Why Capital Is Shifting From Software To Robots

An Old Problem Rome Understood Better Than We Dp

Every declining civilization eventually runs into the same wall: not enough people to keep the machine running. Rome learned this the hard way. At its height, farmers grew the food, craftsmen made the goods, merchants moved them across borders, soldiers guarded the trade routes, and taxpayers funded it all. Then came the invasions, the wars, the food shortages — and with them, a shrinking population that historians have since tied directly to falling agricultural output, stagnant wages, and a state that could no longer collect enough tax to function.

The lesson buried in that history is simple but easy to forget: an economy doesn't just run on money. It runs on people — people to build things, deliver services, and buy what everyone else is building and delivering. Shrink that pool of people for long enough, and the damage compounds across the entire system, year after year.

Economists have formalised this idea too. Stanford's Charles Jones has argued that long-run economic growth ultimately depends on people generating new ideas — meaning a shrinking population doesn't just mean fewer factory hands, it means fewer scientists and researchers pushing innovation forward in the first place.

What Happens When Machines Start Doing The Thinking Too

AI complicates that old equation in a genuinely strange way. If machines eventually get good enough to run research, manage factory floors, and operate other machines, then a country's economic capacity starts looking less like a function of its population and more like a function of its capital — its chips, its factories, its power grid.

That might sound like a distant hypothetical. It isn't, if you've been watching what's happening in China.

Chinese humanoid robotics maker Unitree Robotics listed on the Shanghai Stock Exchange a few days ago, and the stock rose almost sixfold on its very first day of trading — briefly valuing the company at more than $50 billion. For a business that generated roughly 1.7 billion yuan (about $250 million) in revenue in 2025, that's an almost absurd multiple to put on the stock. But the debut wasn't really the interesting part. What matters more is what these robots are already starting to do to manufacturing itself.

Manufacturing Without The Manufacturers

For decades, the recipe for mass manufacturing was straightforward: find the country with the cheapest, largest workforce, and build your factories there. That's precisely how China became "the world's factory" in the 1970s — an enormous labour pool, low wages, and a dense supplier network capable of turning raw materials into everything from toys to electronics.

Automation is quietly rewriting that recipe. China has spent years pushing factory automation to genuinely extreme levels — so extreme that some facilities have earned the nickname "dark factories," places that can run production lines with almost no human presence at all. At BYD's Zhengzhou manufacturing complex, the company says automated robots now handle 98% of welding, assembly and quality inspection.

This isn't exclusively a Chinese phenomenon either. India already has its own example: Polymatech's semiconductor facility in Kancheepuram, Tamil Nadu, runs robots that assemble chips inside a cleanroom through the night, with human engineers largely just monitoring from outside. But this kind of extreme automation comes with its own fragility — Polymatech's own director has estimated that a serious power fluctuation at the facility could destroy roughly $1 million worth of production in one go.

Why Humanoid Robots Specifically Change The Math

This is where humanoid robots enter the picture, and why they're structurally different from the industrial robots factories have used for decades. A conventional industrial robot is typically built to do one very specific job — weld this joint, tighten this bolt — and if you want it to do something else, you often have to redesign the entire production line around it.

A humanoid robot, in theory, doesn't need that redesign. Built to move and work within spaces humans already use, it can walk into a workstation designed for a person and, at least in principle, switch between several different tasks without the factory needing to be rebuilt each time.

If that promise holds up at scale, it changes the fundamental calculation companies use to decide where to manufacture. A factory that can run a robot for longer hours, with more consistency, and without needing a single break, starts to matter more than simply finding the cheapest available worker. Picture two factories — one in China, one in India. The Chinese factory might pay meaningfully higher wages. But if it also has better automation, cheaper robots, a stronger local supplier base, more reliable power, and engineers who know how to actually integrate AI into a production line, it could still end up producing the same good at a lower total cost. In that world, cheap labour alone stops being enough to guarantee that manufacturing gravitates toward India.

China's Head Start Is Turning Into A Self-Reinforcing Loop

The scale gap here is already massive, and it's compounding. China installed roughly 295,000 industrial robots in 2024 alone — 54% of every new robot installation on the planet that year. India, by comparison, installed a record 9,100.

That gap isn't just about robot count. China has built an entire ecosystem around it: manufacturers producing the components that go into robots, engineers writing the software that runs them, factories willing to test and deploy them, and an industrial base large enough to absorb the technology at real scale. Each piece feeds the next — more automated factories create more demand for robotic components, more deployed robots generate more data to improve the underlying systems, and more production volume pushes the cost of the technology down further. That's a genuine feedback loop, and it's precisely the kind of loop that's difficult for a country without a comparably large, human-centric industrial base to replicate quickly.

Investors Have Already Noticed

Venture capital is placing real money behind the idea that this shift is commercially significant, not just technologically interesting. According to Crunchbase, companies working in what's broadly classified as "physical AI" — robotics, autonomous vehicles, drones, industrial automation and sensors — pulled in $47.4 billion in venture funding across 521 deals in the first half of 2026 alone. That's roughly 80% more than the same period last year, and nearly four times what the category raised in the second half of 2025.

Robotics specifically has now overtaken fintech in fresh investment as of H1 2026 — a genuinely notable handover, given how dominant fintech funding has been for the past decade. Taken together, this looks like the AI investment story quietly shifting away from companies that just produce information on a screen, and toward companies that can actually manipulate the physical world.

Where India's Real Opportunity Might Actually Sit

None of this means India needs to try to out-build China at manufacturing humanoid robots — that race already has a significant head start built in. India's more realistic opportunity may sit further up the value chain instead.

Someone still has to design the production system a robot operates within, maintain the machines when they break, train the underlying AI models, integrate different pieces of equipment so they work together, and troubleshoot the inevitable failures on the factory floor. Those are fundamentally different skills from the repetitive manual work automation is replacing. A country that simply imports robots without building the domestic talent to operate, maintain and improve them risks ending up with highly automated factories that it doesn't actually have the capability to run or upgrade on its own terms.

Framed that way, India's real challenge isn't a binary choice between people and machines. It's making sure its workforce moves up the value chain fast enough to stay ahead of the automation displacing the lower end of it.

The Harder Question: Who Actually Benefits?

There's a deeper economic question sitting underneath all of this. If companies can eventually produce more goods with fewer workers, productivity could rise sharply — consumers get cheaper products, companies earn fatter margins, and countries produce more without needing a proportional increase in their labour force.

But those gains rarely spread evenly. A factory worker who loses their job to automation doesn't automatically transform into an AI engineer. Without enough new kinds of jobs being created alongside this shift, an economy can become significantly more productive on paper while a meaningful chunk of its workforce becomes economically stranded.

That tension matters more for India than almost anywhere else. The country's large working-age population has long been one of the strongest arguments for why global manufacturing should shift here. That same population becomes a genuine liability, though, if the economy doesn't generate jobs fast enough to absorb the young people entering the workforce every year.

The Real Race Isn't About Cheap Labour Anymore

The countries that end up winning the next phase of global manufacturing probably won't simply be the ones with the cheapest workers. They're more likely to be the ones that can combine labour, robots, energy, software and industrial know-how at the lowest total cost, taken together as a system.

None of this means the goal should be replacing Indian workers with robots, or trying to shield every existing job from automation. The more useful goal is making the pairing of an Indian worker and an intelligent machine genuinely more productive than either could be on its own.

If China's bet is on factories where robots do most of the physical work, India's opening may be to build an economy where people become exceptionally skilled at directing, maintaining and improving those robots. That distinction — between simply having automation and actually knowing how to run it — could end up determining something fairly consequential: whether India becomes just the next country in line for low-cost manufacturing, or one of the genuinely most efficient manufacturing economies in the world.