I want to introduce you to the largest economic category since the Internet.
This isn’t something I found through industry reports. I stumbled into it speaking with business owners, noticing the same story repeated back to me again and again.
Our economy is a big machine. Gears and cogs work together to make things happen. In large part, it’s the labour of people, the cogs, that cranks out all the stuff we use.
Human labour is in a curious state at the moment. Let me paint an uncomfortable picture for you using three converging trends.
Aging population - most people are aware of this threat. For those who are unfamiliar; in short, as the world population becomes older, it’s unclear who will fill the physical jobs to keep the products/services we need flowing.
Silent Workforce Crisis - slightly lesser known, employers of the hard job (repetitive, dangerous, or physically demanding roles) category are having a difficult time keeping their employees. A very hard time. After doing a consulting job speaking with close to 100 businesses, the phrase I heard most was “I cannot find enough people to work”. Turnover for warehouses in certain cases is 100%.
Minimum Wage Regulation - There is an impact to continually increasing minimum wages. Business owners are smart. When presented with a problem, they begin actively considering alternative (sometimes radical) solutions. In this case their growing balance sheet line item, labour. It’s come to a point for some corporations that automating people out of a job is cheaper than enduring ever increasing minimum wage amounts.
We have converging economic threats building pressure looking for a release valve. But to date, there has been no elegant solution, forcing businesses to continue to operate with inefficiencies.
That is, until recently.
Try and visualize some of the robots you’ve seen to date. Warehousing robots from Amazon, restaurant busboy “trays”, Boston Dynamics Atlas or Spot.
Amazon’s warehousing robots offer a ton of efficiency gains, but have to be segregated from humans because of their capacity to inadvertently maim them. Restaurant busboy trays are cute, but in interacting with them, they feel generally useless. Boston Dynamics’ Atlas literally does parkour - awesome, but where is the utility?
These are all novel products but they are not widely adopted.
Why?
Narrow in their functionality, they lack awareness and adaptability. They are clunky and in some cases dangerous.
The method used to program robotics until recently followed a rules-based approach. Meaning every possible scenario a robot will encounter is meticulously scripted in advance. Hundreds of thousands of lines of code direct the device to, for example, pick this box up here and put it down there.
Rules-based programming is hard, messy, and difficult to upkeep. The lack of nuance produces less-than-ideal results in real-world applications. What if a human walks in front of the robot carrying a box from here to there for example?
At the end of 2023, there was a shift in robotics programming. Large Language Models (think ChatGPT) began to be installed as the brain for robotic bodies, and given cameras to perceive the world. A trend labeled Embodied Ai. With these handy new tools, Roboticists began playing around with a new method of programming their robots.
In contrast to the rules-based approach, they used a Neural Net Training approach. A method of “teaching” a robot an operation rather than programming it. The model (the robot brain) instead learns by watching video. The training allows the robot to role-model tasks through the training video. Think of how a toddler models its parents' physical movement to walk. Neural Net Training is the concept Tesla is using to build its Full Self-Driving v12+ capabilities.
Combined, cameras for eyes, an LLM as a brain, and neural net training provide the robot with awareness of its surroundings, an ability to learn new tasks, and make nuanced decisions, much like humans do in the real world.
Brett Adcock Pace of Innovation
With this shift in approach, like the iPhone did for mobile phones, robots will now suddenly become incredibly useful.
The type of robot I’m talking about is important for context. The world we live in is designed by humans for humans. A human form factor is the best way to extract utility out of it.
Our hands are the primary reason we have evolved beyond any other animal in the animal kingdom. With hands, we grip and manipulate objects, make tools, and have the dexterity to make arts, crafts, and intricate designs. Humans can do awesome things, and in large part, it’s due to the little meat sausages sprouting from the ends of our arms.
The form factor of a general purpose humanoid robot, specifically its hands, give it the ability to interact with the world the way we do. Hands make a robot a Swiss Army Knife. Rather than building robots for a single function, as we do now, a single humanoid robot will have many functions.
Right now, general purpose humanoid robots with awareness are being built to perform tasks in warehousing and manufacturing settings.
Although the function of Figure-01 robots in BMW plants has not been released, it is speculated that the initial purpose of these bots is to perform a simple function like taking a piece of sheet metal from a stack and placing it into a press. A small job, but one that currently requires human labour input. BMW will no longer have to fill that function with a human.
Consider a simple economic example using labour in the manufacturing sector:
Let’s say a medium-sized manufacturer invests in robotics for their assembly line.
Pre-Robotics: The business operates with 100 human workers producing 1000 units per day, at a labour cost of $20 per hour for an 8 hour shift. ($16,000 in human labour)
Post-Robotics: The business is able to operate with 20 human workers and 20 robots, producing 2,000 units per day, with robots running 24/7 at a reduced hourly cost (factoring in maintenance and energy). ($3,200 in human labour, $1,095 in robotics cost)
Cost Reduction: Reducing human labor from 100 to 20 workers, even with the upfront cost of robots, would more than halve the labour costs.
Efficiency Gains: Doubling the output while potentially reducing production time and improving quality will significantly increase market competitiveness and revenue.
Impact: Assuming the manufacturer operates at a profit margin of 10% pre-robotics, efficiency gains, and cost reductions could potentially increase margins to 15-20%, substantially impacting the bottom line.
Robots offer the most compelling value proposition for the manufacturing sector. Not only to enhance productivity but to potentially redefine cost structures and revenue potential.
This is a simple example, in one sector. More sectors are prime for robot enablement, and more sectors will be invented for work that otherwise is not possible for us now.
I know what you’re thinking. The Figure-01 robot taking a sheet of metal from a stack and putting it into a press is not quite the revolution you were promised in the beginning.
Let’s add some colour to the picture.
The beauty of general purpose humanoid robots is that they will be a gift that keeps giving. Typically when we buy a piece of equipment it has a single function. If I buy a lawn mower, I expect that it will mow the lawn, not get up and make me a coffee. General Purpose Humanoid Robots offer a different paradigm.
Humanoid Robots will be able to acquire new skills through training, meaning their value will increase to the purchaser after the purchase. A bot may come out of the box with basic utility to justify the purchase - however over time, with proprietary training from the company, they will learn more and more skills. Companies will build training sets for their robot fleet, increasing their overall utility many fold.
Task by task, job by job, humanoid robots will perform the functions that humans are unable or unwilling to perform.
What does the incremental replacement of roles in these environments mean at scale?
Humans outsourced leg labour to horses before cars, we mastered agriculture allowing us the time to think up a whack of other ideas, and taught millions of computers to communicate with each other, forming a global memory mesh that envelopes the earth.
These were huge leaps in productivity, yet there are still a significant number of roles for which we require human labour.
The ability to replace humans in physical roles will generate nearly unbounded demand for humanoid robots. In the near term - hard, repetitive jobs. In the future, jobs that humans cannot perform.
Productivity and economic output will rise, price of goods will fall - we will see a level of abundance unlike anything humanity has seen to date.
It is anticipated that 1 billion humanoid robots will be in use by 2040. At that time, our economy, running at peak efficiency and scale, will look markedly different.
But, Robo butlers will not be at next year's CES, don’t get your hopes up. It’s tough to predict how this category will flesh out, but this is my best guess.
The design phase will ramp up incredibly quickly. We will see companies go tit for tat on X, showcasing ever-evolving product demos. Round one looked like this:
Figure-01 Makes Coffee
Optimus Folds a Shirt
Going For a Walk With Optimus
Google Aloha Cleans The Home
This is the beginning of the robot arms race.
The end of the design phase will be marked by working general purpose humanoid robot prototypes in warehouses or manufacturing plants. I.e. the robots will have reached a minimum viable utility rate.
I know what you’re thinking. Once there are working prototypes of a few general purpose humanoid robots floating around in warehouses - the production of bots will start to flow. Cool your jets, jefe.
The design phase will be easy relative to the manufacturing phase.
The single most difficult aspect of this category will be the ability to ship products at scale. Elon has repeatedly spoken about the challenges involved in production ramp-ups. Describing design as fun, and manufacturing as hell. I believe the man.
Think about the struggle legacy automotive manufacturers are having to produce EVs profitably. EVs are produced in a different way than combustible engine vehicles. They require different parts and a completely new method of manufacturing. It is not the same traditional model of sourcing parts from a supplier and assembling them together on the plant floor.
Similarly, robot parts cannot be purchased off the shelf at this time. Engineering of the parts, and the manufacturing of them, will be done in-house.
Those who can build manufacturing facilities that scale, and vertically integrate, will be able to meet that frothy, frothy demand.
Material in, Robots out. Figure’s in-house manufacturing.
Of course, the category will evolve, there are other players in the space now and more to come. I’ve compiled a list of all the companies focused on building general purpose humanoid robots below.
When manufacturing is solved, bots will roll out at scale, reshaping warehousing, manufacturing, agriculture, construction, and mining sectors. This hinges on robots' ability to perform complex tasks with precision, and their integration into existing workflows without significant disruptions.
The deployment will be rapid across the traditionally labor-intensive and physically demanding sectors. Tasks considered dangerous, dull, or dirty will transfer from humans to robots.
The impact on the businesses will be immense. Businesses utilizing the bots will operate 24/7, with a workforce that does not tire or need breaks.
Earnings calls for robot producers will look a whole lot like automotive companies do now, with a focus on the number of units shipped growing quarter by quarter.
With fleets of robots in the market doing hard jobs, the producers of these units will be ingesting a huge amount of data from the devices. Getting better and better every day. Kind of like how Apple tracks your usage on their devices and improves it.
With that data, the models, their hardware, and their software will be iterated on and continually refined. New training and design will allow the robots to take on more nuanced roles.
The next labour dominos to fall will be the undesirable ones - retail, and warfare.
In retail, robots will handle stocking, cleaning, and even customer service, improving efficiency and allowing human employees to focus on tasks that require a personal touch. Some organizations will opt not to have any human labour at all.
In warfare, humanoid robots will be used for reconnaissance, logistics, and support roles, reducing the risk to human soldiers. I wouldn’t be surprised to see Anduril take a crack at a humanoid robot over the next couple of years to be used on the battlefield.
Now that we've had adoption in hard jobs, followed by undesirable jobs - our bots will look a lot less like their version 1 predecessor from the Hard Jobs phase. They can operate in an environment with humans, their hands have been refined to the point where they can do all the things we do with ours. They are more adaptable to varied and unpredictable environments and interact more closely and empathetically with humans.
Producers have been manufacturing these units at scale, bringing the unit price down significantly. The price will be affordable for households, and the robots will be easily integrated into daily life.
You can breathe a sigh of relief, this is when you’ll see bots at CES. Caregiving and robot butlers for every home.
Humans get tired. We have bad days, we make mistakes. This is especially unfortunate if the human who is tired and making mistakes, is operating on you. The precision and reliability of humanoid robots will eventually make them indispensable in healthcare, particularly in performing surgeries.
Robots will be able to conduct complex procedures with a level of precision and steadiness that surpasses human capabilities, reducing the risk of complications and improving patient outcomes.
The models, their brains, having ingested thousands of hours of video surgery, and equipped with the complete knowledge of human medicine - will make real-time decisions and adapt to human anatomy.
Much like it will soon be seen as too risky for humans to drive vehicles manually, it will become unthinkable to have a human operate on a human.
Phases 1 through 6 will have dramatic improvements in our day-to-day life. But all the while, there will be people cooking up new ideas for these robots. Schemin’ and Dreamin’, is what humans do best.
They will ask the question: What are we not doing now, that we otherwise would if human life was not a factor?
The final phase of this evolution involves the emergence of jobs and roles that are uniquely suited to robots. Jobs that do not currently exist.
Tasks that require superhuman strength, precision, or endurance. Or, tasks that take place in environments hazardous to humans, like deep-sea exploration, space construction, and disaster response.
We will invent new ways to use these tools to extract more utility out of our environments.
The humanoid robotics category will be astoundingly large. Over the next few years, humans will quickly come to a point where robots are an indispensable part of our daily lives. This will not be an incremental improvement, but an order-of-magnitude leap. Our world will shift to one of abundance.
The category will be so big, there are billions if not trillions in wealth that will be made by companies and investors. The winning producers will be those who are able engineer/manufacture everything in-house at scale. The players will evolve, but at the time of this writing, there are two key producer front runners: Tesla and Figure.
Tesla has the best manufacturing expertise of any company on the planet. Their robot, Optimus, is currently walking around their factory floors.
Figure’s CEO Brett Adcock has experience manufacturing vehicles at scale with Archer Aviation, producing flying electric vehicles. They currently have contracts with large warehousing and manufacturing businesses to produce humanoids for their warehouse floors.
Additionally, both are excellent at marketing - an undervalued but important part of generating demand.
Right now, my gut tells me Tesla will be the winner in this category overall.
I’ve listed the full list of key players in the space as they are today below. Take a gander at some of the businesses who are building the beings that will shape our new paradigm.
I send emails only when I have something worth saying