The CEO Scale Problem

I honestly didn’t mean for this to be a political post.

There’s a question I ask founders when they’re navigating a difficult hire, a reorg, or a leadership-team conflict: Is your company smarter than you?

Not “do you have smart people?” Every founder says yes to that. I mean: do you believe the organization, as a system, can perceive reality more accurately than you can alone?

That answer tells me a lot. The CEO’s job changes when the company starts to know things the CEO does not. Early on, the founder is often the company’s best source of judgment. They have the most customer context, the clearest product intuition, and the deepest sense of urgency. But as the company grows, knowledge spreads. Sales knows things product does not. Finance sees risks the founder cannot feel. Legal, compliance, research, security, and people teams start solving problems the founder cannot see from the CEO’s desk.

The founder’s job is no longer to be the company’s intelligence. It is to build an organization whose intelligence exceeds their own. This transition is difficult because it requires shifting between two very different leadership modes.

Operator mode

Operator-CEOs are the company’s brain, and employees are the hands. They’re expected to be deeply involved in strategy, to hold the anxiety for all outcomes, to be the most central person in almost every important decision. They delegate execution, not thinking.

This model works beautifully at a certain scale. The early-stage startup is essentially designed for it. When you’re ten people trying to figure out product-market fit, you want a founder who is in every customer conversation, reviewing every design decision, opinionated about everything. Intense involvement is a feature, not a bug. Speed, coherence, and conviction are what you need, and one brain calling the shots provides exactly that.

The fatal flaw of operator mode, though, is a specific kind of blind spot. As companies grow, more and more mission-critical work becomes invisible to the founder.

Because an operator cannot personally evaluate the quality of work in specialized areas like finance or legal, they struggle to distinguish between a function that’s genuinely bureaucratic and one that’s doing vital work outside their line of sight. They tend to misread “I can’t evaluate this” as “this is probably unimportant.”

You see it in the small moments. The founder rolls their eyes at the legal review. The new VP of Finance keeps raising risks that feel like “process.” The head of People says the hiring loop is broken, and everyone nods, and then nothing changes because the CEO can’t quite tell whether this is real work or HR theater.

The result is underinvestment in exactly the capabilities the company needs as it scales. Beyond that, the operator is a bottleneck by design. Every significant decision goes through one person, so the organization can only move as fast as that person can. Strong employees figure this out quickly and quit (they want to think, not just execute). Over time, an organization stuck in operator mode ends up with a team optimized for loyalty and execution rather than judgment. (If this sounds familiar, hold that thought.)

Orchestrator mode

The orchestrator accepts, as a baseline, that the organization knows things they don’t. Their job is to set direction, create the conditions for good decisions to happen without them in the room, and hold the whole thing together when those decisions start to conflict.

But orchestration is not abdication. They build the machinery by which judgment improves: clear strategy, explicit decision rights, strong executives, fast feedback loops, and enough direct contact with reality that they can still tell when the system is fooling itself.

Satya Nadella is the obvious contemporary example. He came in with deep technical and organizational credibility, especially from Microsoft’s cloud business, but he did not try to be the smartest person in every Microsoft business. His achievement was not personally out-thinking every division; it was making those divisions add up to something coherent.

The orchestrator has their own failure modes. Strategy can become a collage - the sum of what smart people in each domain want to do - rather than genuine direction. The leader can lose ground truth, becoming so abstracted from execution that they can’t tell when the organization is drifting or, worse, lying to them. And accountability diffuses: when everyone owns a piece of a decision, hard tradeoffs have a way of falling through the cracks (and usually right onto the plates of incredibly frustrated middle managers).

John Sculley at Apple is the classic example of that failure mode. He brought professional management discipline to early Apple, but lacked the unifying product judgment to force hard choices. Under his tenure, Apple’s strategy became a sprawling collage of overlapping product lines designed to appease internal divisions rather than serve a clear vision. He built the machine, but forgot to drive it.

The startup fork

Successful founders are almost always operators, and they should be. The hard part is that the same founder behavior can be exactly right at one stage and exactly wrong at the next. At ten employees, being in every important decision creates coherence. At a hundred, it teaches the company not to think until you enter the room.

If a company is successful, it eventually creates more context than one founder can hold. At some point, a founder hits an inflection point where they have a choice to make - often without fully realizing they’re making it (usually right after a great VP rage-quits).

One option is to become an orchestrator: to genuinely change how you conceive of your role, to hire people who will push back on you, to accept that important decisions are being made every day of which you’re not aware, and to recognize that this is correct. This is hard. It requires a kind of ego flexibility that founders are not typically selected for. It means sitting in a product review, watching a team make a choice you wouldn’t make, and keeping your mouth shut - not because you suddenly agree, but because stepping in to “fix” it would teach them to stop thinking for themselves. (Finding a good coach can also help.)

The other option is to restructure the company to stay small enough for the operator model (aka, “founder mode”). One way to read Brian Chesky’s recent Airbnb reorganization is as a deliberate choice to preserve more operator leverage at scale. After a period of growth into genuine organizational complexity, Chesky reorganized back into something tighter - something he could run the way he ran it when it was small.

When is this a legitimate strategic choice versus a refusal to adapt? Well, product-market fit buys time (though it doesn’t repeal complexity). With extreme PMF, an undisputed market leader can afford the bottlenecks of an overgrown operator because the product’s momentum covers for organizational drag. Given Airbnb’s market dominance, Chesky had the luxury to make that choice.

Steve Jobs is perhaps the best historical example of a founder who successfully built a massive company around his operator mode. When he returned to Apple, he famously slashed the product line to a simple grid - cleaning up the exact collage Sculley had created. He actively constrained the company’s surface area so his operator intuition could still touch everything that mattered. Crucially, though, in the one critical area he knew he couldn’t add value (global manufacturing and supply chain), he handed it to Tim Cook and let him build a machine. Jobs found a third mode: stay close to the parts where your judgment is unusually good, and get out of the way where it isn’t.

Contrast that with Travis Kalanick at Uber. Kalanick had brilliant product and market intuition, but he is the clearest example of the operator’s blind spot. He devalued the specialized functions he couldn’t personally evaluate - HR, legal, compliance, corporate culture. Under his leadership, “I don’t understand why this matters” became functionally identical to “this does not matter.” Uber didn’t implode because the product lacked PMF. It imploded because PMF created a bigger, more dangerous company than Kalanick’s operating model could safely contain.

The scale problem

This is not only a startup problem. The bigger and older the institution, the less forgiving this gets.

The United States federal government is not a startup (tech-exec posts on X notwithstanding). It manages massive complexity across domains where expertise takes decades to develop. Yet the current administration looks a lot like operator logic pushed past its useful range: a small circle of trusted executors is being stacked with more and more responsibilities, not because the work consolidates naturally but because the operator needs people he trusts rather than people with depth. Rubio has been asked to serve simultaneously as Secretary of State and National Security Adviser, while Bill Pulte has been moved into the acting Director of National Intelligence role while retaining his existing housing-finance responsibilities.

This is not just a style preference. It’s a scale error. A way of working that can be useful when the system is small becomes dangerous when the system is too large for any one person, or any small circle, to understand. Jobs proved that an operator can only scale if they aggressively constrain the organization’s scope and selectively orchestrate the domains they don’t understand. This administration is doing the opposite: applying operator logic to an unconstrained organization, while replacing people with domain depth with people whose main qualification is trust.

Operator logic doesn’t build depth. It builds loyalty. And an organization run on loyalty rather than expertise is one that devalues what it doesn’t understand.

The irony is that this is the same trap that catches founders. They don’t fail because they’re not smart or not hardworking. They fail because they’re applying a model that was right for one stage to a context it was never designed for.

The question worth asking

So back to the opening question: does your company know things you don’t?

If you’re a founder, a few years in, and the honest answer is still no, that’s worth sitting with. Either your company hasn’t grown yet, or you’ve built an organization still only as capable as you. That’s a ceiling, not a strategy.

The goal is not to become an orchestrator because orchestration is morally superior. Some businesses can and should remain tightly founder-led. But that should be a conscious choice, not a reflex. If the company is only as capable as the founder, then the founder has not built an organization. They have built a dependency.

PS

One reason this transition is so hard: some leaders come out of apprenticeship cultures (like medicine, banking, or family businesses) where leadership mostly means being better at the same work. If your whole career taught you that seniority means doing the same craft at a higher level, it’s easy to assume the people below you are just less-developed versions of you. Then you enter a genuinely complex organization, where specialized teams do work that is entirely invisible to you, and the old model breaks. You don’t just struggle to orchestrate. You may not even realize orchestration is the job.