Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
This is not a verdict. It is an observation I have been turning over lately, and I am putting it here because it seems to be a pattern — and the fastest way to learn whether it is a real one is to describe it plainly and let additional eyes test it. So here is what I think I am seeing. Tell me where it breaks.
While I was already working through this, a colleague, Tanya Seda, landed on the same shape independently — in a comment on another post earlier today. Agent sprawl, she observed, reminded her of mobile device sprawl — the years when unmanaged BYOD growth forced organizations to bolt on Mobile Device Management to regain control. Her line was the one that stayed with me: the lesson was never “build better MDM.” It was “design systems so management isn’t constantly playing catch-up.” That two people arrived at the same pattern separately is part of why I suspect it is a real one.
MDM wasn’t the cure for device sprawl. It was the response to it. It managed the consequences; it did not remove the conditions that produced them.
Once I had that frame, I began to see it in more than one place — and this is where I want to be careful, because three instances of something are enough to suggest a pattern and nowhere near enough to prove one.
The reflex looks like this: when the promised value of a technology arrives more slowly than expected, organizations tend to reach for something tangible and controllable rather than examine the operating model underneath. MDM was that reach in the device era.
Cloud repatriation looks like the same reach in the cloud era. According to the Barclays CIO Survey, 86% of CIOs planned to move at least some workloads from public cloud back to private or on-premises environments — the highest figure that survey has recorded. The driver, in study after study, is not a failure of the technology but a failure of the expected value to materialize: an IDC survey found 69% of IT decision-makers reporting that actual cloud costs had exceeded their original expectations. Very few organizations are leaving the cloud entirely; public cloud spending is still growing. But the reflex is unmistakable — when the rented promise disappoints, re-own the hardware. Repatriation isn’t the cure for cloud disappointment. It is the response to it.
The same reach appears to be starting in the AI era, and this is the one I would watch most closely. Enterprises are increasingly choosing to build AI systems on their own hardware rather than rent capacity from the hyperscalers, and IDC has forecast a roughly 10% rise in hardware infrastructure sales driven by AI. IBM’s difficult quarter — customers reprioritizing capital toward servers and storage — is a single data point inside that shift, and I would not lean on it too hard, because its causes are specific and contested. But the direction is consistent: under uncertainty about where AI value will actually land, the money moves toward the thing you can put on a purchase order.
I traced this one layer deeper in a recent piece on the sprawl itself, titled “AI Sprawl Is the Fever, Not the Disease.” In that piece I saw the visible sign of something older. As I put it there: “Sprawl is not an oversight gap. It is organizational fragmentation expressed through technology.” What I am describing here is what that fragmentation does once it reaches the balance sheet: when the value it undermines runs late, the reach is for the tangible.
Set against all three is a pairing of numbers that will not leave me alone. Gartner projects the average large enterprise will run more than 150,000 AI agents by 2028, while only 13% of organizations believe they have the governance in place to manage them. And the four largest hyperscalers are guiding to roughly $725 billion in combined AI infrastructure spending in 2026 — up 77% in a single year — a figure that, by many analysts’ own admission, outpaces demonstrable AI revenue.
I have seen where this reflex can end. In February 2000, I gave a radio interview about my own company being acquired for twelve million dollars. In the segment right after mine, the chief financial officer of Nortel assured the audience the company’s climb was intact. You know what happened next — not only to Nortel, whose market value would fall more than 99% from a peak that had briefly made it one of the most valuable corporations in the world and roughly a third of the entire Toronto Stock Exchange, but to the whole optical-infrastructure sector around it. JDS Uniphase, another Ottawa-rooted company, wrote down some forty-five billion dollars in value it had paid for acquisitions made ahead of demand that never fully arrived. What broke them was not a shortage of internet. It was infrastructure built far ahead of realized value, and the reckoning that came when the value lagged the buildout.
So here is the honest difficulty, and it is the reason this is an observation and not a prediction. Buying hardware during a boom is ambiguous. It can be confidence — the technology is working so well, and demand is so real, that owning the infrastructure beats renting it. Or it can be displacement — reaching for the tangible because the expected value has not shown up, and hardware is easier to fund than the harder question of why. From the outside, the two look identical. Only the outcome tells you which one you were watching. The Nortel precedent shows how it ends when it is displacement. It does not tell me that this is displacement.
What it does tell me is which question decides it — and it is not a technology question. It is whether the organization can actually absorb and operationalize what it is buying, at the pace its investment assumes. That is the variable that separates confidence from displacement, and it is the one most easily skipped, because you cannot put “operating-model coherence” on a balance sheet the way you can put a data center on one.
Which points, finally, at the capital funding all of this. If there is a pattern here, the implication for investors is not “avoid AI.” It is that the next real differentiator may not be backing the best model or the largest pipeline, but assessing whether a company’s customers can operationalize AI at the scale the valuation assumes. Readiness as diligence. The dot-com money that survived was not the capital chasing the promise; it was the capital that backed businesses which could execute. In this wave, execution is absorption.
None of this is settled, and I am not pretending otherwise. It is three reversions to the tangible, converging at the same moment, against a precedent that shows where the reflex leads when it is left unexamined. It may be the early shape of a correction. It may be rational optimization that resolves without incident. I genuinely do not know yet — and I would rather arrive at the truth of it with help than be right about it alone.
So I will end where the honest version of this has to end: not with a conclusion, but with a question. Three reversions to the tangible, one unresolved reading, one precedent that remembers how this goes. What are you seeing?
This analysis draws on the Procurement Insights archive — an independent record, carrying zero vendor sponsorships, that I have published openly since 2007 and that consolidates documented client work, lectures, and writing reaching back to 1998. Every claim in it is held to the Provenance Ledger™: a verify-before-publish discipline that traces each assertion to a primary source and never quietly edits the record once it is posted. That record is the evidence base for two working lenses — Invariant Physics™, the constant that however far the technology advances, the operating logic must be in place first; and Implementation Physics™, its per-engagement application: the discipline of doing the readiness work before the platform, not after. Getting it right, rather than being right.
-30-
Related
Not the Cure — the Response
Posted on July 18, 2026
0
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
This is not a verdict. It is an observation I have been turning over lately, and I am putting it here because it seems to be a pattern — and the fastest way to learn whether it is a real one is to describe it plainly and let additional eyes test it. So here is what I think I am seeing. Tell me where it breaks.
While I was already working through this, a colleague, Tanya Seda, landed on the same shape independently — in a comment on another post earlier today. Agent sprawl, she observed, reminded her of mobile device sprawl — the years when unmanaged BYOD growth forced organizations to bolt on Mobile Device Management to regain control. Her line was the one that stayed with me: the lesson was never “build better MDM.” It was “design systems so management isn’t constantly playing catch-up.” That two people arrived at the same pattern separately is part of why I suspect it is a real one.
MDM wasn’t the cure for device sprawl. It was the response to it. It managed the consequences; it did not remove the conditions that produced them.
Once I had that frame, I began to see it in more than one place — and this is where I want to be careful, because three instances of something are enough to suggest a pattern and nowhere near enough to prove one.
The reflex looks like this: when the promised value of a technology arrives more slowly than expected, organizations tend to reach for something tangible and controllable rather than examine the operating model underneath. MDM was that reach in the device era.
Cloud repatriation looks like the same reach in the cloud era. According to the Barclays CIO Survey, 86% of CIOs planned to move at least some workloads from public cloud back to private or on-premises environments — the highest figure that survey has recorded. The driver, in study after study, is not a failure of the technology but a failure of the expected value to materialize: an IDC survey found 69% of IT decision-makers reporting that actual cloud costs had exceeded their original expectations. Very few organizations are leaving the cloud entirely; public cloud spending is still growing. But the reflex is unmistakable — when the rented promise disappoints, re-own the hardware. Repatriation isn’t the cure for cloud disappointment. It is the response to it.
The same reach appears to be starting in the AI era, and this is the one I would watch most closely. Enterprises are increasingly choosing to build AI systems on their own hardware rather than rent capacity from the hyperscalers, and IDC has forecast a roughly 10% rise in hardware infrastructure sales driven by AI. IBM’s difficult quarter — customers reprioritizing capital toward servers and storage — is a single data point inside that shift, and I would not lean on it too hard, because its causes are specific and contested. But the direction is consistent: under uncertainty about where AI value will actually land, the money moves toward the thing you can put on a purchase order.
I traced this one layer deeper in a recent piece on the sprawl itself, titled “AI Sprawl Is the Fever, Not the Disease.” In that piece I saw the visible sign of something older. As I put it there: “Sprawl is not an oversight gap. It is organizational fragmentation expressed through technology.” What I am describing here is what that fragmentation does once it reaches the balance sheet: when the value it undermines runs late, the reach is for the tangible.
Set against all three is a pairing of numbers that will not leave me alone. Gartner projects the average large enterprise will run more than 150,000 AI agents by 2028, while only 13% of organizations believe they have the governance in place to manage them. And the four largest hyperscalers are guiding to roughly $725 billion in combined AI infrastructure spending in 2026 — up 77% in a single year — a figure that, by many analysts’ own admission, outpaces demonstrable AI revenue.
I have seen where this reflex can end. In February 2000, I gave a radio interview about my own company being acquired for twelve million dollars. In the segment right after mine, the chief financial officer of Nortel assured the audience the company’s climb was intact. You know what happened next — not only to Nortel, whose market value would fall more than 99% from a peak that had briefly made it one of the most valuable corporations in the world and roughly a third of the entire Toronto Stock Exchange, but to the whole optical-infrastructure sector around it. JDS Uniphase, another Ottawa-rooted company, wrote down some forty-five billion dollars in value it had paid for acquisitions made ahead of demand that never fully arrived. What broke them was not a shortage of internet. It was infrastructure built far ahead of realized value, and the reckoning that came when the value lagged the buildout.
So here is the honest difficulty, and it is the reason this is an observation and not a prediction. Buying hardware during a boom is ambiguous. It can be confidence — the technology is working so well, and demand is so real, that owning the infrastructure beats renting it. Or it can be displacement — reaching for the tangible because the expected value has not shown up, and hardware is easier to fund than the harder question of why. From the outside, the two look identical. Only the outcome tells you which one you were watching. The Nortel precedent shows how it ends when it is displacement. It does not tell me that this is displacement.
What it does tell me is which question decides it — and it is not a technology question. It is whether the organization can actually absorb and operationalize what it is buying, at the pace its investment assumes. That is the variable that separates confidence from displacement, and it is the one most easily skipped, because you cannot put “operating-model coherence” on a balance sheet the way you can put a data center on one.
Which points, finally, at the capital funding all of this. If there is a pattern here, the implication for investors is not “avoid AI.” It is that the next real differentiator may not be backing the best model or the largest pipeline, but assessing whether a company’s customers can operationalize AI at the scale the valuation assumes. Readiness as diligence. The dot-com money that survived was not the capital chasing the promise; it was the capital that backed businesses which could execute. In this wave, execution is absorption.
None of this is settled, and I am not pretending otherwise. It is three reversions to the tangible, converging at the same moment, against a precedent that shows where the reflex leads when it is left unexamined. It may be the early shape of a correction. It may be rational optimization that resolves without incident. I genuinely do not know yet — and I would rather arrive at the truth of it with help than be right about it alone.
So I will end where the honest version of this has to end: not with a conclusion, but with a question. Three reversions to the tangible, one unresolved reading, one precedent that remembers how this goes. What are you seeing?
This analysis draws on the Procurement Insights archive — an independent record, carrying zero vendor sponsorships, that I have published openly since 2007 and that consolidates documented client work, lectures, and writing reaching back to 1998. Every claim in it is held to the Provenance Ledger™: a verify-before-publish discipline that traces each assertion to a primary source and never quietly edits the record once it is posted. That record is the evidence base for two working lenses — Invariant Physics™, the constant that however far the technology advances, the operating logic must be in place first; and Implementation Physics™, its per-engagement application: the discipline of doing the readiness work before the platform, not after. Getting it right, rather than being right.
-30-
Share this:
Like this:
Related