top of page

The Minimum Viable Company Is Collapsing. Nobody's Talking About What Comes Next.

2 hours ago
11 min read
A builder at a wooden desk at dawn, laptop open to a glowing agent interface, looking out over a neighborhood of small workshops, storefronts and studios each tagged with a floating holographic service icon, city towers on the horizon

Everybody's talking about how many jobs AI is going to eliminate.

Fair enough. That's a big deal.


When a company discovers that five people with AI can accomplish what previously required twenty, the employment implications are pretty obvious. But there's another side of that equation that I think we're spectacularly underestimating.


If five people can now do the work of twenty, how many businesses that previously needed twenty people can suddenly exist with five? Or two? Or one?


That's not just a productivity improvement.

That's a fundamental change in the economics of starting and operating a business. And I think we're beginning to see it. Not in the way the coverage describes it, and not quite where the statistics are looking.


The Other Side of the Employment Story

The Wall Street Journal published a piece in November 2026 titled 20-Somethings Are Ditching Entry-Level Jobs to Become Their Own Bosses.


According to Babson College figures cited in the reporting, at least 23% of Americans ages 18 to 34 were operating a business or a side hustle in 2025, compared with 19% in 2021. Babson's Global Entrepreneurship Monitor data tells the same story from a different angle. Americans between 18 and 34 report the highest rate of early-stage entrepreneurial activity of any age group, and the 18-to-24 rate has roughly doubled in five years. That's interesting by itself. But the really interesting part is why this is happening.


We're watching a generation discover that the traditional sequence of education, entry-level employment, years of experience, and eventually maybe starting a business isn't necessarily the only path anymore. AI can help them design a product, write software, create marketing materials, develop a website, research customers, and automate operations. Things that previously required hiring specialists, learning several disciplines, or finding substantial startup capital.


And this isn't a claim about what AI might eventually do. In Babson's 2025–2026 GEM survey of 11,000 U.S. adults, 67% of entrepreneurs said AI tools are important to their business model and strategy.


Two thirds of the people running businesses, describing the thing they're already using, as important to how the business works. That's not a forecast. That's a headcount of what's already happening. I don't know where the ceiling is on any of this. Definitions matter, especially when we're counting side hustles alongside businesses, and "entrepreneur" has become a word people use to describe a lot of very different activity.

But I absolutely recognize the underlying transformation. Because I've been watching it from the builder's side.


The Collapse of the Specialization Tax

In The Three-Way Mexican Standoff That Nobody Wins — Except the Builder, published May 31, 2026, I wrote about the standoff between engineers, product managers, and designers. Everybody was convinced AI was making somebody else's job unnecessary.


The engineer didn't need the product manager. The product manager didn't need the engineer. The designer didn't need either one.


My conclusion wasn't that any of them had won. It was that the specialization tax was collapsing. For decades, building software required a collection of specialists because the knowledge and execution requirements were too substantial for one person to handle effectively.


Then you needed somebody to coordinate those specialists. Then somebody to manage the coordinators. And before long, you had an organization. Not necessarily because the product itself demanded a large organization, but because assembling the capabilities to create that product demanded one.


AI changes that calculation.


A single individual can increasingly perform or coordinate work across several disciplines, with AI filling gaps in execution. Not perfectly. Not without supervision. And not with the depth of an experienced specialist in every domain. But often well enough to build, test, and deliver something useful. The organizational overhead that used to be necessary just to get started is becoming optional for a growing class of businesses.

And that's an enormous economic development.


The Minimum Viable Market

Here's where I think this gets really interesting. We've spent years talking about the minimum viable product. What's the smallest useful thing we can build to test an idea? But I think AI introduces another concept worth examining.


The minimum viable market.


Imagine a specialized software product that serves 200 customers. Maybe it's for independent marine mechanics. Or a very specific category of medical office. Or a little workflow problem that only a few thousand businesses in the country even experience.


Traditionally, developing and maintaining that product might require enough engineering, support, marketing, and administrative expense that the potential revenue couldn't justify the investment. The market was simply too small. But what if one knowledgeable person can build it, maintain it, support its customers, and automate much of the business?


Suddenly, that tiny market might be economically viable. Not because the market got bigger. Because the cost of serving it got smaller.


Think about the implications. We could see thousands of highly specialized, modestly profitable businesses emerge in places where conventional startup economics never made sense. Not every entrepreneur needs to build a billion-dollar company. Somebody building a useful product that earns $120,000 a year with manageable expenses and very little organizational overhead might be operating a perfectly successful business.


No venture capital.

No massive staff.

No obsession with becoming the next unicorn.


Just somebody who understands a problem and can now afford to solve it.

The funding data backs up the shape of this. GEM found that 55% of American entrepreneurs are financing themselves out of personal savings. That fits the model exactly, because the business has to be small enough to start with what's already in your account. Of course, lower costs don't create customers out of thin air. Somebody still has to want the thing enough to pay for it. But the threshold for testing that question is falling. And that changes what kinds of markets are worth exploring.


It's Not Just the Kids

I also think the generational framing misses half the opportunity.

Sure, younger people who grew up with digital tools may have an advantage in adopting AI quickly. But there's another group sitting on an extraordinary amount of unrealized economic potential. People with twenty, thirty, or forty years of experience doing something. The mechanic who can diagnose a problem by listening to an engine. The accountant who knows exactly where a business is leaking money. The sales professional who understands a particular industry's customers better than anybody.

The retired engineer who has forgotten more about a specialized manufacturing process than most people will ever learn.


I've written about this before in Tribal Knowledge as Capital (And Why Experience Is the Next Frontier of AI). These people possess something increasingly valuable: accumulated judgment. They know which problems are worth solving. What they often haven't had is the surrounding technical and organizational capability to turn that expertise into a scalable product or service.


Now AI can help supply some of those missing capabilities. And here's a thought.


Imagine pairing the AI fluency of a twenty-three-year-old with the domain expertise of somebody who's spent thirty years in an industry.

One knows how to make things happen with the tools. The other knows what actually needs to happen. That could be a far more powerful entrepreneurial combination than either person operating alone.


More Companies Doesn't Automatically Mean More Success

Now, there's an important distinction here. The cost of creating a business is not the same as the cost of sustaining one.


The Census Bureau's Business Formation Statistics release on November 12, 2026 recorded 501,609 seasonally adjusted business applications for October. Its model projected that 27,867 of them would turn into employer businesses, meaning businesses with payroll, within four quarters.


That's about one in eighteen. And the direction of the most recent month is worth noticing.


Applications were down 6% from September. High-propensity applications, the ones the Bureau considers most likely to become real employers, were down 2%.


Then there's the longer arc, which is the part I keep coming back to.

In 2005, the Bureau recorded about 2.5 million business applications, and roughly 427,000 employer businesses formed from the cohorts that filed them. That's about one employer for every six applications.


In 2025, the country filed about 5.7 million applications. Those cohorts produced about 297,000 employer businesses, with the most recent months still partly projected. Roughly one in nineteen.


Read that comparison again, because it's the entire argument of this piece in one line.


The number of people filing to start a business has more than doubled in twenty years. The number of businesses that actually hire anybody has gone down.


Not flattened. Down.


The caveats are real and I'll state them plainly. An EIN application costs nothing, and people file them for reasons that have nothing to do with building a company: consulting gigs, rental properties, single-member LLCs that will never have a second employee. That's why the ratio has been sliding for two decades, and why nobody should read it as "most businesses fail." But the slide isn't noise. It's a signal.


The country has gotten dramatically better at creating business entities and no better at all at creating employers, and the gap between those two numbers has been widening since long before ChatGPT existed.


Which brings me to a piece of research that cuts the other way.

A May 11, 2026 preprint by Hyunso Kim, Hyo Kang, and Jaeyong Song examined more than 160,000 launches on Product Hunt. They found that entrepreneurial entry rose sharply after ChatGPT's release, driven disproportionately by solo founders, and especially in categories that had historically favored teams.


Then came the catch. Much of that growth was low-commitment, experimental entry, and it didn't translate into more solo founders at the top of the quality distribution. Teams remained dominant in the highest ranks of the platform.


Babson's GEM report says something similar in different language. Established business ownership in the United States continued to decline in 2025, part of a long-running trend. Donna Kelley, who co-leads the U.S. GEM team, was blunt about it: fewer ventures are reaching long-term stability. "We have some concerns about the sustainability of businesses, about businesses sticking around."


That's the honest check on everything above, and I'd rather put it on the page than let somebody else raise it in the comments.


AI may make it dramatically easier to start a business without making it dramatically easier to build a successful one.


In fact, if everybody suddenly has access to the same development capabilities, we might end up with extraordinary competition. Producing software becomes easier. Finding customers doesn't necessarily become easier. Designing something becomes easier. Knowing what deserves to exist doesn't. And maintaining what you've built, earning trust, managing liability, and keeping customers happy remain very real responsibilities.

That brings us to the next question.


When Capability Becomes Abundant, What Becomes Scarce?

In The End of Hypothetical Living, I wrote something that seems increasingly relevant: Everyone gets the forge. But not everyone knows what to make.


That's the shift.


When nearly anyone can produce software, content, designs, and business processes, the tools stop being the primary differentiator. Judgment becomes more important. Understanding customers becomes more important.

Reputation, trust, relationships, taste, and the willingness to accept responsibility for what you create become more important.


I think we're going to see a renewed appreciation for people who genuinely understand an industry or a problem. Because those people know where the value is. AI gives them a much more practical way to capture it.

And that's a very different proposition from simply replacing employees with software.


There's Still a Giant Behind the Curtain

There's one uncomfortable part of this story we shouldn't ignore. These increasingly independent entrepreneurs aren't operating in some magical world without infrastructure. They're relying on enormous data centers, cloud platforms, AI providers, payment processors, distribution networks, and marketplaces. A one-person company might have almost no employees while depending on billions of dollars of infrastructure owned by somebody else. So we could potentially see two seemingly contradictory developments happening simultaneously:

More decentralized business ownership at the edge, and more concentrated infrastructure ownership at the center.


That's a strange economy. Millions of individuals operating increasingly capable microbusinesses, supported by a relatively small number of massive technology providers. It raises serious questions about pricing power, platform dependence, data ownership, and who ultimately captures the economic value. The gatekeepers may not disappear. Some of them may simply move deeper into the infrastructure. That's worth watching very closely.


The Return of the Builder

For most of the industrial era, increasing productive capacity generally meant assembling larger collections of people, capital, machinery, and organizational structure. We built institutions around the limitations of what individuals could accomplish. And those institutions made extraordinary things possible.


But AI introduces another direction.


What happens when the capabilities of an institution can increasingly be accessed by an individual? What happens when you don't need to assemble a department before you can test an idea? Or when a small company can serve a specialized market that would never have supported a conventional business? Or when decades of professional experience can be packaged into a product without hiring a development team?


We're not eliminating the advantages of large organizations. We're expanding the range of things that smaller organizations, and even individuals, can realistically attempt. And I suspect this is going to reshape the economy in ways that the employment statistics alone won't capture. Because we're accustomed to measuring economic change in jobs created, jobs eliminated, corporate headcount, and productivity per employee. But what if a meaningful part of the next economic expansion comes from businesses that never needed to hire those employees in the first place?


What if the most important number isn't how many jobs a company creates, but how many people can now create a company? And what if some of those companies are small by design, economically durable, and perfectly content to stay that way? That's a different model of economic participation. It doesn't eliminate the risks of automation, the disruption of traditional careers, or the need for economic security. But it does suggest that the story doesn't end with displaced workers.


It may also include millions of newly empowered builders.


Here's the part that makes me think this is further along than we give it credit for. The gap between how many businesses people file to create and how many of those actually hire anybody has been widening for twenty years. AI didn't open that gap. AI may be the first thing that makes living in it sustainable.


We've been asking what happens when AI can do the work of ten people.

I think the more interesting question is what happens when one person can finally build the business that used to require ten. Because maybe the biggest effect of AI won't be the number of jobs it destroys. Maybe it'll be the number of businesses that never could have existed without it. And maybe, just maybe, we're watching the beginning of a new entrepreneurial economy without quite recognizing it yet.



Sources & Further Reading

External reporting and research

Related articles by Rich Washburn


Editorial note: Business applications are not equivalent to successful businesses. The available statistics do not establish AI as the cause of increased entrepreneurship, and the Census formation series counts employer businesses only, which excludes businesses that never add payroll.


Rich Washburn is a technologist, strategist, and Founder & Chief AI Architect of ARIA AI Labs, working at the intersection of AI, infrastructure, communications, and capital. He also serves as Managing Partner and Chief AI Officer at Eliakim Capital.

Comments


Animated coffee.gif
cup2 trans.fw.png

© 2018 Rich Washburn

bottom of page