Your Next Competitor Has No Employees

A guy I know who runs a small marketing shop lost a retainer this spring to a company he had never heard of. He went looking for them afterward, the way you do, half curious and half annoyed. No office. No team page. One person, a clean website, and a client list that looked a lot like his.

He kept circling back to the same question. How are they charging that?

They were not better than him. They just were not carrying what he was carrying.

The number nobody was expecting

Citadel Securities published a research note this month called AI-merican Exceptionalism. Sitting inside it is a Census Bureau figure worth a minute of your time: Americans filed 5.6 million business applications in 2025. That is up 24% since ChatGPT launched, and the strongest growth is showing up in exactly the sectors most exposed to AI.

The caveat first, because it matters. A business application is an EIN filing, not a business. The Census Bureau only classifies roughly three in ten as high-propensity, meaning they show real signs of turning into something with payroll. Plenty of those 5.6 million are side projects, LLCs opened for a tax reason, and ideas that die quietly in somebody's notes app.

But the direction is not ambiguous. More people are starting things. And they are starting them where AI does the most work.

What actually changed

The phrase from that note I keep coming back to is minimum efficient scale. It sounds like something from a business school lecture. It is not.

Minimum efficient scale is the revenue level at which hiring a specialist starts to make sense. It is a line every owner has crossed without naming it. Somewhere around a certain number, you stopped doing your own books. Somewhere later, you stopped writing every proposal yourself. Somewhere after that, you hired a person whose whole job was answering the phone.

That line moved. Citadel's macro strategist put it this way: AI lets a business access scale economics before it has actually scaled, pooling capabilities across functions and lowering the output threshold at which expansion becomes economic. What that means on the ground is that a founder who used to need four separate people for bookkeeping, marketing, service, and compliance can now spread all four across a much smaller team.

So businesses that never would have penciled out are penciling out now. Not because the work got easier. Because the overhead required to do the work got smaller.

The version of this story you should not believe

There is a story going around right now about a founder who built a company doing hundreds of millions in sales with essentially no employees. Just AI tools and his brother. It gets shared as proof that the future landed early.

What does not usually travel with it: an FDA warning letter, allegations about how the ads were made, and a class action. I am not telling you that to score a point. I am telling you because someone is going to send you that story as a reason to change how you run your company, and the story has holes in it.

The honest read is narrower and a lot more useful. AI did not make it possible to run a billion-dollar company by yourself. It made it possible to run a real, small, profitable company with fewer people than that same company would have needed in 2019.

That is a far less exciting sentence. It is also the one that is going to show up in your market.

What this looks like from your side of the desk

Two things worth doing, and neither one is panic.

The first is to look at your own cost structure, and not to cut it. Just to notice it. Every business that has been running for a while is carrying line items that exist because, at the time they were added, there was genuinely no other way to get that job done. Some of those are still true. Some of them stopped being true about eighteen months ago and nobody went back to check. The new shop underbidding you is not smarter. They just built their cost structure after the tools changed, and you built yours before. That is the whole gap.

The second is to be honest about where you are actually spending the time you free up. If AI clears six hours a week and those six hours go into more email, you have not changed your position. You have just moved the pile.

The part that did not get cheaper

Here is the other finding in Citadel's work, and it is the one nobody quotes. Across S&P 500 earnings calls, companies describing AI as a complement to hiring outnumber the ones describing it as a substitute, in nearly every job function they looked at. The headcount is not being deleted. It is being deferred, until the business hits the point where human judgment becomes the binding constraint.

That last phrase is your entire business.

Human judgment is the binding constraint. It is knowing what a job actually costs before you quote it, because you have been burned on that exact job. It is the subcontractor who picks up at six on a Friday because it is you calling. It is the customer who refers you without being asked, who has never once looked at your website. It is the estimate you walked away from because something felt wrong, and you were right.

None of that has a minimum efficient scale. None of it got 24% cheaper since ChatGPT launched. It is the one line on your balance sheet that a two-person shop with a great AI stack cannot download.

The one-person competitor is real, and they will beat you on price for the commodity work. Let them have some of it. Then go be the thing they cannot be yet, which is a business that has already made every expensive mistake once.

The bar for starting a company dropped. The bar for lasting did not move an inch.

If you are looking at your own overhead wondering which parts of it the last two years quietly made optional, that is a good conversation to have with someone who does this every day. That is the work we do at Nexera Intelligence.