HUMANOIDS 2050
Humanoid robots: 14x by 2030, then $12 trillion in robot work.
Key figures
Today: about 13,000 to 16,000 humanoids shipped in 2025, more than 4 in 5 built in China.
By 2030, my call: the humanoid market grows 14x, from about $2.7 billion in 2025 to about $37 billion, above the analysts' $11 to 25 billion.
By 2040, my call: fewer than 1 billion humanoids, against Elon Musk's and Peter Diamandis' 10 billion.
By 2040, the economics, my math: $3 per productive robot hour vs $26 for a human worker.
By 2050, my call: $12 trillion a year in robot work, against the banks' $1.4 to 7 trillion.
Humanoid robots will be to physical work what the cloud and large language models are to office work.
In 2025 the world shipped about 13,000 to 16,000 humanoids. More than 4 in 5 came from China, led by AgiBot, Unitree and UBTech.
The money is in: Apptronik raised $520 million at a valuation above $5.5 billion, with Google among the lead investors. I am an investor in Apptronik.
Elon Musk says 10 billion humanoids by 2040. Peter Diamandis says the same, and talks a $50 trillion market.
The banks are calmer: UBS $1.4 trillion, Morgan Stanley $5 trillion, Citi $7 trillion by 2050.
I say it's bullshit. We will not even hit 1 billion humanoids by 2040. We cannot even power and build them.
Building 1.64 billion EVs to replace all dirty cars takes 28 years, even with every car brand at full EV production. And the power grid would explode. Production-wise, electricity-wise, it's all bullshit.
By 2030: the market grows 14x, from about $2.7 billion to about $37 billion, from demos into factories and warehouses.
By 2040: the crossover. My math: about $3 per productive robot hour against roughly $26 for a human worker.
By 2050: 1 billion humanoids, about 4,000 working hours a year each, at $3 an hour. That is $12 trillion a year in robot work.
Let the machines take the heavy, dirty, and dangerous work.
