In every boardroom we've sat in for the last eighteen months, the conversation has the same shape.
Many of our enterprise partners have overinvested in AI technology, buying ahead of their organization's ability to absorb it. Others have done the opposite, paralyzed by governance complexity, risk frameworks, and the politics of who owns what.
They began at different starting points, but are headed in the same destination. The capital is out the door, or stuck in committee, and the return is not yet showing up in the numbers that matter.
When boards ask where the value is, the answers come back in minutes saved per ticket, per claim, per email. Locally credible. Aggregated across the enterprise, they do not come close to justifying the investment thesis that was presented twelve to eighteen months ago.
The instinct is to revisit the technology, the vendors, or the sequencing. In our experience, the variable that explains most of the shortfall is the workforce, the people expected to absorb these tools, change how they work, and turn pilots into operating leverage.
It is the largest unaccounted line item on the corporate balance sheet today.
By the end of 2026, global enterprises will have spent more than $2.5 trillion on artificial intelligence (Gartner, 2026). Boston Consulting Group puts the share of organizations seeing meaningful financial returns at roughly one in four. MIT and BCG, working together, found that 63% of AI transformation failures are caused by human factors, leadership misalignment, capability gaps, change fatigue, and the slow erosion of trust between the people building the technology and the people expected to use it.
And yet, when companies allocate AI budget, the split is roughly 80% to strategy and technology, 20% to the people who will actually have to make it work.
This is the 80/63 mismatch. Eighty percent of investment going to one side of the equation. Sixty-three percent of failures coming from the other.
It is the single most expensive misallocation in enterprise IT today.
There was a moment, maybe three years ago, when buying the right technology was the differentiator. That moment has passed. Foundation models are converging in capability. Tooling is commoditizing. Implementation partners are saturated.
Today, AI advantage is not built by the company with the best stack. It is built by the company with the leaders who can lead in an AI-shaped operating environment, the workforce that can actually use the tools, and the organizational nervous system that can absorb continuous change without burning out.
Or, said in plain language: the company has to be ready for what it is buying.
Most are not. Not because they are foolish or under-resourced. They are ready in the places they have always been ready: strategy decks, vendor selection, infrastructure. They are not yet ready in the places that matter most now: leadership behaviour, decision speed, capability density, and organizational trust.
The choice for senior leaders in 2026 should be clear. You can keep investing on the 80% side and hope the 20% works itself out. Or you can deliberately build the 20%, and watch the 80% finally start delivering.
Here is what that looks like in practice.
The first move is not technical. It is a change in posture at the top.
In organizations where AI is delivering, senior leaders treat AI the way they treat capital allocation: a strategic instrument they understand well enough to challenge, not a black box they delegate. They do not need to write Python. They do need to read an AI roadmap with the same fluency they read a P&L.
The fastest path to that fluency is not a course. It is a Leadership Circle, a recurring, cross-functional working forum where the executive team, a handful of operators, and a small number of external experts meet monthly to debate live decisions. These are not training sessions or strategy theatre. They are working sessions on the decisions that are actually live this quarter.
We have seen this single ritual shift the AI conversation in DAX-listed and FTSE-listed companies more than any single training programme. It is also, not coincidentally, the lowest-cost intervention on this list.
Try this: In the next 30 days, install a 90-minute monthly Leadership Circle on AI decisions. Three rules: a live decision on the table every meeting, no slides over 10 minutes, an external voice in the room every time.
The second move is the one most companies get wrong, and it is the one with the highest compounding return.
The default response to the AI capability gap is the AI module: a one-off training, an LMS course, a town hall with a vendor demo. We have studied dozens of these programs. They produce awareness. They do not produce capability. And they almost never produce behaviour change.
What works is the opposite of a module. It is an AI mandate: a structured, hands-on capability-building programme tied to real business challenges your teams already own. People learn by solving the problems they actually have, with the tools they actually use, in the contexts they actually work in. Theory disappears. Capability sticks.
This is the design principle behind every workforce activation programme we run with our partners: learning is not a place you go. It is the way work happens.
Try this: Identify the three business challenges in your organization that AI could meaningfully change in the next two quarters. Build your capability programme around those exact challenges. If your L&D team can't tell you what those challenges are, that is your starting problem.
The third move is what stops AI from getting stuck in the pilot phase. It is also the one most often skipped.
We call the failure mode the Pilot Paradox: AI pilots succeed in carefully chosen, well-resourced corners of the organization, and then fail to scale into the rest. Not because the technology stops working, it works fine. They fail because the organization around the pilot was never designed to absorb it.
Scaling AI is a change management problem, not a technology problem. It requires what we call the change muscle: the deliberate, trained capability of the organization to adapt repeatedly without exhausting itself. That muscle is built through three components: networks of change agents who carry the work peer-to-peer, communication that frames change as ongoing rather than episodic, and leaders who model the behaviour they expect.
Without the muscle, every AI rollout starts from zero. With the muscle, each rollout teaches the organization how to do the next one faster.
Try this: Map the last three transformations your organization has run. For each, ask: what did we learn that we are now consistently using? If the honest answer is "nothing," your change muscle is atrophied. That is the first thing to fix, before the next AI rollout.
The fourth move is what makes the first three sustainable: change how you measure success.
The current default, productivity gain per task, is the wrong frame. It captures a fraction of the actual value AI is creating, and it badly under-measures the leading indicators of long-term return. A more useful frame, and the one we use with our partners, is the human dividend: the compounding return on capability, decision speed, and engagement that AI investment unlocks when the human side is healthy.
Concretely, this means tracking three things alongside productivity: leadership AI fluency (can your top 200 leaders make AI-informed decisions without a translator?), capability density (what percentage of your workforce can productively use AI in their current role?), and change-readiness scores (how prepared is your organization to absorb the next rollout?).
Companies that track these numbers move faster. Companies that don't track them learn, eventually and expensively, that their productivity numbers were a lagging indicator of something they had stopped investing in.
Try this: Add three lines to your next AI quarterly review. Leadership fluency. Capability density. Change-readiness. They take an afternoon to build a baseline. They will tell you in 90 days whether your AI investment is on a curve or on a plateau.
The companies that win the next phase of AI will not be the ones with the most technology. They will be the ones whose leaders, workforces, and organizations are ready to lead it.
That readiness is not a posture. It is a capability you build deliberately. It has four components, and a 90-day starting move on each:
None of this requires a new vendor. None of it requires more strategy. It requires a deliberate decision to spend the next quarter investing in the side of the equation that has been quietly draining the value out of every AI investment you have already made.
The 80/63 mismatch is not destiny. It is a choice. And it is the most consequential choice your leadership team will make in 2026.
If you are a CEO, CHRO, or transformation leader running an AI agenda in 2026, we run a 90-minute Human-AI Collaboration board diagnostic with senior teams. It is live-facilitated, designed for 4–8 leaders, and covers the four moves in this article applied to your specific context.
You will leave with: a current-state read on your 80/63 mismatch, a prioritized 90-day plan, and a shared language across your leadership team.
To book the diagnostic with us below.
