The Robot Bubble That’s Quietly Becoming A $5 Trillion Reality
First it’s a viral video. Then it’s your coworker.
TLDR;
Morgan Stanley projects the humanoid robot market will hit $5 trillion by 2050.
Figure AI is valued at $39 billion with one robot doing one task at one factory.
Chinese manufacturer Unitree launched a humanoid for $5,900. That’s less than some laptops.
China recorded 610 robotics investment deals in just 9 months of 2025.
The numbers look like a bubble. The infrastructure being built says otherwise.
The numbers don’t lie (but they’re not everything..)
Figure AI raised over $1 billion at a $39 billion valuation.
Their robot has been operating on a BMW production line for five months. Ten hours a day. Every day.
Sounds impressive until you realize it’s one robot, doing one task, at one factory.
You do the math.
But here’s the thing…
I’ve been watching the robotics space closely.
And what I’m seeing is reminiscent of something we’ve all experienced before. Remember when electric vehicles were “a niche”? Remember when smartphones were “just for techies”?
The pattern is the same.
Overvalued companies. Viral videos. Skeptical analysts. And then, slowly, the infrastructure catches up to the hype.
Why the “bubble” might be different this time
The China Factor
This is where it gets interesting.
Unitree Robotics is preparing an IPO at a $7 billion valuation. Their latest humanoid, the R1, costs $5,900. For reference:
Boston Dynamics’ Atlas: Over $200,000
Tesla’s Optimus (projected): $20,000
Unitree R1: $5,900
That’s not a typo.
A Chinese company is selling humanoid robots for less than some laptops. And people are buying them.
Two Unitree humanoids were sold on JD.com to actual consumers. First documented consumer sales in the sector.
Morgan Stanley forecasts that by 2050:
China will have 302.3 million humanoid units
The US will have 77.7 million units
China is building the robots. China is buying the robots. And China is already deploying the robots.
Their robot density is 470 robots per 10,000 manufacturing workers. That’s second in the world, behind only South Korea.
The Figure AI reality check
Brett Adcock, Figure AI’s CEO, has been aggressive with his marketing. Videos showing robots in factories. Claims about “fleets” working at BMW.
Fortune did some digging.
The reality? A single robot doing a limited task. For a while, only during off-hours.
But here’s where it gets nuanced.
That same robot has now been running for five consecutive months during live production. Ten hours daily. It picks up sheet metal parts and places them onto fixtures.
Boring? Yes.
Valuable proof of concept? Also yes.
The gap between marketing and reality is shrinking. Just slowly.
The production targets are real
This is what separates robots from past bubbles.
Tesla is targeting around 5,000 Optimus units in 2025. By 2026? They want 50,000 to 100,000.
BYD isn’t far behind. They’re planning 1,500 humanoids this year, ramping to 20,000 by 2026.
Agibot is going for 5,000 units in 2025.
Agility Robotics already built a factory capable of churning out 10,000 Digit robots per year.
Figure AI is constructing a facility targeting 12,000 humanoids annually.
These aren’t projections. These are factories being built. Production lines being installed. Workers being hired.
Tesla just started construction on a dedicated Optimus factory at Giga Texas. Target: 10 million robots per year. Musk wants $20,000 manufacturing cost per unit.
Is it ambitious? Extremely.
Is it impossible? They said the same about electric vehicles.
The barista is here
While everyone debates valuations, robots are already making coffee.
ADAM, built by Richtech Robotics, operates at Clouffee & Tea in Las Vegas.
It makes roughly 100 different drinks.
It speaks four languages.
It uses machine vision to recognize customers.
MUJI has robot baristas in Portland, New York, Vancouver, and Toronto. Videos went viral on TikTok and Instagram.
In San Jose, an AI-powered humanoid barista named “C” just debuted at NCM Café.
These aren’t prototypes. These are operating businesses.
The supply chain opportunity
This is where I’d focus if I were investing.
Every humanoid robot needs:
Semiconductors (AI chips for thinking)
Sensors (cameras, LiDAR, touch)
Actuators (motors, reducers)
Batteries (power)
AI software (decision-making)
It doesn’t matter who wins the humanoid race. Tesla, Figure, Unitree, whoever. They all need the same components.
Nvidia made $567 million from robotics last quarter. Jensen Huang calls it their “biggest opportunity after AI.” They’re projecting $5 billion by 2026.
That’s 10x growth in two years.
My take
The valuations are insane.
$39 billion for Figure AI.
$7 billion for Unitree’s planned IPO.
But the infrastructure is real:
Factories are being built
Production lines are being installed
Robots are actually working in factories
Consumer units are being sold in China
The pattern I’m seeing:
Viral videos create awareness
Venture capital floods the space
Valuations outpace reality
But... production catches up
Prices collapse
Mass adoption begins
We’re somewhere between steps 3 and 4.
The question isn’t whether humanoid robots will become mainstream. Morgan Stanley thinks there will be 1 billion by 2050 — roughly one for every person on Earth.
The question is: who captures the value along the way?
My bet: the picks and shovels. The component manufacturers. The companies everyone ignores while chasing the flashy robot headlines.
The entrepreneurs and investors who win are the ones thinking in decades, not quarters…