How Figure AI and the Global Race for Physical Intelligence Could Reshape Every Dollar of Human Labor on Earth
Humanoid robots that actually work in the real world are no longer a science fiction fantasy — they are walking factory floors, cleaning living rooms, and hauling parts inside BMW plants right now in 2026.
That single reality is what sits at the center of one of the boldest bets in technology history.
Brett Adcock, founder and CEO of Figure AI, has raised nearly $2 billion at a valuation of $39 billion to build robots that can do everything a human can do, from picking up objects to reasoning about what comes next.
And the number he keeps coming back to is not a small one.
A little under half of the entire world’s GDP is human labor.
That means the total addressable market for humanoid robots is not measured in billions.
It is measured in tens of trillions of dollars.
If you are someone who builds income online and wants to understand where the next decade of money is going to flow, tools like AmpereAI and ReplitIncome are already helping everyday creators and builders ride the wave of AI automation before most people even realize the tide has shifted.
We strongly recommend that you check out our guide on how to take advantage of AI in today’s passive income economy.
Table of Contents
Brett Adcock’s Big Claim — And Why He Might Be Right
Brett Adcock is not a first-time founder who stumbled into robotics.
He scaled a software company, sold it, then founded Archer Aviation, took it public, and led all engineering and aircraft design from the ground up.
When he left Archer to start Figure AI, the move looked risky to outsiders.
But to Adcock, it was obvious.
He had been watching the humanoid space for two decades and kept seeing the same pattern repeat itself.
The wrong hardware choices, the wrong software approach, or companies treating it like a research hobby instead of a commercial product.
Boston Dynamics had Atlas, a hydraulic humanoid that dripped oil, lasted about twenty minutes on a charge, and was so large and dangerous that it could never stand next to a human being on a factory floor.
Adcock believed the world needed a group willing to treat humanoid robots like a real product built for real deployment at real scale.
So he self-funded the company from day one.
Within four months, Figure AI was burning one million dollars a month and had a forty-person team working one hundred hours a week.
From Zero to BMW — How Figure AI Got Robots Working in the Real World
A Small Batch, a Big Lesson
The first real commercial test for Figure AI came through a partnership with BMW.
A small batch of Figure 1 robots went into a BMW facility and ran every single day for six months.
The robots worked.
They learned a tremendous amount from that deployment, and Adcock says the experience forced them to completely rethink how they commercialize their software and AI systems.
That process led directly to Helix 2, their second-generation AI model for robotic control, which they launched in early 2025.
H3: What the Robots Are Doing Today
By 2026, Figure robots are cleaning living rooms, performing commercial tasks, handling parts, and operating with a level of autonomy that would have looked impossible just a few years ago.
The robots are not teleoperated.
Nobody is sitting behind a joystick guiding every move.
The system runs end-to-end on AI models that reason, observe, and act in real time.
Figure AI designs everything in-house — motors, rotors, stators, sensors, joint structures, battery packs, kinematics — the entire vertical stack.
Adcock believes this is the only way to control your destiny in hardware.
If a supplier has a problem and you do not understand every part of your own system, you are stuck.
Figure AI is never stuck.
The OpenAI Breakup That Went Viral
One of the most talked-about moments in robotics in recent memory came when Adcock publicly confirmed that he had fired OpenAI from a collaboration agreement.
H3: How the Partnership Started
OpenAI led Figure AI’s Series B funding round.
They brought in Satya Nadella and Microsoft as co-investors.
As part of the deal, Figure and OpenAI agreed to collaborate on developing next-generation AI models for humanoid robots.
For a period, the two teams worked closely together on how language models could function inside a physical robotic system.
H3: Why It Ended
After roughly a year of collaboration, Adcock’s internal team — most of whom came from robot learning research backgrounds spanning over a decade — was outperforming OpenAI on every relevant benchmark.
They were better at training the models, better at testing on actual robots, and better at understanding what the AI needed to do inside a physical body.
Adcock made the call to end the relationship.
He fired OpenAI.
The comment went viral, and it is not hard to see why.
But the decision itself was not emotional.
It was strategic.
Figure AI’s edge is in physical intelligence — the ability to train models that interact with the real world in real time, not models that answer text prompts in a browser window.
If you are building income streams that depend on AI, tools like AmpereAI are designed specifically to help you stay ahead of exactly this kind of rapid AI shift, putting smart automation directly in your hands.
The $10 Trillion Revenue Argument — Is It Real?
Breaking Down the Numbers
Let’s look at this clearly and honestly.
Global wages paid to workers in commercial environments total somewhere between thirty and forty trillion dollars every single year.
If humanoid robots can perform even a significant fraction of that work, and if they are sold at scale like any other industrial product, the revenue potential is staggering.
Adcock’s argument is straightforward.
Tech companies trade at ten to twenty times revenue.
If Figure AI eventually generates hundreds of billions of dollars in annual revenue, the business is worth ten trillion dollars or more at standard market multiples.
That is not a fantasy number pulled out of thin air.
That is a multiplication problem applied to an already-known market.
H3: The Production Ramp
Figure AI hit record production in March 2026.
Adcock’s target is to triple that output by May 2026.
The parts to do it are already in-house.
The commercial demand exists right now — Adcock says he could fill customer orders for robots immediately if only the robots were ready to operate fully autonomously at scale.
The bottleneck is not demand.
The bottleneck is reliability.
Getting robots to run seven to ten hours a day, every day, without failure and without human intervention, is the hardest unsolved problem in the industry.
What “General Robotics” Actually Means
Adcock talks about a goal he calls general robotics.
Imagine a robot inside a bodysuit shaped like a human body.
You can talk to it.
It looks at you.
It reasons about its environment.
It understands what it sees.
You drop it into any location — a warehouse, a kitchen, a hospital corridor — and it figures out what needs to be done, then does it.
No custom programming per environment.
No human supervisor watching every move.
Just a machine that generalizes the way a human worker does on their first day at a new job.
That is the goal behind Helix 2 and whatever comes after it.
Figure AI wants to be the first company to demonstrate AGI — artificial general intelligence — operating inside a physical body in the real world.
That is a bold claim.
But they are the only company that currently has robots running inside commercial customer facilities every single day at meaningful scale.
For online builders who want to understand how AI is reshaping income, ReplitIncome offers a practical entry point into the world of AI-powered income tools built for the current era of automation.
The Security Arms Race Inside Figure AI’s Offices
Spies, Drones, and Tinted Glass
Running a company that is designing one of the most commercially valuable technologies on the planet comes with unusual problems.
One afternoon at Figure AI’s Bay Area offices, a staff member looked up and noticed a drone hovering outside the top corner of a window, pointing directly into the main workspace.
Nobody ever found out who sent it.
After that, Figure AI tinted all of its glass and implemented strict physical and digital security protocols.
The company’s engineering CAD files and software are locked down with serious cybersecurity measures.
Visitors who wander into the open office areas see robots and hardware on display, but the really sensitive systems are behind layers of access control.
Adcock describes the Bay Area as an environment full of competitive intelligence gatherers.
He is not wrong.
When your robots are potentially worth trillions of dollars in future revenue, everyone wants a look.
AmpereAI is built on the same principle — protecting the advantage of early movers in the AI economy by giving builders access to automation tools before the crowd catches on.
Competition, Optimus, and Why Adcock Is Not Worried About Tesla
When asked directly whether he is worried about Tesla’s Optimus robot project, Adcock’s answer was immediate and precise.
This is not a manufacturing problem.
This is an intelligence problem.
Tesla is extremely good at manufacturing.
Elon Musk has built some of the most efficient production lines in the history of automotive and energy industries.
But building a robot that can learn to do useful autonomous work across unpredictable environments is a fundamentally different challenge.
It requires training data from real robotic deployments.
It requires iterating AI models on real physical systems.
It requires years of learning what breaks, what fails, what the robot does not know how to handle.
Figure AI has that runway.
They have been doing this since 2022 and have already shipped commercial robots into BMW facilities.
They have already rebuilt their entire AI model strategy based on what they learned in that deployment.
Adcock believes Figure AI is a few years ahead of every other team globally at this level of humanoid development.
Whether that lead holds is the central question of the next five years.
The Personal Cost of Building Something This Hard
One Hundred Hours a Week and Home by Six
Brett Adcock goes home every evening in time for dinner with his children.
He is back working by the time they go to sleep.
He is at the office until midnight on a regular basis.
He gave up annual golf trips.
He stopped going to dinners for old friends passing through town.
He does not attend trade shows or sit on panels.
Every hour that is not with his family is pointed directly at Figure AI, Cover, or Hark — the three companies he is currently running simultaneously.
Cover is building terahertz imaging radar systems originally developed at NASA’s Jet Propulsion Laboratory to detect weapons on students entering schools from a distance of ten to twenty meters.
Hark is an AI lab building next-generation personalized intelligence models and new hardware devices designed specifically for interacting with AI — because Adcock believes phones and laptops are twenty-year-old interfaces that were never built for the AI age.
He describes his biggest risk clearly: the funnel of the hardest, most difficult problems he has to solve every day is never empty.
Getting a robot to run autonomously for seven to ten hours without a single failure is a problem nobody has ever solved.
And he is trying to solve it while simultaneously scaling production to thousands of units.
That is what tens of trillions of dollars worth of ambition actually looks like in practice.
If building at that scale feels out of reach for most people, tools like ReplitIncome exist precisely to give everyday builders a real way to earn from the AI revolution without needing a billion-dollar war chest.
Why Humanoid Robots Are the Most Important Business of Our Lifetime
The Meta Problem in Robotics
Adcock uses one phrase over and over again when explaining why he left Archer Aviation to start Figure AI.
He calls it the meta problem in robotics.
If you can solve the humanoid robot, you solve the biggest business in the world by a large factor.
Not the biggest robotics business.
Not the biggest AI business.
The biggest business — full stop.
Because a humanoid robot is the only machine in history that can walk into any environment built for a human being and do what a human being would do.
No special equipment.
No custom conveyor belts.
No redesigned facility layouts.
Just a robot shaped like a person, doing a person’s job, in a world already built for people.
Every factory on earth.
Every warehouse.
Every retail store.
Every hospital.
Every home.
All of it is already designed for human-shaped workers.
And AmpereAI is one of the tools helping forward-thinking builders tap into the AI income wave that is rising as all of this unfolds — putting smart automation tools in the hands of people who are ready to move early.
What Comes Next — Scaling Into the Millions
The Road From Thousands to One Million Units Per Year
Figure AI’s near-term target is thousands of robots shipped this year to multiple commercial customers.
The announcement of those customer relationships is expected within ninety days of this writing.
After that, assuming deployments go well, Adcock plans to compound the scaling aggressively.
Tens of thousands, then hundreds of thousands, then the stated goal of one million units per year.
The company already has commercial demand that exceeds its current production capacity by a wide margin.
Adcock says if the robots were all ready right now, he could place them with commercial customers immediately.
The gap between supply and demand is not a demand problem.
It is an execution problem.
And Jeff Bezos — one of Figure AI’s investors — is not the kind of person who backs companies that cannot execute.
For anyone looking to build income using AI tools while this revolution unfolds, ReplitIncome and AmpereAI both offer ways to participate in the AI economy at a level that does not require a robotics lab or a nine-figure funding round.
The humanoid robots era is not coming.
It is already here.
The only question is whether you are watching it happen or building something with it.
We strongly recommend that you check out our guide on how to take advantage of AI in today’s passive income economy.
