To be fair, the graphic also lists five practical steps, and McKinsey’s post opens with a line I agree with: “scaling AI and creating value from AI are not the same thing.” But the number and the question come first.
It is a number built to be remembered for its scale. The precise figure fades, but the impression remains: enormous amounts of money are moving into AI, and your organization had better move with them.
But how does $7 trillion invested in AI infrastructure help any particular organization succeed with AI?
It does not.
The figure describes projected capital investment. It does not establish that your organization needs AI, is ready for it, has selected the right application or will produce a measurable return from it.
It tells you that other people are jumping.
As my father used to ask me:
If your friends jumped off the Brooklyn Bridge, would you?
What the $7 trillion actually represents
McKinsey’s underlying analysis is more careful than the promotional graphic.
In The AI value curve, one of the articles in its collection The new economics of AI, McKinsey estimates that by 2030 roughly $7 trillion in capital will have been invested in AI infrastructure, and that the technology will need to generate about $760 billion in annual profit pools to justify it.
Those are significant economic questions for infrastructure providers, investors, hyperscalers, semiconductor companies and energy suppliers.
They are not evidence that an individual enterprise will create value by deploying AI.
McKinsey’s longer analysis acknowledges this distinction. It warns that many organizations are still measuring AI adoption instead of value, tracking users, copilots deployed and tokens consumed, and that effective AI costs are already climbing as usage grows faster than unit prices fall.
That is sensible guidance.
Its first recommendation is equally sensible: map the opportunity at the task level, starting with a task-level audit of your top five cost centers. That is necessary but not sufficient. Every task can succeed while the operation fails, because the failure lives between the tasks, in what we at Hansen call the seams.
The graphic’s practical steps come after the number and the question, and that sequence carries its own message:
An extraordinary amount of money will be invested.
AI therefore represents an unavoidable economic transformation.
Your strategy should be aligned with it.
If you are not acting, you risk being left behind.
The detailed analysis asks whether AI produces value. The promotional frame first creates pressure to participate.
That is the structure of FOMO, whether or not it is intended.
The man who put bacon on the breakfast table
An email arrived this morning from Mark Manson’s weekly newsletter, Your Next Breakthrough, describing the work of Edward Bernays, widely regarded as one of the founders of modern public relations. It was promoting the latest episode of Manson’s podcast, Solved, on manipulation.
From Mark Manson’s newsletter, Your Next Breakthrough.
In the 1920s, the Beech-Nut Packing Company wanted to sell more bacon. Bernays reportedly had the company’s physician write to some 5,000 doctors, asking whether a substantial breakfast was healthier than a light one. About 4,500 agreed that it was.
Bernays turned the responses into a national story. Newspapers reported that thousands of physicians recommended a hearty breakfast. Bacon and eggs became the commercially useful embodiment of that recommendation.
The doctors had not established that Americans needed to buy more Beech-Nut bacon. They had expressed a view about breakfast.
The campaign supplied the association:
Doctors recommend a substantial breakfast. Bacon belongs in a substantial breakfast. Therefore, eat more bacon.
The authority was real. The conclusion was commercially constructed.
Bernays later applied a similar method for the American Tobacco Company. In 1929, he arranged for women to smoke cigarettes during New York’s Easter Parade and presented the cigarettes as “torches of freedom.”
He did not lead with tobacco.
He connected smoking to women’s independence and allowed the audience to carry the political meaning into the commercial product.
Bernays called the broader practice the engineering of consent.
The surveys in our own industry
Procurement has its own version of the physicians’ poll.
Our industry is informed and educated largely through surveys. Consulting firms such as Deloitte and The Hackett Group publish them. So do analyst firms such as Gartner and Ardent Partners, communities and events such as Art of Procurement and ProcureCon, and solution providers that sponsor or distribute the research. Ardent Partners’ 2026 study of 311 CPOs, for example, was sponsored by GEP.
Most of this research is legitimate and useful, and sponsorship is not wrongdoing. But two patterns should look familiar.
First, the figures overwhelmingly measure adoption and intent, not outcomes. One independent review of the leading studies found that every figure it examined measured adoption or intent, and none measured a time, cost or accuracy outcome. That is the same gap McKinsey itself warns about.
Second, the same question produces very different answers. Ardent Partners’ 2026 research reports that 58% of procurement organizations are actively using or piloting AI. The 2026 ProcureCon CPO Report finds 11% implementing AI with measurable impact. Both can be accurate, because they measure different things. But a reader who sees only one headline comes away with a very different picture of where the market stands.
The authority is real, as it was with Bernays’ physicians. The data may be sound. But the headline figure carries an implication, that everyone is moving, and the reader supplies the urgency.
That is why our own labs work the other way around: we lock a baseline before the outcome is known, measure the outcome against it, and publish the result even when it rejects our own hypothesis.
The number does not have to prove the conclusion
The McKinsey graphic does not tell executives to invest merely because others are investing.
It does not need to.
The structure carries the implication:
$7 trillion will be invested in AI. Is your strategy aligned?
The figure supplies scale. McKinsey supplies institutional authority. The question supplies the possibility that your organization is out of alignment with the future.
The reader supplies the anxiety.
This does not mean McKinsey fabricated the estimate or consciously set out to manipulate anyone. Motive is not required to recognize the mechanism. The question is what the evidence proves and what the presentation encourages its audience to conclude.
The $7 trillion figure proves that a massive infrastructure buildout is expected.
It does not prove:
that your organization should invest in AI;
which operation should use it;
whether the organization understands that operation;
whether AI can change the outcome it currently produces;
whether the required data and authority exist;
whether the investment will create value;
or whether the organization can recognize when the system is confidently operating inside the wrong frame.
Market magnitude is not organizational readiness.
The warning about manipulation used manipulation
There was another revealing feature in the email about Bernays.
After explaining that manipulation surrounds us, it warned that merely knowing the techniques offers little protection. It then promised:
It uses a compelling historical story, authority, fear, curiosity, novelty and an unresolved information gap to produce the click.
The podcast may provide valuable information. The point is that recognizing persuasive methods does not prevent us from responding to them. Even a message warning about manipulation can use the architecture it describes.
Executives therefore need more than media literacy or a checklist of cognitive biases. They need an operating method that prevents an attractive premise from becoming the organization’s premise without being tested.
What would you decide without the number?
There is a simple question every executive should ask when presented with a large market forecast:
What would we decide if this number had never been shown to us?
Would the organization still have identified the same operating problem?
Would it still have selected the same investment?
Would it still expect the same outcome?
Could it trace the evidence connecting that investment to the outcome?
If the business case weakens when the market-size statistic is removed, the statistic may be carrying more of the decision than the operating evidence.
That is particularly dangerous with AI because the technology can turn an untested premise into action at unprecedented speed.
A persuasive frame no longer remains inside a presentation. It can be encoded into an ontology, embedded in an agent, enforced through a workflow and repeated across thousands of decisions.
A second brain can institutionalize the first assumption
In my previous article, How the Hansen-McKinsey AI Operating Systems Address Choudary’s “Human-in-the-Loop Fallacy”, I argued that a second brain can still learn the wrong organization. The same is true of an accepted premise. If the operating context begins with “AI is inevitable,” or with the documented process standing in for the real one, the system can reason flawlessly inside an incomplete frame, and the human asked to click approve inherits accountability for a frame they did not establish.
Phase 0 reopens the frame
Bernays’ method worked by establishing the frame before the audience considered the action.
Phase 0™ works in the opposite direction.
It does not begin by asking:
How should we deploy AI?
It begins with:
What outcome is the organization producing now, and what is actually producing it?
The organization then traces backward through:
decisions;
incentives;
informal handoffs;
shadow functions;
workarounds;
external agents;
authority boundaries;
and dependencies absent from the documented process.
Only after that operating reality has been established should the organization ask:
What in this operation should an agent take over, and why?
What could an agent enhance or extend that the operation could not do before?
This is more than readiness assessment. It is a defense against framing becoming architecture.
The contemporary engineering of consent
The AI industry does not need to tell executives that a specific implementation will succeed.
It only has to establish three ideas:
AI investment is enormous.
Competitors are moving.
Delay represents strategic danger.
Once those premises are accepted, the debate moves immediately to models, agents, platforms, governance and cost.
The most important question disappears:
What operating evidence established that this organization should act?
That is why $7 trillion is such an effective headline. It does not answer the question. It makes executives feel that asking it may already place them behind.
From engineered consent to evidenced authority
The defense is not cynicism. McKinsey’s analysis contains useful economic and architectural work, just as a persuasive message may contain accurate information.
The defense is separating the evidence from the conclusion it is being used to carry.
The Phase 0™ sequence I set out in my previous article does that structurally:
Start with the outcome being produced now.
Trace backward to establish the real operation.
Identify shadow functions, incentives, workarounds and external agents.
Establish the Firm Fence around that operating reality.
Use independent AI models to challenge the understanding.
Preserve disagreement through separate Main and Shadow panels.
Assign decision authority and escalation to a named Human Orchestrator.
Feed outcomes back through the Infinity Loopback without rewriting what the system originally understood.
This does not make executives immune to persuasion.
It prevents persuasion from silently becoming operating truth.
Everyone may be jumping. That is not your business case.
The projected $7 trillion investment matters. It will shape energy markets, semiconductor capacity, infrastructure financing and the economics of AI services.
It may also create bubbles, dependencies and concentrated exposures that procurement and supply-chain leaders should trace carefully.
But it is not a reason for an organization to deploy AI.
The business case begins with the organization’s own outcome, its real operation and evidence that the proposed intervention can change both.
Bernays engineered consent by controlling the frame.
Phase 0 reopens the frame before the organization acts.
And my father’s question remains as useful in the AI era as it was when I first heard it:
If your friends jumped off the Brooklyn Bridge, would you?
Before following the $7 trillion crowd, find out why everyone is jumping, and whether there is any water below.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
Technology Surveys: A Repeating Phenomenon
Each technology era arrived with an authoritative forecast and a sense that everyone was moving. Three of the four headlines were about $7 trillion. Forecasts and outcomes as reported at the time and in later reviews; failure rates vary by study and definition.
$7 Trillion Will Be Invested in AI. What Does That Prove About Your Organization?
Posted on November 5, 2026
0
Jon W. Hansen, FCIPS | Procurement Insights | Hansen Models™
What does Bernays’ bacon have to do with the latest survey in our industry?
A recent McKinsey graphic on LinkedIn, promoting its collection The new economics of AI, opens with an arresting number:
Below it appears the question:
Graphic: McKinsey & Company, on LinkedIn, promoting The new economics of AI and its article The AI value curve.
To be fair, the graphic also lists five practical steps, and McKinsey’s post opens with a line I agree with: “scaling AI and creating value from AI are not the same thing.” But the number and the question come first.
It is a number built to be remembered for its scale. The precise figure fades, but the impression remains: enormous amounts of money are moving into AI, and your organization had better move with them.
But how does $7 trillion invested in AI infrastructure help any particular organization succeed with AI?
It does not.
The figure describes projected capital investment. It does not establish that your organization needs AI, is ready for it, has selected the right application or will produce a measurable return from it.
It tells you that other people are jumping.
As my father used to ask me:
What the $7 trillion actually represents
McKinsey’s underlying analysis is more careful than the promotional graphic.
In The AI value curve, one of the articles in its collection The new economics of AI, McKinsey estimates that by 2030 roughly $7 trillion in capital will have been invested in AI infrastructure, and that the technology will need to generate about $760 billion in annual profit pools to justify it.
Those are significant economic questions for infrastructure providers, investors, hyperscalers, semiconductor companies and energy suppliers.
They are not evidence that an individual enterprise will create value by deploying AI.
McKinsey’s longer analysis acknowledges this distinction. It warns that many organizations are still measuring AI adoption instead of value, tracking users, copilots deployed and tokens consumed, and that effective AI costs are already climbing as usage grows faster than unit prices fall.
That is sensible guidance.
Its first recommendation is equally sensible: map the opportunity at the task level, starting with a task-level audit of your top five cost centers. That is necessary but not sufficient. Every task can succeed while the operation fails, because the failure lives between the tasks, in what we at Hansen call the seams.
The graphic’s practical steps come after the number and the question, and that sequence carries its own message:
The detailed analysis asks whether AI produces value. The promotional frame first creates pressure to participate.
That is the structure of FOMO, whether or not it is intended.
The man who put bacon on the breakfast table
An email arrived this morning from Mark Manson’s weekly newsletter, Your Next Breakthrough, describing the work of Edward Bernays, widely regarded as one of the founders of modern public relations. It was promoting the latest episode of Manson’s podcast, Solved, on manipulation.
From Mark Manson’s newsletter, Your Next Breakthrough.
In the 1920s, the Beech-Nut Packing Company wanted to sell more bacon. Bernays reportedly had the company’s physician write to some 5,000 doctors, asking whether a substantial breakfast was healthier than a light one. About 4,500 agreed that it was.
Bernays turned the responses into a national story. Newspapers reported that thousands of physicians recommended a hearty breakfast. Bacon and eggs became the commercially useful embodiment of that recommendation.
The doctors had not established that Americans needed to buy more Beech-Nut bacon. They had expressed a view about breakfast.
The campaign supplied the association:
The authority was real. The conclusion was commercially constructed.
Bernays later applied a similar method for the American Tobacco Company. In 1929, he arranged for women to smoke cigarettes during New York’s Easter Parade and presented the cigarettes as “torches of freedom.”
He did not lead with tobacco.
He connected smoking to women’s independence and allowed the audience to carry the political meaning into the commercial product.
Bernays called the broader practice the engineering of consent.
The surveys in our own industry
Procurement has its own version of the physicians’ poll.
Our industry is informed and educated largely through surveys. Consulting firms such as Deloitte and The Hackett Group publish them. So do analyst firms such as Gartner and Ardent Partners, communities and events such as Art of Procurement and ProcureCon, and solution providers that sponsor or distribute the research. Ardent Partners’ 2026 study of 311 CPOs, for example, was sponsored by GEP.
Most of this research is legitimate and useful, and sponsorship is not wrongdoing. But two patterns should look familiar.
First, the figures overwhelmingly measure adoption and intent, not outcomes. One independent review of the leading studies found that every figure it examined measured adoption or intent, and none measured a time, cost or accuracy outcome. That is the same gap McKinsey itself warns about.
Second, the same question produces very different answers. Ardent Partners’ 2026 research reports that 58% of procurement organizations are actively using or piloting AI. The 2026 ProcureCon CPO Report finds 11% implementing AI with measurable impact. Both can be accurate, because they measure different things. But a reader who sees only one headline comes away with a very different picture of where the market stands.
The authority is real, as it was with Bernays’ physicians. The data may be sound. But the headline figure carries an implication, that everyone is moving, and the reader supplies the urgency.
That is why our own labs work the other way around: we lock a baseline before the outcome is known, measure the outcome against it, and publish the result even when it rejects our own hypothesis.
The number does not have to prove the conclusion
The McKinsey graphic does not tell executives to invest merely because others are investing.
It does not need to.
The structure carries the implication:
The figure supplies scale. McKinsey supplies institutional authority. The question supplies the possibility that your organization is out of alignment with the future.
The reader supplies the anxiety.
This does not mean McKinsey fabricated the estimate or consciously set out to manipulate anyone. Motive is not required to recognize the mechanism. The question is what the evidence proves and what the presentation encourages its audience to conclude.
The $7 trillion figure proves that a massive infrastructure buildout is expected.
It does not prove:
Market magnitude is not organizational readiness.
The warning about manipulation used manipulation
There was another revealing feature in the email about Bernays.
After explaining that manipulation surrounds us, it warned that merely knowing the techniques offers little protection. It then promised:
That is an effective promotion.
It uses a compelling historical story, authority, fear, curiosity, novelty and an unresolved information gap to produce the click.
The podcast may provide valuable information. The point is that recognizing persuasive methods does not prevent us from responding to them. Even a message warning about manipulation can use the architecture it describes.
Executives therefore need more than media literacy or a checklist of cognitive biases. They need an operating method that prevents an attractive premise from becoming the organization’s premise without being tested.
What would you decide without the number?
There is a simple question every executive should ask when presented with a large market forecast:
Would the organization still have identified the same operating problem?
Would it still have selected the same investment?
Would it still expect the same outcome?
Could it trace the evidence connecting that investment to the outcome?
If the business case weakens when the market-size statistic is removed, the statistic may be carrying more of the decision than the operating evidence.
That is particularly dangerous with AI because the technology can turn an untested premise into action at unprecedented speed.
A persuasive frame no longer remains inside a presentation. It can be encoded into an ontology, embedded in an agent, enforced through a workflow and repeated across thousands of decisions.
A second brain can institutionalize the first assumption
In my previous article, How the Hansen-McKinsey AI Operating Systems Address Choudary’s “Human-in-the-Loop Fallacy”, I argued that a second brain can still learn the wrong organization. The same is true of an accepted premise. If the operating context begins with “AI is inevitable,” or with the documented process standing in for the real one, the system can reason flawlessly inside an incomplete frame, and the human asked to click approve inherits accountability for a frame they did not establish.
Phase 0 reopens the frame
Bernays’ method worked by establishing the frame before the audience considered the action.
Phase 0™ works in the opposite direction.
It does not begin by asking:
It begins with:
The organization then traces backward through:
Only after that operating reality has been established should the organization ask:
This is more than readiness assessment. It is a defense against framing becoming architecture.
The contemporary engineering of consent
The AI industry does not need to tell executives that a specific implementation will succeed.
It only has to establish three ideas:
Once those premises are accepted, the debate moves immediately to models, agents, platforms, governance and cost.
The most important question disappears:
That is why $7 trillion is such an effective headline. It does not answer the question. It makes executives feel that asking it may already place them behind.
From engineered consent to evidenced authority
The defense is not cynicism. McKinsey’s analysis contains useful economic and architectural work, just as a persuasive message may contain accurate information.
The defense is separating the evidence from the conclusion it is being used to carry.
The Phase 0™ sequence I set out in my previous article does that structurally:
This does not make executives immune to persuasion.
It prevents persuasion from silently becoming operating truth.
Everyone may be jumping. That is not your business case.
The projected $7 trillion investment matters. It will shape energy markets, semiconductor capacity, infrastructure financing and the economics of AI services.
It may also create bubbles, dependencies and concentrated exposures that procurement and supply-chain leaders should trace carefully.
But it is not a reason for an organization to deploy AI.
The business case begins with the organization’s own outcome, its real operation and evidence that the proposed intervention can change both.
Bernays engineered consent by controlling the frame.
Phase 0 reopens the frame before the organization acts.
And my father’s question remains as useful in the AI era as it was when I first heard it:
Before following the $7 trillion crowd, find out why everyone is jumping, and whether there is any water below.
Sources and related reading
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
Technology Surveys: A Repeating Phenomenon
Each technology era arrived with an authoritative forecast and a sense that everyone was moving. Three of the four headlines were about $7 trillion. Forecasts and outcomes as reported at the time and in later reviews; failure rates vary by study and definition.
Sources for the table
Share this:
Like this:
Related