FOUNDATIVE INC.
Thesis

Thesis

The Future of Work

By Aaryan Kushwah

In 1900, 50% of the United States was operated by independent operators. By 1970, this number dropped to 7%. In 1955, roughly one in every two hundred working people in the country received a paycheck from General Motors (wtf). The company’s revenue was 3% of the US GDP. Its CEO told the Senate that what was good for the country was good for General Motors, and meant it as a tautology, not a slogan. William Whyte called that ‘one in 200’ person the Organization Man. The white-collar middle manager who had “left home, spiritually as well as physically, to take the vows of organization life”.

The Organization Man is now dead. He survived the Reagan recessions, the 1990s downsizings, the offshoring of the 2000s, and the COVID layoffs of 2020. He will not survive what is coming next.

Most essays about AI and the future of work are written from inside the world the Organization Man built. They ask whether Claude will replace a paralegal, or whether an LLM can outperform a junior accountant. These are pretty interesting questions, but they’re the wrong questions. They assume the institutional form of the firm (a hierarchical entity that pays salaries to coordinated workers) is the constant, and that AI is the variable acting on it. It is, (un)fortunately, the other way round.

The firm itself is the variable. The companies being built right now are smaller, faster, more profitable per head, and less dependent on hiring than any companies in the history of American capitalism. Today, the creator running a media company is the same shape as the founder running a SaaS company is the same shape as the operator running an e-commerce brand. They are all the same animal now.

It’s probably a good time to mention I lied. This won’t be a forward looking essay. The future is much too similar to the past to pretend as if I made these predictions. Mass employment by giant corporations has always been the weird thing. The rest of human commercial history has been small operators, all the way down. We are about to remember.

In three generations, America went from a nation of small operators to a nation of company-men. What happened in between was the creation and total cultural triumph of the Modern Corporation: the assembly line, the org chart, the HR department, the pension fund, the pyramid-shaped management hierarchy, the lifetime-employment social contract, the holy trinity of GM, AT&T, and IBM. That format is now so familiar that it feels like the natural state of human economic organization. But it isn’t. It was a sixty-year industrial accident produced by a stack of one-off conditions that no longer exist:

Mass production demanded mass capital, which only large entities could raise. Distribution was slow and physical, which rewarded firms with national logistics. Information moved through corporate hierarchies, because there was nowhere else for it to move. Communication had a marginal cost, so coordination required co-location. Marketing was broadcast, so reaching customers required a budget only big firms could afford. Each of these was a fortress wall around the corporate form. And each of them has been breached, individually, over the last forty years.

The first wall came down with the personal computer. The second came down with the internet and overnight logistics. The third came down with email and Slack. The fourth came down with Zoom and Notion and GitHub. The fifth wall came down with Google AdWords, Facebook, Shopify, TikTok. By 2010, every individual cost barrier to running a business had collapsed by ninety percent or more. And yet the self-employment rate in the United States had only crept back up to about ten percent. Why?

The operational overhead of running a serious business is multidimensional, and each tooling revolution only collapsed one dimension at a time. The PC made document production cheap, but you still needed accountants. The internet made distribution cheap, but you still needed customer support. AWS made software infrastructure cheap, but you still needed marketers. Shopify made selling things cheap, but you still needed designers. Each revolution turned one wall into a doorway, but the building was still mostly walls.

Agents are the first technology in human history that can collapse all of those dimensions simultaneously.

There’s a pretty interesting pattern here. When the cost of a productive activity collapses by a hundredfold, the number of people doing it goes up by a hundredfold within a generation, and the median producer shrinks dramatically in size. It happened with books. With clothes. With software. With commerce. With media. It is about to happen with companies themselves.

Why this time really is different

In 1930, Keynes predicted a fifteen-hour workweek by 2030. That is clearly not happening. Wassily Leontief, the Nobel laureate, wrote in 1982 that humans would “go the way of the horse: first diminished and then eliminated.” That clearly did not happen (hopefully lol). Daniel Pink wrote Free Agent Nation in 2001 and predicted that within twenty-five years the W-2 lifetime job would be a historical curiosity. We are now within months of his deadline, and W-2 employment is doing fine. Every previous generation has had its prophet of mass solopreneurship, and most of them have looked silly within twenty years.

So the contrarian (me) must explain, specifically, what is different now. The honest answer is not “AI is smart.” Smart was never the binding constraint. The binding constraint was always bundling. Previous tooling revolutions each collapsed one dimension of the operational stack of running a business. The press made writing cheap. The PC made layout cheap. AWS made hosting cheap. Shopify made selling cheap. Stripe made payments cheap. Each one of those collapses, individually, was an enormous productivity gain. None of them, individually, was sufficient to make a one-person company viable at scale.

AI agents are the first technology in history that compresses all dimensions of the operational stack on a single S-curve. The minimum viable team for a $10M ARR internet business is about to drop to one. Pink was right that people wanted to work for themselves. He was wrong about when they could afford to. Working for yourself has, until now, meant working without the operational support that an employee at a big firm gets for free. Now, for the first time, the small operator has access to support functions that until recently only a Goldman Sachs partner had. And, soon, support functions that no one had.

The per-employee revenue of the highest-value tech companies has gone from about $300,000 in 2000 to $640,000 today (the S&P 500 average). For frontier-AI companies, it is an order of magnitude higher. For Midjourney specifically, around $5 million per employee. For Telegram, higher. For OnlyFans, hilariously higher. This is the trailing edge of a fifty-year compounding trend in which the value-per-employee of a successful internet company has gone from roughly $1M (1990s) to $5M (2010s) to $50M+ (2020s). The trend will accelerate. And it will broaden out of the long tail of tail outcomes, and into the median.

Predictions

If the argument is right, here are the things that should happen in the next thirty-six months. I have tried to be specific enough to be falsifiable and aggressive enough to be useful.

By the end of 2026

By the end of 2027

By the end of 2028

The objections

The strongest objection is the historical one: every generation predicts the death of W-2 employment, and every generation is wrong. Why won’t this be Pink-2001 redux? It might be. I could be totally wrong. The sixty-year arc from 50% self-employment to 7% was not driven by a single technology; it was driven by the institutional infrastructure that grew up around large-scale industrial production. That infrastructure is still mostly in place. Healthcare runs through W-2 employment. Mortgages go to W-2 borrowers. Visas favor W-2 workers (this writer’s stake in the question is not zero). Risk pools, retirement plans, status hierarchies, the entire sociology of the upper-middle class remains organized around the assumption that you work for an employer.

So the prediction is not that the corporation disappears. It is that the corporation stops growing as a share of economic activity, and the long tail of small operators starts growing very fast. America has 33 million businesses today. By 2030, it will have 60 to 80 million. The corporation will still exist. It will be a smaller part of an enormous, much more fragmented whole.

The second objection is the cultural one. Maybe people don’t want to be solopreneurs. Maybe they want a boss, a team, a Slack to belong to, an office to commute to, a sense of being part of something larger. This is the real objection. Most of the people who became solopreneurs in the last decade did so reluctantly, after layoffs. Most say they would prefer a normal job. This is probably the biggest risk of the bet we’re taking.

But the historical answer to that is that what people prefer often loses to what is economically inevitable. The German peasant of 1830 preferred the village to the textile mill. By 1860, he was in the textile mill anyway. The 1955 typist preferred the typewriter to the word processor. By 1985, she was using a word processor anyway. Preferences adapt to the available options. The option to “have a normal corporate job” is going to feel, to a 23-year-old in 2030, the way the option to “be a typist in a typing pool” felt to a 23-year-old in 1995. It is not that the option doesn’t exist. It is that nobody who has any choice is taking it.

The Organization Man was, it turned out, a temporary shape.

America was a nation of small operators for most of its history. We forgot for sixty years. We are about to remember.

This is the thesis behind result.dev and why we’re building it.