In June, Amazon CEO Andy Jassy had a blunt message for his 350,000 corporate employees: There were going to be fewer of them in the near future, thanks to the “efficiency gains” he expected from AI. The proclamation generated big headlines and an uproar from staff. But it struck me as merely honest. He was acknowledging something that pretty much every CEO who sits atop a large white-collar workforce is quietly hoping to achieve sooner or later.

After all, Jassy hasn’t been the only executive to hint at a future of lower headcount. The head of JPMorgan’s consumer and community business predicted in May that AI will reduce the number of employees in its operations division by 10%. That same month, the CEO of Klarna said that the company’s investments in AI has already driven the company’s headcount to shrink by 40%. And the CEO of Ford — a company that employs tens of thousands of white-collar professionals — declared that AI will wipe out “literally half” of all white-collar jobs. Meanwhile, Kian Katanforoosh, the CEO and founder of the software startup Workera, tells me that he never wants to have much more than the 80 or so employees he has today, no matter how successful his business ends up becoming. “I truly believe we can go super super far without growing more,” he says. “I’m an engineer. I don’t want to have to manage so many people if I don’t need to.”

It’s not like CEOs ever enjoyed shelling out for the salaries or navigating the personnel headaches that come with the sprawling bureaucracies they employ. But for more than a century, armies of office workers were a necessary cost of doing business. To grow from tiny upstarts into titans of industry, companies needed an ever-multiplying number of HR reps, accountants, marketers, engineers, analysts, and project managers.

In recent months, that 100-year-trend is starting to come undone. Everywhere you look, AI appears to be helping leaner teams take on work that used to require more people. And executives are talking about their large workforces — once their greatest competitive advantage — as if they’re an unfortunate holdover from a bygone, bloated era.

If today’s corporate giants shrink their ranks, and if tomorrow’s giants never need to bulk up in the first place, we may well be witnessing the end of a defining feature of corporate America: the mega-employer. That could give rise to a whole new generation of nimble companies that innovate faster — but also leave workers navigating a world of diminished career paths and fewer jobs.

Before the Industrial Revolution, most Americans worked for themselves as farmers or craftsmen. And those who didn’t worked for very small operations — say, a few journeymen training under a master shoemaker. The resulting economy was a patchwork of all these tiny businesses.

That started to change with the advent of capital-intensive industries like textile manufacturing, which required organizing larger groups of people under a single employer. Then came railroads in the late 19th century. With projects that took many years to realize and stretched over thousands of miles, vast numbers of workers needed to be on the same page. “If you mess it up, there’s a big explosion,” says Louis Hyman, a political economy historian at Johns Hopkins. “You needed to really coordinate your mechanisms and make sure that people are doing things exactly the same way.”

In the 1930s, about a tenth of the labor force worked for companies with at least 10,000 people. By the end of World War II, that share had surged to a third.

As mass production developed, Hyman says, many of the most consequential innovations during this time weren’t so much technical breakthroughs: They were social inventions to coordinate the labor of all the people it took to get the most out of the new machines. The assembly line broke down complicated work into simple, repeatable, standardized tasks; scientific management emphasized the importance of monitoring, measuring, and optimizing everyone’s performance; and the M-form corporate structure created a blueprint to manage sprawling bureaucracies through a clear chain of command. In the 1930s, about a tenth of the labor force worked for businesses that employed at least 10,000 people. By the end of World War II, that share had surged to about a third.

By the 1970s, some of that bigger-is-better ideology started to change. A new management philosophy set in, normalizing layoffs that took aim at bloat. And as robots automated many blue-collar jobs, IBM mainframes and word processors eliminated a whole set of white-collar clerical roles as well. Still, there was plenty of work that technology couldn’t automate, which meant that companies needed large teams of college-educated professionals to keep them going. Even the most tech-forward companies saw their people — especially their coders — as mission-critical to their success. “Hiring great people — especially engineers — is one of the biggest challenges that any technology company has,” Mark Zuckerberg lamented in 2013. “Our country doesn’t produce the volume of engineers that the companies would want to hire.” Tech giants often hired more than they needed to make sure they had a steady supply of talent, and to attract and retain the best of the best, they treated their employees like gods.

If you were to pinpoint one moment the gods turned mortal, it would probably be November 9, 2022 — the day Meta laid off more than 11,000 employees. From there, virtually every tech company followed suit, with employers across other industries close behind. At first, the cuts were chalked up to overhiring in the pandemic. But two and a half years later, the layoffs haven’t stopped and hiring is still down. More and more, AI appears to be driving those austerity measures. In an industry that once hoarded talent like gold, the shift is striking. CEOs no longer seem to view the bulk of their workforce as indispensable, and they say as much: A common refrain among tech leaders from Mark Zuckerberg to Elon Musk to Dara Khosrowshahi now is some version of “If you don’t like it here, you should leave.” Companies like Microsoft, Meta, Google, and Salesforce had reliably increased their headcounts year after year. Now, according to the workforce analytics provider Live Data, all of them employ fewer people than they did at their 2022 peak.

J. Scott Hamilton, Live Data’s CEO, says this is probably just the beginning. To gauge how much deeper the cuts could go, his team recently analyzed the detailed responsibilities of most roles at Microsoft to estimate the share of tasks that could, in theory, be done by AI. Their conclusion: If Microsoft were to offload all of those automatable tasks to AI, it would eliminate 36% of the work currently done by employees. That would mean the company could lay off some 80,000 employees.

On the one hand, that’s an aggressive scenario: Companies are rarely able to overhaul their workflows to take full advantage of a new automation technology’s capabilities. If they do, that transition takes a very long time. And besides, some work is simply too high-stakes to entrust to error-prone AI — even if it’s technically possible. On the other, the estimate may prove conservative: Live Data’s predictions assume that AI will remain at 2025-level capabilities. Given how much better the leading large language models have become over the last two years, the best tools will almost certainly be able to handle more than what they can today.

“The optimists are saying that the good companies will simply redeploy the assets elsewhere now that they can be more efficient,” Hamilton says. “But I think an equal argument can be made that they’ll just say, ‘We’re going to do the same amount with fewer people.'”

If Microsoft offloaded all automatable tasks to AI, it could eliminate some 80,000 jobs, Live Data found.

If that sounds like a far-off hypothetical future, consider what’s already happening today at startups. OpenAI CEO Sam Altman says he’s making bets with his friends on when we’ll get the first “one-person billion-dollar company.” And Arthur Kaneko, a general partner at Coreline Ventures, tells me he’s noticed that early-stage founders are raising their initial rounds of funding with fewer employees than they would have had in the past — among the AI-fluent founders, perhaps with less than half.

“The way companies are being built is just fundamentally changing right now because of AI,” Kaneko says. “Through the use of AI coding, AI marketing, AI sales, people are able to do a lot more work with way fewer people.”

And he thinks these startups will stay lean as they scale into successful businesses. “They just won’t hire the people that Meta and Microsoft had to hire to get to where they are,” he says. “I do think per-company headcount will permanently be depressed in startups.”

There are reasons to be hopeful about a new era of smaller employers. If AI makes it cheaper and easier to launch companies, we’ll probably see more of them — and that would be great for the long-term health of the economy in all kinds of ways. New businesses tend to employ people with less experience and fewer credentials who get passed up by the bigger companies. They’re more willing to try new things, which drives innovation. And they create more competition for the established giants, which is good for consumers.

Smaller companies may also be good for the workers inside them. There’s a lot that people hate about working at big organizations: the constant turf wars, the endless layers of approval, the meetings before the meetings, the sense that you’re just one tiny inconsequential cog in a giant machine. Smaller bureaucracies would minimize that, which is one reason why people often feel more motivated in leaner workplaces. According to Gallup, employees at small companies report the highest engagement, with scores dropping below the national average once organizations hit 500 employees. On the same stage where Altman made his one-person unicorn prediction, Reddit co-founder Alexis Ohanian raved about the benefits of this possibility. “CEOs and founders are going to be so excited to get up and go to work with much smaller, much more performant, much more culturally strong teams,” he said.

But a world of shrunken employers could also rob workers of something essential: the long-term career paths that big companies used to offer. With so many roles under one roof, big companies made it possible for workers to try new things, move up, and build careers. Smaller firms don’t offer the same range of opportunities, which means people will likely need to switch companies a lot more in the future. Smaller firms are also less likely to invest in on-the-job training — a shift that would hit early-career professionals hard, just as their roles face the greatest risk from AI.

The big question is what this all means for college-educated workers. If enough startups emerge, they might create new jobs to offset the ones disappearing from big companies. But that would require an unprecedented boom in entrepreneurship — one enormous enough to make up for the retrenchment of the giants. In 2022, 29% of the American workforce worked for an organization that employed at least 10,000 people. Meanwhile, the country’s education system is churning out ever more college grads, who studied hard with the expectation of a stable future in white-collar work. If big companies hire less, and small companies also hire less, where will they all go?

The usual reassurance is that AI, like every disruptive technology before it, will eventually create more jobs than it destroys. That glosses over an important detail, according to Carl Benedikt Frey, an economist at Oxford. In the early stages of the Industrial Revolution, most innovations simply made existing work faster and cheaper — like the loom, which automated the work of skilled weavers but still produced more or less the same fabric. That made a handful of industrialists very rich, but for the average worker, wages barely budged for the first 80 or so years of industrialization. It was only later — with inventions like electricity and the automobile that gave rise to entirely novel industries — that economic growth surged and better, high-paying jobs emerged. Had that second wave never arrived, we’d remember the Industrial Revolution very differently. “Most productivity gains over the long run,” Frey says, “come from doing new and previously inconceivable things.”

Right now, corporate America seems stuck in that first phase. So many executives are laser-focused on using AI to do the same work with fewer people, rather than applying it to problems we couldn’t solve before — the kind of breakthroughs that would open up new lines of business and generate more demand for labor, not less. “A real risk is that we’re getting leaner organizations, but they’re not really creating that much new,” Frey says. “That would be a bleak future, and I do worry we’re moving in that direction.”

Correction: August 11, 2025 — A previous version of this story incorrectly stated Louis Hyman teaches at Cornell, his former employer. He now teaches at Johns Hopkins.

Aki Ito is a chief correspondent at Business Insider.