The Last Generation of Majority Humans

We are the last generation where humans are the majority labor species on Earth. Not in 200 years. Not in 50. Within one working lifetime.

I know how that sounds. But the numbers aren't science fiction -- they're in Citi's December 2024 GPS report, and they're brutal in their clarity: 1.3 billion AI robots by 2035, over 4.1 billion by 2050. A population curve that looks less like technology adoption and more like a species expansion. If this were a biological organism, epidemiologists would call it an outbreak. Robotics analysts call it "CAGR."

This is the robot population explosion. And almost no one is thinking about it correctly.

The conversation keeps getting trapped in the wrong frame. We keep asking will robots take jobs? -- but that's the small question. The real question is what happens when the fundamental assumption behind every economic system humans have ever built simply stops being true: that humans are the primary labor input. From the first wheat field to the Industrial Revolution to the SaaS era, the unspoken rule was always the same. If work needs doing, a human must do it. Machines helped, but humans were the invariant. The baseline. The unit of production.

For the first time in 300,000 years of human history, that invariant is breaking. And the speed at which it's breaking will catch civilization flat-footed.

Every few centuries, a society hits what demographers call an inversion -- when the young no longer outnumber the old, or rural populations flip with urban ones. These transitions are quiet but profound because they rewrite how everything functions. We're about to hit the most extreme inversion since the agricultural revolution: a world where non-human workers outnumber human workers.

Rob Garlick, who ran technology and innovation research at Citi, put it bluntly:

"We will have gone from less than 1 percent of the working population to more than the working population with non-human workers, and they do this cheaper and cheaper and cheaper."

Not metaphorically. Literally.

The slope tells the story. There were 354 million AI robots operating globally in 2024. By 2030, Citi projects 749 million. By 2035, 1.3 billion -- enough to exceed the entire workforce of the developed world. By 2050, 4.1 billion, roughly twice as many working robots as working humans. And those remaining human workers will be older, fewer, and dramatically more expensive. The inversion doesn't happen gradually. It happens all at once, and then it's simply the new reality.

Humanoids -- the category designed to directly replace human physical labor -- are growing at 60.7% annually, faster than smartphones at their peak. And smartphones weren't designed to walk into a warehouse and immediately cover the night shift.

What makes this inevitable isn't capability. It's cost.

The most dangerous force in economics is a cost curve you can't fight. Citi's report contains one. A modern humanoid robot runs about $15,000, and when deployed against a $41-per-hour human worker -- nurses, skilled tradespeople, warehouse leads -- the robot pays for itself in under four weeks. Four weeks. There are enterprise software tools with longer onboarding periods. Even against minimum wage, the payback is under six months.

Garlick was direct about the implication:

"You can already buy a humanoid today which gives you a payback period versus human workers of less than 10 weeks. Humans can't compete on this basis."

This is the quiet part no one wants to say out loud: you don't need AI that's as smart as a human if the unit economics beat a human into the ground. People keep debating capability while capital is doing math. And the math says this: if a robot costs less per hour than payroll taxes, it wins.

RaaS -- Robots as a Service -- is the accelerant. Instead of buying hardware outright, companies can now rent physical workers like cloud compute. A metered labor force. RaaS pricing runs $2 to $8 per hour. The average U.S. factory worker costs $28. That's not a gap -- that's a canyon. No business survives ignoring a 4-to-14x cost differential, not in competitive markets, not globally. The second one competitor deploys robot labor, everyone deploys robot labor or dies. And the robots get cheaper every quarter.

This is why the population explosion is inevitable. It's not ideology. It's not futurism. It's compounding economics, and humans have never won a fight against compounding.

If you want to know how real this is, stop watching the debates and start watching what companies do with their most valuable assets.

Tesla quietly retired its Model S and Model X production lines -- some of the most sophisticated automotive manufacturing infrastructure on Earth -- to make room for Optimus humanoid assembly. That is not a research project. That is a bet. Elon Musk said it openly at Davos: "My prediction is there'll be more robots than people." Most observers heard futurism. Manufacturing executives heard strategy.

One viral tweet captured a recent Musk exchange:

"Will Optimus robots take away jobs from people at Tesla? No. Elon explains: 'The best useful robots in the beginning will be any continuous operations, any 24/7 operation because they can work continuously.'"

The framing is careful. "Continuous operations." Translation: anywhere humans need to sleep, eat, or take breaks -- which is everywhere.

Meanwhile, McKinsey -- the consultancy that advises half the Fortune 500 -- now has 20,000 AI agents working alongside its 40,000 human consultants. A year ago, they had 3,000. Bob Sternfels, McKinsey's global managing partner, expects agent-human parity within 18 months. The consultants advising the world's largest corporations on workforce strategy are already replacing their own workforce. That's the tell. When a firm with four decades of human-capital DNA starts tipping toward synthetic labor, the floodgates are open.

One Tesla investor put the scale in perspective:

"Elon Musk betting big on Robotaxi and Optimus to take Tesla to entirely new levels. $5T from autonomous fleets. $25T from humanoid robots. It sounds wild until you remember people once said EVs would never go mainstream."

$25 trillion. That's not a product line. That's a new sector of the global economy.

Now imagine what happens when labor itself is no longer scarce.

Every economics textbook starts with the same assumption: labor is limited. Humans can only work so many hours, train so fast, scale so far. Wages, prices, margins, growth -- the entire economic structure is built on that constraint. But what happens when labor becomes effectively infinite? Cheap. Compounding. Instantly deployable. Globally portable. Upgradable with a software patch.

That's not a different market. That's a different civilization.

When you can deploy 10,000 new workers with an API call, headcount stops being a constraint. Growth stops being labor-limited. Geography stops mattering. Demographics stop mattering. This is the reversal no policymaker is ready for: aging populations won't cause economic decline if the majority of the workforce isn't human. Japan doesn't have a labor crisis -- Japan is simply first to full robotic labor saturation. Countries with shrinking human populations won't collapse; they'll reallocate. Countries with booming populations won't automatically win; they'll be competing with a species that doesn't sleep.

The old axis of labor-rich versus labor-poor collapses entirely.

For 10,000 years, if a nation wanted more workers, it needed more births or more immigration. That link is now severed. If China wants to add 200 million workers, it doesn't need 200 million people -- it needs 200 million robots. If India wants to maintain cost advantage, it won't do it through a billion citizens; it'll do it through RaaS pricing. If the U.S. wants to reshore manufacturing, it doesn't need to rebuild human labor pipelines -- it needs electricity, land, and access to the robot supply chain.

Labor was the currency of nations. Now it's a commodity.

The geopolitical order that defined the 20th century -- built on population size, workforce education, immigration flows -- is about to be rewritten by whoever controls synthetic labor production.

And then there's the question no one wants to confront honestly.

We keep asking whether robots will take jobs, but the deeper issue is this: what happens when the majority of productive entities in your society are not biological? Who earns the income? Not the robots. Who gets bargaining power when management can scale its workforce infinitely at $4 an hour? What does unionization even mean in that world? What does "full employment" mean when employment is no longer the main way humans participate in the economy?

We built every institution -- taxes, pensions, insurance, education, currency policy -- around the idea that human labor is the engine. When labor detaches from humans, every one of those institutions becomes unmoored. The economic flywheel keeps accelerating, but fewer and fewer humans are connected to the axis. That tension will define the next 30 years.

Everyone keeps talking about AI models. Quiet money is moving into robots.

Because software changes industries. Labor species change civilizations. The biggest companies of the 2030s will be the ones that control robot manufacturing, robot operating systems, robot labor marketplaces, and robot-compatible infrastructure. The internet created trillion-dollar companies. Robot labor markets will create multi-trillion-dollar supply chains -- and reshape the power structure of nations.

We are not entering the age of automation. We are entering the age of a second industrial population -- billions strong, compounding, tireless, and increasingly humanoid. For 300,000 years, humans were the default unit of work. In a single generation, we won't be.

The strangest part is that the future won't be shaped by the people who fear this. It will be shaped by the ones who read the census correctly -- and build accordingly.