faster. It isn’t that they have more budget. The distinguishing factor is a measurement framework.

The 20% ask: what can we now do that we couldn’t do before?

The 80% ask: where are we saving time and cost?

Those are not the same question. One optimises existing capacity. The other builds new capacity. One produces arithmetic return. The other produces compound return. And the compound return, over time, looks like 7.2 times more AI-driven revenue.

Not a speed gap. A question gap.

First-Order: The Productivity Story Everyone Is Telling

The first-order effect of AI investment is the one on every roadshow deck, every investor call, and every conference keynote this year.

AI compresses timelines. AI improves synthesis. AI assists knowledge workers, accelerates content, flags patterns in data that human analysts would take weeks to surface. For organisations that have integrated AI well, the difference in output quality and volume is real and measurable.

None of that is wrong. The productivity story is true.

What it misses is that productivity gains don’t stay evenly distributed. They never have, with any technology adoption cycle. The gains flow toward the organisations with the right measurement framework, the right integration depth, and the right strategic question. And then they compound.

That’s the shift from first-order to second-order. And it’s the shift most boardrooms aren’t tracking.

I’ve written about this pattern before, in the context of financial technology. In early 2026, I traced how the same value concentration mechanism played out in the fintech and infrastructure layers of global capital markets: the productivity gains from a technology wave were real, widely shared in the first phase, and then concentrated in the second phase as the organisations that understood what they were actually building pulled away from those that understood only what they were automating. The mechanism is the same now. The speed is different.

Second-Order: The Concentration That Nobody Is Naming

The second-order consequence of the AI productivity story isn’t visible in any single organisation’s dashboard. It only becomes visible when you look across organisations and across time.

Value is concentrating.

Seven point two times. That’s not a modest advantage. That’s not a gap that closes itself when the laggards catch up on adoption speed. That is the beginning of a compound separation — and compound separations don’t level off. They accelerate.

Here is the specific mechanism the PwC data reveals: the organisations in the 20% are not primarily using AI to reduce costs. They are using AI to do things they could not do before. New services. New analytical capabilities. New speed-to-market on product development. New capacity to hold and synthesise complexity at a scale that wasn’t previously possible. These are capability gains, not efficiency gains. They create structural advantage that builds on itself.

The organisations in the 80% are — in the majority of cases — using AI primarily to reduce costs. To generate content faster. To compress meeting notes. To automate customer service scripts. To make existing processes cheaper.

These are legitimate gains. They show up on the dashboard. They vindicate the investment case in the next budget review. But they are not building new capacity. They are optimising existing capacity. And optimised existing capacity does not compound against new capacity.

The measurement framework question matters enormously here. If you are evaluating your AI investment primarily by costs saved and time compressed, you are making decisions that look rational in the short term and create a strategic problem in the medium term. You are not building toward the 20%. You are building toward a more efficient version of the 80%.

Third-Order: The Organising Begins

This is the part of the analysis that most boardroom conversations haven’t reached. Not because it requires unusual insight, but because third-order effects are always easier to see once they’ve already arrived.

They have arrived.

When value concentrates in a pattern that is sufficiently visible and sufficiently sustained, the people and organisations outside the concentration begin to respond. They adapt strategy. They negotiate differently. They build new structures to capture a share of what existing structures are not distributing.

Three events this week alone confirm this pattern is already in motion.

Samsung’s semiconductor workforce — the workers who manufacture the physical chips that power every AI system in the world — successfully negotiated a €350,000 profit-sharing arrangement specifically tied to AI productivity gains. They did not receive this through goodwill. They received it through organised leverage and a credible documented position: we make the infrastructure without which none of your AI value exists. The company agreed. This is a third-order consequence of AI value concentration: the workers who enable the concentration are beginning to formalise their share of it.

Wikipedia’s editors are on strike against AI-driven cost reduction at the foundation of open knowledge infrastructure. The Wikimedia Foundation is processing tens of millions of AI queries per month against content that was created and maintained by volunteer intellectual labour. The editors organising the strike are not anti-technology. They are naming the third-order consequence that the cost-reduction conversation missed: if the organisation’s cost base drops because AI is now doing work that used to require human maintenance, and the volunteers who created the knowledge base see none of that value, the conditions that made that volunteer labour sustainable are undermined.

The EU research published this week is more structural than either of these. AI patenting density — where AI intellectual property is being created and owned — correlates directly with declining labour income share in the same geographic regions. Not declining wages at a single company. Declining share across entire economic zones. In regions where AI patents are concentrated, capital is capturing a growing fraction of the value that labour used to capture. Africa’s AI readiness gap is widening simultaneously: access to the infrastructure, talent, and capital needed to participate in the 20% is not evenly distributed globally, and the gap between where the value is being created and where it isn’t is already a structural economic divide.

Not separate stories. Not isolated incidents. Third-order effects of a value concentration that started with first-order decisions that looked completely reasonable at the time.

This is exactly what the Ripple Effects framework predicts: second and third-order consequences are set in motion by first-order decisions. They don’t appear immediately. They appear on a delay. And by the time they are visible, the first-order decisions that caused them are often locked in.

The Compound Problem for the 80%

I want to be specific about the compound problem, because I think it is more serious than the headlines suggest.

It has three layers.

The strategic layer. If the 20% are building compound advantage — capability that creates further capability, integration that produces further integration, data assets that grow more valuable as they grow larger — and the 80% are building optimised existing capacity, the gap does not hold at 7.2 times. It grows. This is what compound returns do. The organisations that are asking the right question now are building toward a position that will be significantly harder to approach in three years’ time.

The talent layer. When AI absorbs the entry-level tasks that teach people how to think professionally — the drafting, the research, the synthesis, the first-pass analysis that junior team members used to do as their training ground — the next generation of senior thinkers doesn’t get built. The first-order cost saving — fewer junior staff, faster output, lower operational overhead — has a third-order talent consequence most organisations will not see until it is too late to address. You optimise your current workforce while quietly defunding its replacement.

The measurement layer. Organisations measuring AI success primarily by cost reduction are not just missing the strategic picture. They are actively rewarding the wrong decisions. When the measurement framework reinforces efficiency over capability, every internal funding conversation, every project approval, every AI investment choice will trend toward the 80% game. The measurement framework becomes self-reinforcing.

Three layers of compound problem, each reinforcing the other. And the organisations experiencing this are not negligent. They are asking reasonable questions and getting reasonable answers. The problem is that reasonable answers to the wrong question still take you the wrong direction.

The pattern is not new. I wrote about it in the context of a much older technology shift: the organisations that struggled most weren’t the ones who adopted slowly. They were the ones who adopted quickly and stopped thinking about what they had set in motion. First-order success is real. It can also be a distraction.

The Question Nobody Is Asking in the Boardroom

Here is the question the PwC data is pointing toward that most boardrooms aren’t raising.

Not “how do we move faster?” Not “how do we increase our AI adoption rate?” Not even “how do we get from 80% to 20%?”

The question is: what are we actually building, and what does it set in motion?

It’s a foresight question, not a technology question. It requires leaders to look past the dashboard at what the dashboard is not measuring. It requires holding the preparation window open even when the first-order story is good. It requires the capacity to sit with second-order and third-order consequences before they arrive, not after.

And it requires a specific kind of intellectual honesty about the measurement framework.

Run this diagnostic on your organisation’s current AI investment. What percentage of your active AI projects answer the question “what can we now do that we couldn’t before?” versus “where are we saving time and cost?”

If the answer is weighted heavily toward efficiency, that is not a catastrophe. It is information. The preparation window is still open. But it tells you that you are building the right infrastructure for the wrong strategic question. And that is the kind of problem that looks manageable now and becomes structural later.

There is a second question worth raising alongside this one, and it is the question I come back to most often in leadership conversations this year. What is the human consequence for your people of the AI investment choices you are making now?

Not as a welfare question. As a strategic one.

If your people are watching AI absorb tasks that used to teach them things, and nobody is naming that, the second-order consequence arrives in the form of disengagement, reduced judgment capacity, and eventually a workforce that can operate the tools but cannot think beyond them.

Not in 2030. It’s an Immediate Futures signal. It’s already arriving in organisations that moved fast without thinking carefully about what they were optimising for.

The 20% organisations are not just using AI differently. They are leading through AI differently. Their senior teams are holding questions that their counterparts in the 80% are not holding: who decides how this changes the work? What do we owe the people whose roles are shifting? What capability are we building that compounds? Where do we draw the line?

These are not technology questions. They are leadership questions. And the organisations treating them as technology questions are delegating decisions about their strategic future to their IT departments.

What the Preparation Window Looks Like Now

I want to be careful about turning this into a checklist. The Ripple Effects framework is a diagnostic tool, not a remediation plan. What it does is show you where you are in the wave — whether you are at the first, second, or third-order stage of a set of consequences — so that your response is calibrated to the actual problem.

The actual problem in the AI concentration story is this: the third-order effects are arriving now, and most organisations in the 80% are still having first-order conversations.

The Samsung deal is done. The Wikipedia editors are organised. The EU data on capital-labour share is published. The Africa gap is widening. These are not future scenarios. They are this week’s evidence. The preparation window for the compound problem is not years. It is quarters.

The organisations that will be in the 20% in three years are the ones asking the capability question now, before the compound advantage of the current 20% becomes structurally insurmountable. The measurement framework question is not a future exercise. It is a current governance decision.

And for leaders in the 80% who are reading this: the most important move is not a new AI investment. It is a new question in the room. What can we now do that we couldn’t do before? Put that question alongside your current AI ROI metrics. See what it reveals about where your investment is actually pointed.

The conversation in the boardroom needs to change. Not the pace of it. The frame of it.

Ripple Effects Stop at the First Order at Your Peril

One of the most consistent patterns I see across 30 years of working with organisations through technology adoption cycles is this.

The organisations that struggle most are not the ones who adopted slowly.

They are the ones who adopted quickly and stopped thinking about what they had set in motion.

First-order consequences are easy to name and easy to defend. They show up in the metrics. They vindicate the investment case. They close the loop on the decisions that seemed reasonable at the time.

Second and third-order consequences require a different posture. They require leaders who are willing to look past the dashboard at what the dashboard is not measuring. They require the preparation window to be held open even when the first-order story looks strong.

The 74 and the 20 is a second-order finding. The organising that is beginning to respond to it — the Samsung deal, the Wikipedia strike, the EU research, the widening global readiness gap — is third-order. The compound talent and capability problem is third-order. All of it was set in motion by first-order decisions that seemed reasonable at the time.

The question for every leadership team sitting with the PwC data this week is not “how do we move faster?” It is “what are we actually building, and what does it set in motion?”

The preparation window is open.

What you do in it will determine whether you look at the next PwC study from the 20% or from further back in the 80%.

Choose Forward.

---

Morris Misel is a foresight strategist who works with organisations, boards, and leadership teams to interpret what’s arriving and make better strategic choices. He is based in Melbourne, Australia.