Gartner recently published a diagram that has been circulating widely — The AI Adoption Gap. It is a good piece of work, and the gap it describes is real. Two curves on a field of rising capability and time: an upper “AI innovation race” run by technology providers, climbing toward AGI; a lower “AI outcome race” run by the customers meant to benefit. Between them, a widening wedge. The accompanying webinar names the cause plainly: AI is ready, but most workforces aren’t — with 71% of CIOs reporting their people are not prepared.
Here is the part most readers move past: you can train every employee, hire the most capable people available, achieve complete uptake of the tool — and still produce nothing durable. Because the variable that determines the outcome was never on the talent axis to begin with. The gap is real. The word on it — adoption — is quietly doing the wrong work.
[GRAPHIC 1 — Gartner “The AI Adoption Gap”]
The word is not neutral
“Adoption” frames the gap as a question of uptake. And uptake points at people: readiness, skills, training, change management. Choose that word and you have already decided what the problem is and where the fix lives — in the workforce. Reskill them, upskill them, manage the change. The “workforce isn’t prepared” conclusion does not follow from the evidence so much as from the vocabulary. It was baked into the framing before the first data point was plotted.
I would call the same gap something different: an absorption gap. And the distinction is not semantic.
An adoption gap asks have the people taken up the tool? An absorption gap asks something prior and larger: can the organization’s actual operating conditions convert this capability into a sustained outcome? Adoption is whether people use it. Absorption is whether the business can metabolize what the technology produces — whether the processes, incentives, decision rights, and structures around the tool let its output turn into a result that holds.
That is why you can close the adoption gap completely and leave the absorption gap wide open. Full uptake, capable people, the best tool available — and no durable result, because the conditions that decide the outcome sit off the talent axis entirely. The remedy you reach for is determined by the word you start with. Start with “adoption,” and you spend your budget and your eighteen months on training while the binding constraint goes untouched.
What 1998 settled
I have the cleanest possible disproof of the talent framing, and it is twenty-seven years old.
In 1998, the Department of National Defence was hitting 51% on a contract requiring 90% next-day delivery, and the service provider was at risk of losing it. The brief was to automate procurement. The people were capable. The suppliers were capable. The systems were capable. The processes were capable. Every input the adoption framing points to was already present.
The breakthrough did not come from better technology, more training, or a more prepared workforce. It came from a single question no one thought was a technology question: what time of day do orders come in?
The service technicians were holding their MRO orders until 4 p.m. to hit a front-end metric — and by 4 p.m. the parts cost more and could not clear customs in time. The 51% was never a capability problem or a talent problem. It was a misaligned-incentive problem: a hidden condition, entirely off the capability axis, that determined the outcome. Surface it, and delivery went from 51% to 97.3% in three months and held for seven years. Technology was not introduced until the fourth and fifth month — after the conditions were right.
If the talent was already present and the gap stayed open anyway, then by definition it was never a talent gap. A single field case settles it. And most AI “adoption” gaps look exactly like this one: the tools, the people, and the data are present, while a misaligned incentive or an unmapped structure quietly holds the outcome flat.
The gap that predates AI
This same gap has opened in every prior technology era, at every level of capability — ERP, e-procurement, the SaaS and cloud era, digital transformation, and now AI. Each arrived with new capability and a new vocabulary. The question of whether the organization could absorb what the technology produced did not change across any of them. And that recurrence is the proof: if a gap reappears at every level of capability, the variable that determines it cannot live on the capability axis. It is absorption.
[GRAPHIC 2 — Hansen “The Gap That Predates AI”]
Gartner’s own data is pointing past the word
The most telling development is that Gartner’s newest research has begun describing absorption while still calling it adoption.
In its Q1 2026 global workforce survey of more than 12,000 employees and managers, Gartner found leaders “mistaking basic access or adoption metrics for transformation” — an “enablement illusion,” in their words, “hiding risks and draining ROI.” They have stated that AI adoption is “a culture issue, not just a training issue,” and that “standard software training and technical learning do not improve workforce sentiment or build trust.” On the financial reality: in a survey of 506 technology leaders, 72% reported their organizations were breaking even or losing money on their AI investments.
Read those carefully. Adoption metrics that do not convert into transformation. Training that does not produce the result. Money not made despite the tools being in place. That is not a description of an adoption gap. It is a description of an absorption gap — uptake achieved, outcome absent. The data has already moved. The vocabulary has not.
Why it matters
The risk in the adoption framing is not that it is wrong about the gap existing. It is that it routes you to the wrong remedy. Believe you have an adoption gap and you spend on reskilling and tool enablement — and may close that gap while still missing the outcome, because you never touched the conditions that determine it.
Understand it as an absorption gap and you do something different first. Before deploying the capability, you diagnose the operating conditions that decide whether it converts — the incentives, the decision rights, the process realities, the structural alignments. You ask the question nobody thinks is a technology question. Then you introduce the technology, onto a foundation that can absorb it.
Capability has rarely been the binding constraint. For nearly three decades, across every technology era, the binding constraint has been organizational absorption. The technology changes. The question does not.
The gap is real. The word we use to name it decides whether we fix it.
Based on observations documented continuously in the Procurement Insights archive since 1998 — twenty-seven years, 3,300+ independent documents, zero vendor sponsorships. The 1998 Department of National Defence engagement is the origin of the diagnostic discipline referenced here. Gartner figures cited from its Q1 2026 global workforce enablement survey and 2025 CIO AI-investment surveys.
-30-
Related
Gartner Calls It an Adoption Gap. The Evidence Says It’s an Absorption Gap.
Posted on May 30, 2026
0
Gartner recently published a diagram that has been circulating widely — The AI Adoption Gap. It is a good piece of work, and the gap it describes is real. Two curves on a field of rising capability and time: an upper “AI innovation race” run by technology providers, climbing toward AGI; a lower “AI outcome race” run by the customers meant to benefit. Between them, a widening wedge. The accompanying webinar names the cause plainly: AI is ready, but most workforces aren’t — with 71% of CIOs reporting their people are not prepared.
Here is the part most readers move past: you can train every employee, hire the most capable people available, achieve complete uptake of the tool — and still produce nothing durable. Because the variable that determines the outcome was never on the talent axis to begin with. The gap is real. The word on it — adoption — is quietly doing the wrong work.
[GRAPHIC 1 — Gartner “The AI Adoption Gap”]
The word is not neutral
“Adoption” frames the gap as a question of uptake. And uptake points at people: readiness, skills, training, change management. Choose that word and you have already decided what the problem is and where the fix lives — in the workforce. Reskill them, upskill them, manage the change. The “workforce isn’t prepared” conclusion does not follow from the evidence so much as from the vocabulary. It was baked into the framing before the first data point was plotted.
I would call the same gap something different: an absorption gap. And the distinction is not semantic.
An adoption gap asks have the people taken up the tool? An absorption gap asks something prior and larger: can the organization’s actual operating conditions convert this capability into a sustained outcome? Adoption is whether people use it. Absorption is whether the business can metabolize what the technology produces — whether the processes, incentives, decision rights, and structures around the tool let its output turn into a result that holds.
That is why you can close the adoption gap completely and leave the absorption gap wide open. Full uptake, capable people, the best tool available — and no durable result, because the conditions that decide the outcome sit off the talent axis entirely. The remedy you reach for is determined by the word you start with. Start with “adoption,” and you spend your budget and your eighteen months on training while the binding constraint goes untouched.
What 1998 settled
I have the cleanest possible disproof of the talent framing, and it is twenty-seven years old.
In 1998, the Department of National Defence was hitting 51% on a contract requiring 90% next-day delivery, and the service provider was at risk of losing it. The brief was to automate procurement. The people were capable. The suppliers were capable. The systems were capable. The processes were capable. Every input the adoption framing points to was already present.
The breakthrough did not come from better technology, more training, or a more prepared workforce. It came from a single question no one thought was a technology question: what time of day do orders come in?
The service technicians were holding their MRO orders until 4 p.m. to hit a front-end metric — and by 4 p.m. the parts cost more and could not clear customs in time. The 51% was never a capability problem or a talent problem. It was a misaligned-incentive problem: a hidden condition, entirely off the capability axis, that determined the outcome. Surface it, and delivery went from 51% to 97.3% in three months and held for seven years. Technology was not introduced until the fourth and fifth month — after the conditions were right.
If the talent was already present and the gap stayed open anyway, then by definition it was never a talent gap. A single field case settles it. And most AI “adoption” gaps look exactly like this one: the tools, the people, and the data are present, while a misaligned incentive or an unmapped structure quietly holds the outcome flat.
The gap that predates AI
This same gap has opened in every prior technology era, at every level of capability — ERP, e-procurement, the SaaS and cloud era, digital transformation, and now AI. Each arrived with new capability and a new vocabulary. The question of whether the organization could absorb what the technology produced did not change across any of them. And that recurrence is the proof: if a gap reappears at every level of capability, the variable that determines it cannot live on the capability axis. It is absorption.
[GRAPHIC 2 — Hansen “The Gap That Predates AI”]
Gartner’s own data is pointing past the word
The most telling development is that Gartner’s newest research has begun describing absorption while still calling it adoption.
In its Q1 2026 global workforce survey of more than 12,000 employees and managers, Gartner found leaders “mistaking basic access or adoption metrics for transformation” — an “enablement illusion,” in their words, “hiding risks and draining ROI.” They have stated that AI adoption is “a culture issue, not just a training issue,” and that “standard software training and technical learning do not improve workforce sentiment or build trust.” On the financial reality: in a survey of 506 technology leaders, 72% reported their organizations were breaking even or losing money on their AI investments.
Read those carefully. Adoption metrics that do not convert into transformation. Training that does not produce the result. Money not made despite the tools being in place. That is not a description of an adoption gap. It is a description of an absorption gap — uptake achieved, outcome absent. The data has already moved. The vocabulary has not.
Why it matters
The risk in the adoption framing is not that it is wrong about the gap existing. It is that it routes you to the wrong remedy. Believe you have an adoption gap and you spend on reskilling and tool enablement — and may close that gap while still missing the outcome, because you never touched the conditions that determine it.
Understand it as an absorption gap and you do something different first. Before deploying the capability, you diagnose the operating conditions that decide whether it converts — the incentives, the decision rights, the process realities, the structural alignments. You ask the question nobody thinks is a technology question. Then you introduce the technology, onto a foundation that can absorb it.
Capability has rarely been the binding constraint. For nearly three decades, across every technology era, the binding constraint has been organizational absorption. The technology changes. The question does not.
The gap is real. The word we use to name it decides whether we fix it.
Based on observations documented continuously in the Procurement Insights archive since 1998 — twenty-seven years, 3,300+ independent documents, zero vendor sponsorships. The 1998 Department of National Defence engagement is the origin of the diagnostic discipline referenced here. Gartner figures cited from its Q1 2026 global workforce enablement survey and 2025 CIO AI-investment surveys.
-30-
Share this:
Like this:
Related