Traditional consulting is under pressure from many directions, but two of them matter more than the rest. Artificial intelligence is compressing the value of coordination, and it is compressing the value of junior labor. Everything else follows from that.
That is the real story.
The superficial story is that firms will become more efficient, more global, more standardized, and more digitally enabled. That is true, but it is not yet the point. The deeper point is that the economic logic that justified the modern consulting firm is being quietly rewritten. The old model was built on a world in which coordination was expensive and human cognition was scarce. AI changes both conditions at once. It lowers the cost of making many people act like one mind, and it lowers the cost of producing the kind of first-draft intelligence that used to require platoons of young analysts.
That is not a cyclical disruption. It is architectural.
The First Compression: Coordination Is Losing Its Scarcity Premium
For decades, one of consulting’s most defensible forms of value was not merely expertise. It was coordinated expertise. Large firms could gather specialists from different disciplines, offices, and geographies, align them under a single brand, and deliver something a fragmented market of independents could not. The client was not simply buying answers. The client was buying orchestration.
This mattered because coordination is hard.
Organizations are full of hidden friction. People do not use the same language. Teams do not frame problems the same way. Data sits in different systems. One office knows one methodology, another office knows another. Someone has to reconcile definitions, normalize formats, align timelines, and present a coherent output. That work has always been enormously valuable because it is the difference between intelligence and institutional paralysis.
But AI reduces the price of that reconciliation layer.
It does not eliminate the need for judgment, politics, or trust. It does something subtler and more destabilizing. It takes many of the translation costs out of the system. It summarizes. It standardizes. It maps one vocabulary onto another. It detects inconsistency. It generates comparable drafts. It makes dispersed teams easier to integrate because much of what kept them apart was linguistic and procedural rather than truly intellectual.
This is the underappreciated part. When people say AI helps with productivity, they often mean it helps an individual work faster. That is true, but the more important effect may be that it helps institutions cohere faster. A firm does not need to be as locally idiosyncratic when its knowledge assets, workflows, and interpretive layers can be continuously harmonized by machines.
That strikes directly at the old premium attached to massive federated organizations.
The historical advantage of the global consulting partnership was that it could do something no single office, boutique, or independent expert could do alone: mobilize the network. But if AI lowers the coordination cost of the network, then the network itself becomes less rare as a source of value. The client no longer needs to pay as much simply because a large number of smart people can be made to sing from the same sheet of music. The sheet music itself can now be generated, translated, updated, and enforced at much lower cost.
In other words, the firm used to be valuable because it could assemble a temporary intelligence. AI increasingly allows intelligence to be assembled on demand.
That does not mean the large firms disappear. It means their center of gravity changes. Their value can no longer rest as heavily on their ability to coordinate human specialists across silos, because the cost of that coordination is falling. Their new challenge is to become platforms of applied judgment rather than networks of managed fragmentation.
That is a very different identity.
The Second Compression: Junior Labor Is Losing Its Margin Structure
The second attack is even more uncomfortable because it reaches into the internal economics of consulting itself.
The great engine of traditional consulting has always been leverage. A relatively small number of senior people sell, frame, and defend the work. A much larger number of junior people research, synthesize, model, deck-build, benchmark, document, and prepare the analysis. The business works because the junior layer can be deployed at scale, managed tightly, and billed upward through a branded hierarchy.
This is not a cynical observation. It is simply how the machine works.
Junior labor has historically done more than one thing at once. It has produced billable output, but it has also served as the apprenticeship layer through which firms manufacture future partners and principals. The same analyst who spends late nights cleaning data and polishing slides is also being socialized into the language, standards, and instincts of the firm. So the junior labor pool is not just labor. It is production capacity and cultural reproduction wrapped into one.
AI begins to split those apart.
Much of the output once created by junior staff is now vulnerable to synthetic generation. Research summaries, first-pass market maps, benchmarking tables, draft memos, deck structures, meeting notes, issue trees, variance explanations, and even parts of quantitative analysis can now be produced in minutes rather than days. Not perfectly, and not autonomously in every case, but sufficiently well to change the economics of staffing.
That phrase matters: sufficiently well.
AI does not have to outperform the best associate on their best day to be disruptive. It only has to perform the median first pass at a cost and speed that make the old staffing pattern hard to justify. Once that happens, the firm starts to ask a dangerous question: how much of our junior pyramid exists because it is strategically necessary, and how much exists because it was historically convenient?
That question cuts to the bone.
The old model depended on the fact that junior cognition was expensive enough to bill and abundant enough to scale. AI changes that equation by making certain forms of cognition both cheaper and more elastic. The first draft becomes abundant. The basic analysis becomes abundant. The presentational polish becomes abundant. Even the labor of “getting smart quickly” becomes abundant.
And when abundance enters a billing model designed around scarcity, margins move.
This does not mean young professionals become useless. Far from it. It means the basis on which they are economically valuable changes. The junior consultant of the old model was valuable because they could execute a large volume of disciplined mental labor under supervision. The junior consultant of the emerging model will be valuable to the extent that they can supervise synthetic output, detect subtle errors, ask better questions, manage ambiguity, and connect the output to client reality.
That sounds like an upgrade, but it is also a contraction. The number of people required to produce the same amount of first-pass work may fall. The demand curve for entry-level consulting labor may not disappear, but it may steepen sharply toward a smaller, more cognitively selective cohort. Less “smart and hardworking” may be needed. More “conceptually sharp, taste-driven, and judgment-capable” may be required.
That is not just automation. It is a redefinition of what the first rung is for.
My View: Consulting Will Survive, But Its Theater Will Change
My view is that consulting is not dying. Its theater is changing.
Clients still want trust. They still want accountability. They still want someone to say, “This is the right decision, here is why, and we will stand behind it.” AI does not naturally provide that social function. It can inform judgment, scale judgment, and sometimes imitate judgment, but it does not yet occupy the same status position in the client relationship. When a board, CEO, or public-sector leader hires a consulting firm, they are not only buying analysis. They are buying borrowed confidence.
That part remains.
What changes is the machinery underneath. The visible consulting experience may still feel human, strategic, and bespoke. But beneath the surface, the production engine will become much more standardized, much more centralized, and much more machine-assisted. The human brand will remain at the front end while the cognitive factory behind it becomes increasingly synthetic.
So the future firm is likely to look less like a labor pyramid and more like a judgment layer sitting on top of an intelligence substrate.
That distinction matters.
The firms that win will not simply be the firms with the most AI tools. They will be the firms that understand what AI does to the value chain. Coordination becomes less scarce. Junior output becomes less scarce. Therefore the premium migrates elsewhere. It migrates to trust, framing, interpretation, political navigation, institutional memory, problem selection, and the capacity to convert abundant analysis into decisive action.
That is where price will hold.
Everything that can be made abundant by AI will tend to lose pricing power over time. Everything that remains socially rare, organizationally difficult, or politically consequential will retain it longer. Consulting firms that mistake AI for a back-office efficiency initiative will miss the magnitude of the change. AI is not just making consultants faster. It is changing which parts of consulting deserve to be expensive.
The Hidden Threat to the Brand
There is also a quieter risk here. The old consulting model hid a great deal of complexity inside the firm. The client saw the polished output, not the messy internal effort required to create it. AI makes some of that effort dramatically easier, which is wonderful operationally but dangerous symbolically.
Why?
Because when the client starts to sense that parts of the work are being generated rather than painstakingly handcrafted, the mythology of the fee comes under pressure. A ten-week engagement billed at traditional rates feels different when the client suspects that the first eighty percent of the analytical scaffolding may have been built in a fraction of the time.
This is where consulting firms will face a legitimacy problem, not just an efficiency problem.
If AI lowers the cost of production but the client still experiences the old pricing model, firms will need a stronger explanation of what exactly they are charging for. The answer cannot simply be “the work.” Increasingly, the answer will have to be “the judgment.” The confidence. The context. The responsibility. The adaptation of generic intelligence to a specific institutional reality.
That is a credible answer. But it is a different answer than before.
The Real Divide Ahead
The real divide in consulting will not be between firms that use AI and firms that do not. That is too shallow. The real divide will be between firms that still think they are selling coordinated human effort and firms that understand they are now selling trusted judgment on top of machine-amplified cognition.
Those are not the same business.
One is still organized around labor management. The other is organized around decision support at scale.
One still believes the moat is the size of the team and the reach of the network. The other understands that network reach matters less when intelligence can be standardized, translated, and distributed more fluidly.
One still monetizes the difficulty of producing analysis. The other monetizes the difficulty of knowing what to do with analysis once it is cheap.
That is where I think this is headed. AI is not merely making consulting more efficient. It is exposing that much of consulting’s historical value came from managing scarcity that is no longer as scarce. Coordination used to be hard enough to bill at a premium. Junior cognition used to be scarce enough to bill through a pyramid. AI attacks both assumptions at once.
That is why this moment feels so consequential.
The firms that understand it early will not just cut costs. They will redesign themselves around a new truth: when intelligence becomes abundant, the only durable premium is knowing where it matters, how it should be used, and being willing to take responsibility for the answer.