China Is Winning the Robot War on Price: Unitree's $6.2B IPO and the $4,290 Humanoid

Let me start with two numbers that, side by side, tell you almost everything about where the humanoid robot race actually stands right now.

Figure AI, the American darling of the robot world, carries a private valuation around $39 billion — with very few robots actually working in the wild. Unitree, the Chinese leader, just fast-tracked a Shanghai IPO at a target valuation of roughly $6.2 billion — and it's already shipping thousands of humanoids, turned its first profit last year, and sells a working humanoid for as little as $4,290.

One company is valued at six times the other and ships a fraction of the product. If that strikes you as backwards, you're paying attention. And today, June 1, the Shanghai Stock Exchange's listing committee is reviewing Unitree's IPO application — which makes this the perfect moment to walk you through what's really happening, and why it matters to you whether you're an investor, a business owner, or just someone trying to understand the decade ahead.

China already won the early round

I want to give you the number that reframes everything, because most Western coverage buries it: China now controls roughly 80% of global humanoid robot shipments. Not "is catching up." Already controls. The early market is theirs.

Here's how 2025 actually shook out when you count units that shipped, not valuations on paper:

Roughly 16,000 humanoid units were deployed worldwide in 2025, and Chinese makers dominated the count. Output is projected to surge about 94% in 2026, with Unitree and AgiBot together expected to capture nearly 80% of all shipments. This isn't a forecast about who might win someday. It's a description of who's already shipping today.

The weapon is price, and it's brutal

Here's the part that should make every Western robotics investor a little nervous. China isn't winning on some secret breakthrough in intelligence. It's winning the way China has won before — on price, manufacturing scale, and supply chain control.

Look at what you can actually buy from Unitree right now: an upper-body humanoid starting at 26,900 yuan, about $4,290. The R1 model with 26 joints went for around $5,900. Even the full-size, more capable G1 lands near $16,000 — and that's still the most affordable real humanoid you can purchase on the planet.

Now hold that against the American approach, where the working assumption has been that a useful humanoid would cost somewhere between $30,000 and six figures. China just walked in and priced the entry models like high-end appliances. When you can buy a working humanoid for the price of a used car instead of a house, the entire adoption curve changes — and the company that sets that price gets to define the market.

This is the same playbook that let China dominate solar panels, EVs, and drones. Get the cost so low that scale becomes inevitable, then ride the volume. I've written before about how Beijing's $150B national AI strategy is deliberately engineered to win exactly these manufacturing-heavy categories, and humanoid robots are turning into the cleanest example yet.

Why Unitree's IPO is the moment to watch

The Unitree IPO is worth your attention even if you'll never buy a share, because it's a stress test for the whole story. Here's what's actually on the table.

Unitree is aiming to raise about 4.2 billion yuan ($610 million) at a target valuation near $6.2 billion. It plans to use the money for R&D and to massively scale manufacturing — the five-year plan calls for producing 75,000 humanoids and 115,000 quadruped robots every year. And unlike most of the robot field, this is a real business: Unitree posted an adjusted net profit of about $90 million in 2025, its first profitable year, up a staggering 674% from the year before.

But — and I promised you the honest version — there's a yellow flag. Ahead of the IPO review, Unitree reportedly posted a 52% year-over-year drop in Q1 profit. That tells you something important: even the leader's margins are getting squeezed by the very price war it started. Winning on price is a double-edged sword. You grab the market, but you bleed margin doing it, and public investors will eventually ask when the profit catches up to the growth.

If you're thinking about how to play this as an investor without betting the farm on one Chinese IPO, the saner route is usually a basket. I broke down that approach in the guide to AI ETFs — there are now humanoid-robotics-specific funds that hold Unitree-adjacent names, which spreads the risk of any single company's margins blowing up.

The plot twist: Nvidia just picked Unitree

Here's a development that complicates the tidy "US vs China" framing, and you should sit with it for a second.

Nvidia — the most important American AI company there is — just selected Unitree's nearly six-foot H2 humanoid as the body for the first robotics system Nvidia is selling to researchers, pairing it with Nvidia's own Jetson Thor hardware. Read that again. The crown jewel of American AI chose a Chinese robot as its reference platform.

That tells you the lines here aren't as clean as the headlines suggest. The American strength is the brain — the chips and AI models, where Nvidia just posted an $81 billion quarter. The Chinese strength is the body — cheap, mass-produced, actually-shipping hardware. For now they need each other, and the most capable humanoid in the near term might well be a Chinese body running an American brain. Whether that partnership survives the next round of export controls is one of the biggest open questions in tech.

What this means for you

Let me make this practical, because "China leads humanoid robots" is useless to you until it changes a decision.

If you run a business that touches physical work — warehousing, manufacturing, logistics, retail stocking — this is the early warning that the economics are about to shift under you. UBTech has already deployed 1,000+ industrial humanoids into automotive plants, electronics fabs, and logistics. When a working humanoid costs $16,000 instead of $100,000, the math on automating repetitive physical tasks flips fast. I walked through the parallel shift in how AI is already reshaping logistics costs, and physical robots are the next chapter. Start thinking now about which tasks in your operation are repetitive, physical, and expensive — because those are the first to go robotic.

If you're an investor, the lesson is to separate shipped reality from valuation hype. A $39 billion company that ships little and a $6.2 billion company that ships the most units in the world can't both be priced right. One of those bets is wrong. I'm not telling you which — I'm telling you the gap is the opportunity, and the broader robotics startup landscape is where you go to understand who actually has product versus who has a pitch deck.

If you're thinking about your career or your kids', understand that the cheap-humanoid wave changes the jobs conversation in a way software AI alone didn't. Software automated knowledge work. Cheap humanoids extend automation into physical work that was supposed to be safe. That's not a reason to panic — it's a reason to aim at the parts of work that direct, design, and maintain these systems rather than compete with them.

The honest take

China didn't win the robot race because it out-invented America. It won the early round the way it wins most hardware races — by making the thing cheap, making a lot of it, and controlling the supply chain that everyone else depends on. Unitree's $6.2 billion IPO, its $4,290 humanoid, and its 80% market share alongside AgiBot are all the same story told three ways: the future of physical labor is being mass-produced in China right now, at prices designed to make adoption unstoppable.

The American counter is the brain — the chips and models that make these bodies actually smart, which is exactly why Nvidia reaching for a Unitree body is so telling. The next few years come down to a single question: does the value in humanoid robots live in the body or the brain? If it's the body, China already won. If it's the brain, the race is still wide open. My honest guess is that the real money ends up split — and the smartest move you can make is to stop watching the valuations and start watching who actually ships.

So here's what I'd leave you with: the next time someone shows you a flashy humanoid demo with a nine-figure valuation, ask the only question that matters in this market — how many did you actually ship, and what does one cost?