By Bryan J. Kaus
The limit to the size of the firm is set where its costs of organizing a transaction become equal to the cost of carrying it out through the market.
Ronald Coase
The Census Bureau counted 531,728 business applications in the United States in August.
That number caught my attention, though not because I think 531,728 new companies suddenly opened their doors. They didn’t.
The Bureau is careful about the distinction. A business application signals that someone may be preparing to conduct business. Separately, Census estimates how many of those applications are likely to become businesses with payroll tax liabilities. For the August cohort, that estimate is 28,501 within four quarters.
Those two numbers measure different things. Divide one by the other, call the result a national startup success rate, and you have a good social-media graphic and bad analysis.
But there is a useful truth inside the distinction.
Starting something and building something are not the same thing.
I have been thinking about that a lot lately because I am living some version of it myself.
Part of my time goes to advising businesses, studying companies and capital allocation, and talking with management teams about growth, integration and the ordinary operating problems that never make the press release. The other part goes to actually building things: an advisory firm, research systems, publishing platforms, and a few commercial ideas that began the way most businesses do, with somebody staring at a blank page and deciding whether there is enough there to make something real.
The difference between doing that today and doing it twenty years ago is remarkable.
I can sit at a desk on a Sunday morning and acquire capabilities that once required a surprisingly large organization. Research a market. Build a financial model. Design a website. Draft an agreement. Analyze competitors. Set up payments. Stand up a CRM, produce the sales material, test the idea and start reaching customers before the coffee is cold.
Some of that is AI.
A lot of it isn’t.
Cloud computing replaced the server room. Software-as-a-service replaced installed systems and much of the infrastructure required to run them. Shopify and Amazon collapsed the cost of distribution. Stripe simplified payments. Canva democratized design. LinkedIn changed prospecting. Contract manufacturers, outsourced logistics and fractional professional services let a company rent capabilities that once had to sit inside it.
Artificial intelligence is simply pushing the same logic further into cognitive work.
The result is extraordinary.
The cost of acquiring the capabilities of a company has fallen dramatically.
The cost of building a good one has not.
The Company You No Longer Have to Own
Ronald Coase asked a deceptively simple question nearly a century ago: if markets are so good at coordinating economic activity, why do firms exist at all?
His answer was that markets have costs.
Someone has to find the supplier, negotiate the terms, write the contract, monitor performance, coordinate timing and sort out the problems when the thing that arrives is not quite the thing that was ordered. Sometimes it is cheaper and more reliable to put those capabilities inside the company than to buy them, again and again, from outside.
That idea still holds.
What has changed is the arithmetic.
Technology has been lowering the cost of coordinating outside capability for decades. It is easier now to find expertise, evaluate suppliers, communicate, monitor work and buy specialized services without putting all of those people on one payroll.
So the efficient boundary of the firm can move inward.
From a capital-allocation seat, that can be a very good thing.
The purpose of a business is not to accumulate employees or own every capability it uses. Size has never been evidence of quality.
A company that turns fixed cost into variable cost, preserves flexibility and reaches excellent outside capability without giving up quality may build a better economic model than one that owns everything because that was simply how companies used to be organized.
Return on invested capital can improve. Break-even points can fall. The business keeps its optionality, and management can point scarce capital at the few things that actually create advantage instead of rebuilding commodity functions in-house.
That is the financial case for the smaller modern firm.
It is a good one.
But it hides a trap.
Renting capabilities is not the same thing as building capability. Having access to the functions of an enterprise is not the same thing as creating one.
The website can be excellent. The analysis can be sophisticated. The software can work. The deck can look like it came from a company with three floors in a downtown tower.
Then Monday morning arrives, and someone still has to get a customer.
Tuesday morning arrives, and someone has to deliver what was promised.
The invoice has to be collected. The next customer has to show up. Quality has to hold. The economics have to survive taxes, slow stretches, working capital, a few bad decisions and the occasional customer who treats “net 30” as a philosophical suggestion rather than a payment term.
If the business is going to grow beyond the founder, eventually somebody else has to know what good looks like.
That is when it starts to get interesting.
The Bottleneck Moves
Markets teach the same lesson over and over: when something becomes abundant, it stops being a source of advantage.
There was a time when simply having a professional website meant something. Good design cost real money. Sophisticated analysis required specialized people. Software required engineers. Distribution required physical infrastructure. Even access to information could be an edge.
Those things still matter.
They just tell you less than they used to about the company behind them.
A beautiful website can sit in front of a wonderful business or a terrible one. The website does not know the difference.
The same is becoming true of competent writing, ordinary software, market analysis, polished presentations and a long list of administrative work.
When the inputs become abundant, scarcity moves.
It moves toward judgment: which problem is actually worth solving.
Toward trust: whether a customer will put money, reputation or career risk in our hands.
Toward execution: whether we can do tomorrow what we promised yesterday.
Toward capital allocation: which opportunity earns another dollar, another hire or another month of management attention.
And toward leadership: whether capable people can make good decisions without routing every one of them back through a single person.
Those have always been scarce capabilities.
Technology is simply making them easier to see.
We have democratized the tools of entrepreneurship far faster than the disciplines of enterprise.
I Know the Temptation
This is not a criticism I am making from the outside.
I understand the temptation because I am in it.
The tools let me do things myself that I would once have handed straight to someone else. Often that is the right answer.
Why add permanent cost when technology can do the job well?
Why build an internal capability before demand justifies it?
Why hire ahead of the economics because an org chart says a company of a certain size ought to have that function?
I am biased toward keeping fixed costs low until the economics prove they deserve to become fixed.
Capital should earn its place.
So should organizational complexity.
But there is another side to that discipline.
The ability to do something yourself can become an excuse never to build the organization that could do it without you.
That is its own kind of capital-allocation mistake.
The scarce capital in a young business is not only cash.
It is management attention.
When every meaningful decision, every customer relationship, every pricing question, every quality issue and every exception ends up on the founder’s desk, the company has concentrated its most important operating risk in its most expensive and least replaceable resource.
The founder.
AI can make that founder extraordinarily productive.
That does not solve the problem.
It can hide it.
A highly capable person surrounded by increasingly capable technology can produce the output of an organization while remaining the nerve center through which everything important still passes.
Every difficult sale comes back to the founder.
Every strange customer issue comes back to the founder.
Every real judgment comes back to the founder.
Every new opportunity competes for the same calendar.
The business gets more productive.
The founder gets busier.
And the enterprise quietly grows more dependent on one person, not less.
There is nothing wrong with that, as long as you know what you have built.
The danger for the founder may not be that AI replaces him.
It may be that AI makes him so productive he never has to replace himself.
Business Is Not a Headcount Contest
This is where the argument needs some discipline.
America has nearly 30 million businesses without paid employees. They generated about $1.7 trillion in receipts in the Census Bureau’s 2022 data.
Those are not pretend businesses.
A consultant with no employees can have a very good one. So can a carpenter, chef, designer, writer, tradesperson or adviser who deliberately organizes around personal skill and reputation.
There is dignity and real economic value in that.
Scale is not virtue.
Neither is complexity.
I have seen enough large companies to know that adding people can just as easily add meetings, handoffs, misaligned incentives and fresh places for accountability to disappear.
The question that matters is what the owner is trying to build.
If you want a craft practice that produces an excellent income and gives you control over your life, build that.
If you want a business whose distinguishing asset is your own skill, protect that.
But if you want a company that eventually outgrows your own capacity, the test changes.
It stops being:
Can I do this?
And becomes:
Can the organization do this?
That transition is harder than forming the LLC, standing up the website or shipping the first product.
It is also where much of the value is created.
From Founder Leverage to Enterprise Leverage
I think about this as the difference between founder leverage and enterprise leverage.
Founder leverage makes one person dramatically more productive.
That is what the new tools do exceptionally well.
Enterprise leverage lets the organization create value without that person’s continuous, direct intervention.
Eventually you need both.
A useful test starts with demand.
Would customers still arrive if the founder stopped personally creating every opportunity?
Then delivery.
Can someone else, or some system, produce work at a standard the founder is willing to put his name behind?
Then economics.
Does the business throw off attractive cash after paying for the people and systems required to run it properly?
That one matters more than it looks.
Apparent margins built on unpaid founder labor are not business margins.
They are compensation the founder has not yet accounted for.
Then judgment.
Can people make ordinary but important decisions without turning the founder into an approval queue?
Then institutional memory.
When the company learns something expensive, does the lesson become part of how the company operates, or does it stay trapped in one person’s head?
And finally the simplest test of all:
What happens if the founder disappears for sixty days?
Not permanently.
No dramatic succession event.
Take a long trip. Get sick. Spend two months buried in a book or camped at a customer’s site. Pick your reason.
Do customers still get served?
Does cash still get collected?
Can suppliers resolve a problem?
Does anyone understand the economics?
Can people tell an exception worth escalating from one they should simply handle?
Does the company slow down?
Or does the machinery stop?
That answer tells you something headcount cannot.
It tells you whether you have built leverage around the founder or leverage beyond the founder.
Good Leadership Changes the Arithmetic
This is also where entrepreneurship becomes a leadership problem.
Founders get rewarded early for exactly the behavior that constrains them later.
Move fast. Solve everything. Know the customer better than anyone. Catch the mistake. Make the sale. Jump into the problem. Protect the standard.
Indispensable when the business is small.
A ceiling when it is not.
The fix is not simply delegation.
I have always thought “just delegate more” misses most of the real issue.
Delegation without standards is abandonment.
Delegation without information is guessing.
Delegation without authority is theater.
Delegation without trust just means the founder redoes the work afterward.
Good leadership builds the conditions under which judgment can move outward.
People need to know what matters. They need enough context to make intelligent trade-offs. They need to know where their authority begins and ends. They need to know what a good outcome looks like.
And they need to believe that an honest mistake made while exercising sound judgment will be treated differently from carelessness.
That takes time.
There is no prompt for it.
Technology can help document the process.
Leadership is what turns the process into an institution.
The Financial Question Beneath the Organizational One
There is a financial reason to care about all of this.
Businesses are assets, and the quality of an asset depends in large part on how durable and predictable its cash flows are.
If revenue disappears when one person steps away, the cash flow carries key-person risk.
If every incremental dollar of revenue demands another equivalent hour of founder labor, scalability is limited.
If quality erodes every time volume rises, growth destroys value instead of creating it.
If the business needs more permanent overhead than its economics can support, scale becomes a liability.
And if the founder adds people mainly to look larger, fixed cost eats the very optionality that made the company attractive.
So the objective is not to build a big company.
It is to build the smallest organization capable of reliably creating the value you intend to create.
Own what differentiates you.
Rent what does not.
Automate what should be automated.
Standardize what should be repeatable.
Keep judgment close enough to the work to stay intelligent, and distribute enough of it that one person never becomes the operating system.
That is good leadership.
It is also good finance.
Technology May Help Build the Institution Too
There is an obvious objection to all of this.
The same technology making it easier to start a business may also make it easier to build the systems that turn one into an enterprise.
I think that is probably true.
AI can help document processes, train people, preserve institutional knowledge, monitor quality, automate the back office and put specialized expertise within reach of a small team.
Organizations that once needed a bureaucracy to do those things may soon do them with a handful of people.
That is exciting.
The company of the future may employ fewer people, own fewer support functions and rent far more capability than the twentieth-century corporation did.
That does not make it less of an enterprise.
It changes what the enterprise has to own.
The most valuable thing inside the company may become less the capability itself than the judgment required to coordinate it.
The system can generate the analysis; someone decides whether it makes sense.
The software can recommend a price; someone understands the relationship with the customer.
The model can produce a staffing plan; someone remains accountable for what kind of institution results.
Technology can distribute information. It can support judgment. It can automate more of the routine every year.
It cannot remove responsibility.
That part stays ours.
The Point Taken
This is an extraordinary moment to start something.
That deserves to be said plainly.
People can reach capabilities that used to require serious capital, a payroll and real infrastructure. The barriers to experimentation are lower than they have ever been, and the ability of one talented person to turn knowledge into economic activity has probably never been greater.
I am benefiting from it too.
But building a business is still different from assembling its parts.
Customers still have to trust you.
Promises still have to be kept.
Cash still has to be managed.
People still have to know what good looks like.
Capital still has to be allocated.
Someone still has to decide when the facts are incomplete.
And if what you want is an enterprise rather than a more productive version of yourself, the organization eventually has to carry weight without your hands underneath every piece of it.
Technology is moving the boundary of the firm.
It may sharply reduce the amount of organization required to create value.
That is progress.
But the disciplines that let a business endure stay stubbornly familiar: trust, judgment, execution, capital allocation, and the willingness to build a system strong enough to run when the founder is no longer standing in the middle of it.
Starting is increasingly a transaction. Building is still a discipline.
© 2026 23.5 Strategies; The Point Taken™