People are still talking about AI as if the future will split neatly between adopters and laggards.
It won’t.
The real split will be between organisations that build intelligence into how they operate and those that simply lease someone else’s cognition layer.
One group is building intelligence into the core of how it thinks, decides, operates, and compounds. The other is subscribing to someone else’s cognition layer, wrapping process around it, and calling that transformation. Those are not the same thing. One is strength. The other is dependency with better branding.
The New Galt Question
Who is John Galt? (Atlas Shrugged)
Most people read it in the construct of the time of writing: the strong leave, the capable disappear, the builders withdraw and let the broken world collapse under the weight of its own mediocrity.
The more apt interpretation for the AI age is less theatrical. The modern Galt does not need to vanish. It is enough to stop giving your best thinking to systems that have chosen the path of least resistance.
The strong performers aren’t always the ones who storm out. Often they are the one’s who stay just long enough to realise the organisation around them no longer wants real judgement, real challenge, or real capability. The organisation wants faster output, lower cost, cleaner decks, more automation, less friction, and above all, less resistance, it wants the appearance of intelligence without the inconvenience of thought.
At that point, the withdrawal begins. Not always physically. More often cognitively, strategically, and economically. The people are still there. The badge still works. The calendar is still full. But the best thinking has already left the building.
That is the brain drain of the AI era. Not people leaving countries. Not talent changing companies. Capability withdrawing from organisations that have chosen synthetic competence over the real thing.
Bought Capability Is Not Capability
This is the mistake many companies are making now, and they are making it with enormous confidence. They think access is advantage.
They think because they can buy licenses, connect copilots, deploy assistants, build wrappers, and integrate model APIs, they have acquired intelligence.
They have not. They have acquired proximity to intelligence. That is different.
The problem is not using tools. Switching models, adopting new platforms, and moving between systems is fine. The intelligence, the playbooks, and the value can travel with you. The problem is renting a solution that gives the illusion of those things. That offers the appearance of a working intelligence layer without the underlying understanding, discipline, or design that makes it durable.
When the tool is doing the thinking and no one inside the organisation understands why, what the edge cases are, or what happens if the upstream provider changes the model, changes the pricing, or simply gets better at abstracting away the value the company thought it owned — then the organisation is not intelligent. It is dependent.
If your capability depends on a model you do not control, a stack you did not build, a reasoning layer you do not understand, and economics set by someone else, then your advantage is contingent by definition.
You are not sovereign. You are a tenant, and tenants do not own the upside. They rent it, briefly, until the market catches up and everyone else gets the same tools, the same outputs, the same templates, and the same synthetic fluency. This is where the next collapse begins. Not because the tools are weak, but because the adoption is shallow.
Most companies will not build intelligence. They will procure it. They will bolt it onto weak operating models, feed it bad process, ask it to compensate for thin thinking, and then act surprised when the result is fast, polished, and fundamentally average.
Borrowed cognition still leaves you intellectually bankrupt.
Enterprise Tribalism
This is where enterprise tribalism starts to emerge. Not as culture. Not as branding. As structure.
Skilled operators and ambitious companies are beginning to cluster around the model ecosystems that matter — OpenAI-first environments, Nemoclaw operators , Claude-anchored stacks, closed proprietary systems, and verticalised AI built around specific infrastructure and data control. These are not just technology choices. They are economic and strategic alignments forming around different standards of reasoning, different tolerances for openness, and different views on what should be owned versus rented.
But the more important division is not which ecosystem a company backs. It is what they do inside it.
Some are building with real depth — building enduring context, shaping workflows, restructuring operating models, owning integration logic, retaining human judgement, and converting tool access into durable capability that would survive a model change, a pricing shift, or a platform pivot.
Others are decorating themselves with the logo of the empire they rent from.
That is not tribal strength. That is feudal allegiance.
A company built on bought capability is weaker than it looks. It may move quickly. It may sound current. But its intelligence is upstream. Its economics are upstream. Its constraints are upstream. Its future is upstream. It does not own the engine. It just sits in the carriage and calls itself modern. The important thing is not which banner a company flies. It is whether that ecosystem is helping it build internal strength or merely rent external fluency.
The Quiet Exit of the Strong
The strongest people see this before most boards do.
They see the slide from judgement to automation theatre. They see organisations confusing faster drafting with better thinking. They see leaders wanting AI because they think it will remove friction, when friction was often the point. They see institutions quietly optimising themselves out of depth. And many stop fighting it.
This is where the Galt comparison earns its place. Not because the best people are all disappearing into mountain compounds. Most are not. That is fantasy. But because many are making a quieter move: withdrawing from organisations that have unwittingly chosen obsolescence.
- They reduce exposure.
- They narrow contribution.
- They invest elsewhere.
- They build personal capability in private.
- They reserve real thought for places where it still matters.
From the outside, it can look like disengagement. It is refusal. A refusal to subsidise institutional decline with real human capability.
The Middle Gets Squeezed First
The companies most at risk are not the obvious laggards. It is the middle.
The companies that adopt enough AI to look modern but not enough to become structurally different. The ones that buy capability but never metabolise it. That use the language of transformation while preserving the instincts of bureaucracy. That mistake implementation for reinvention.
For a while they will look efficient — more output, faster turnaround, lower costs, brighter dashboards. And underneath it, their distinctiveness will rot.
Once capability is rented rather than built, every competitor with budget can buy the same lift. Once intelligence is upstream, differentiation collapses downstream. The layers that made them distinct — the analysts with process but not judgement, the managers whose value was coordination not clarity, the teams whose output was presentable but not defensible, the AI strategy that was a procurement programme with better adjectives — all of that becomes replaceable.
That is when markets stop rewarding participation and start rewarding ownership.
- Ownership of the model
- Ownership of the data
- Ownership of the workflow
- Ownership of the distribution
- Ownership of the intellectual discipline needed to direct all four
Everyone else converges into the same average.
The Coming Technocracy
If this keeps moving in the same direction, value will not vanish. It will narrow. Into the intellectual and technological capital of a select set of enterprises.
Power pooling around those who own the intelligence layer, or understand it deeply enough to bend it to their will. They will not just sell tools. They will define the grammar of work — deciding what gets abstracted, what gets priced, what gets automated, and what becomes economically invisible.
That is real power. Not the ability to use the machine. The ability to decide how everyone else should use it. And beneath that layer will sit thousands of companies still congratulating themselves for deploying features.
Final Word
The AI age will not simply separate adopters from non-adopters. It will separate builders from renters.
Some companies will build real internal capability on top of these systems and become harder, sharper, and more sovereign. Others will buy access to borrowed intelligence, mistake it for strength, and slowly discover they have become dependent on the very platforms they thought gave them an edge.
The same split will happen in people.
- Some will use AI to extend judgement.
- Some will use it to avoid thinking.
- Some will stay and fight for standards.
- Others will quietly withdraw from institutions that have chosen comfortable decline.
In Atlas Shrugged, John Galt asked what the world would look like if the people who actually drove it simply stopped. Not sabotage — withdrawal. That is the real question for the AI era. Not who disappeared. Who stopped giving their mind to systems that traded depth for synthetic fluency. When enough of them do, the market won’t fall apart. It will consolidate around those who never outsourced the thinking.
For more on artificial intelligence, autonomy, leverage, and the irreversible changes now reshaping work and modern life, explore the other posts, subscribe or contact me.
