XPENG Raises Over $900 Million to Mass-Produce Humanoid Robots
Chinese electric vehicle manufacturer XPENG has raised more than US$900 million for its robotics business, valuing the unit at more than US$6.3 billion.
According to XPENG, the financing represents the largest single-round private capital raise in China’s embodied AI industry to date. The funding will accelerate development and mass production of its humanoid robot platform, IRON, while supporting research into Physical AI models and global commercialization.
And this is much bigger than another robot demonstration.
XPENG is making a serious bet that the next major AI revolution will not simply live inside computers and smartphones.
It will walk, move, manipulate objects, work inside factories and eventually operate alongside humans.
Welcome to the age of Physical AI.
What Is Physical AI?
Generative AI systems such as ChatGPT operate primarily in the digital world.
They understand text, images, audio, video and software.
Physical AI takes intelligence one step further.
It gives AI a body.
Instead of simply answering a question, a Physical AI system can potentially:
walk through a factory
pick up objects
operate equipment
navigate buildings
interact with customers
perform repetitive physical work
respond to changes in its environment
Humanoid robots are therefore becoming one of the biggest battlegrounds in artificial intelligence.
And XPENG wants to become one of the leaders.
Meet IRON
At the center of XPENG's robotics strategy is IRON, its advanced humanoid robot.
IRON was designed with a human-like body structure that allows it to navigate environments originally built for people.
The robot reportedly has 76 degrees of freedom across its body, while each hand contains 21 degrees of freedom, giving it significant mobility and dexterity.
That matters.
A robot working inside a human environment must do much more than walk.
It needs to reach.
Grip.
Turn.
Balance.
Manipulate objects.
Use tools.
And eventually adapt to situations it has never encountered before.
XPENG believes the human-like structure of IRON could help it collect behavioral data from real-world environments and progressively improve its capabilities.
The Robot Has Serious Computing Power
IRON is not simply a mechanical machine connected to a remote computer.
XPENG has equipped the robot with three of its internally developed Turing AI chips, providing up to 2,250 TOPS of effective computing power.
TOPS means trillions of operations per second.
That computing power allows significant AI processing to happen directly inside the robot.
Instead of constantly sending information to a cloud server and waiting for instructions, IRON can potentially analyze its surroundings and make decisions locally.
That means lower latency and faster reactions.
For a chatbot, a slight delay may be annoying.
For a robot carrying something heavy, operating machinery or moving beside a human, milliseconds can matter.
XPENG Is Bringing Car Manufacturing Into Robotics
XPENG already knows how to manufacture complicated machines at scale.
It builds electric vehicles.
That experience gives the company an interesting advantage in humanoid robotics.
Modern electric vehicles already require technologies that overlap heavily with intelligent robots:
batteries
motors
sensors
cameras
AI chips
autonomous navigation
software
control systems
manufacturing automation
supply chains
XPENG is now attempting to transfer much of that expertise into humanoid robot production.
The company says it wants to apply automotive-grade manufacturing and quality standards to IRON.
And the target is aggressive.
Reuters reports that XPENG is aiming for production capacity of around 1,000 IRON humanoid robots per month by the end of 2026, with broader commercial sales in China and international markets expected in 2027.
If XPENG can achieve that scale, humanoid robots begin moving from laboratory experiments toward actual manufactured products.
Tencent, Alibaba and IDG Capital Are Betting on the Robot
This is also not XPENG funding the project alone.
The financing round was led by IDG Capital, with participation from Gaorong Ventures.
Chinese technology giants Tencent and Alibaba joined as strategic investors.
That investor list matters because it shows how seriously major technology and investment companies are taking Physical AI.
The money will reportedly be used for:
robot hardware development
AI software development
training Physical AI models
generating high-quality robotics data
building mass-production facilities
employee incentives
international expansion
XPENG will retain controlling ownership of the robotics business.
But There Is Another Side to the Story
The robotics investment arrives while XPENG's core electric vehicle business faces significant competitive pressure.
China's EV market has become brutally competitive.
XPENG must compete against Chinese manufacturers including BYD, NIO, Li Auto and others, while also facing Tesla in both Chinese and international markets.
XPENG recently issued a third-quarter revenue forecast that fell below market expectations, with Reuters reporting its U.S.-listed shares were down roughly 40 percent year-to-date at the time of the company's August results.
As of September 1, XPENG's U.S.-listed shares closed at approximately US$11.17, near the lower end of their 52-week trading range.
That creates an interesting contrast.
Investors are valuing XPENG's robotics business at more than US$6.3 billion, even while its automobile business operates inside one of the world's toughest vehicle markets.
XPENG is therefore not simply developing a robot.
It is potentially building its next major business.
Robots Could Become More Profitable Than Cars
XPENG appears increasingly confident about that possibility.
The Wall Street Journal recently reported that XPENG believes robotics could eventually become more profitable than its automobile business.
The company expects IRON to enter mass production, initially targeting commercial environments such as offices, retail establishments and other business applications.
That makes the economics of humanoid robots extremely important.
Cars are expensive to manufacture.
Their margins can be squeezed by price wars, materials, logistics and competitors.
Humanoid robots could eventually generate revenue not only from hardware sales but also from software, AI services, maintenance and continuous capability upgrades.
Imagine buying a robot today.
Then downloading new abilities tomorrow.
Warehouse assistant.
Security patrol.
Receptionist.
Inventory checker.
Factory helper.
Customer service agent.
That could create an entirely different business model.
The Real Gold Is the Data
There is another reason companies want robots operating in the real world.
Data.
AI models improve through data.
Chatbots learned from enormous quantities of text.
Image generators learned from enormous collections of images.
Humanoid robots require something different.
They need information about:
movement
balance
force
physical interaction
hand manipulation
objects
spaces
human behavior
This kind of high-quality real-world robotics data is far harder and more expensive to collect.
If XPENG eventually deploys thousands of IRON robots, each robot could potentially contribute experience that helps improve future models.
XPENG describes this as a:
Data → Model → Application flywheel
More deployed robots create more useful data.
Better data improves AI models.
Better models improve robots.
Better robots create more applications.
More applications mean more robots.
And the cycle accelerates.
But Humanoid Robots Are Still Not Ready to Replace Everyone
The hype requires some reality.
China is moving aggressively into humanoid robotics, but the technology still has major limitations.
Recent reporting from Reuters found that many humanoid robots remain significantly less capable than viral demonstration videos suggest.
Robots may perform impressive choreographed movements while still struggling with unpredictable factory environments, delicate manipulation and autonomous decision-making.
Traditional industrial robots remain faster, cheaper and more reliable for many specialized manufacturing tasks.
So the humanoid revolution is not happening tomorrow morning.
But billions of dollars are now being invested in making it happen.
Why Filipinos Should Pay Attention
This story matters to the Philippines.
Not because thousands of humanoid robots will suddenly arrive in Manila next month.
The bigger issue is the direction of technology.
The Philippines has built major industries around human labor:
BPO. Customer service. Manufacturing. Logistics. Retail. Hospitality. Overseas employment.
The first wave of generative AI began automating intellectual and digital tasks.
Physical AI targets something different.
Physical labor.
Once AI can reliably understand the world, walk through it and manipulate objects, automation expands beyond computers.
Warehouses.
Factories.
Hotels.
Restaurants.
Hospitals.
Shopping centers.
Airports.
Construction sites.
Homes.
The conversation changes from:
"Can AI do my computer work?"
to:
"Can AI physically do my job?"
That is a much bigger transformation.
The Next AI Race Has a Body
For the past several years, the global AI race has focused on who could build the smartest large language model.
OpenAI.
Google.
Anthropic.
Meta.
DeepSeek.
Alibaba.
But another race is accelerating underneath it.
Tesla is building Optimus.
Figure is developing humanoid robots.
Unitree is rapidly expanding.
Chinese robotics companies are receiving enormous investment.
And now XPENG has secured more than US$900 million to push IRON toward mass production.
The next generation of artificial intelligence may not remain trapped behind a screen.
It may have cameras for eyes.
Motors for muscles.
AI chips for brains.
Hands capable of manipulating objects.
And eventually, enough intelligence to understand the physical world around it.
