If you're reading this on a desktop, we suggest you open the PDF-version, but if on mobile phone - continue reading below.
Could general-purpose humanoid robots become the next big product category in society and yield multiple trillions of dollars in economic value? Most likely, this is an inevitable future in the very long-term. When general-purpose robots become more capable than humans in performing basically any task you could think of, our global economy will transform in ways still unfathomable for most people. Many traditional product categories will be eliminated, and trillions of human labor cost will transform into robotic labor cost. For investors, being early and very precise in which vertical they attack – will be paramount.
When Steve Jobs unveiled the iPhone almost 15 years ago, few of us imagined a future where the mobile phone industry would be completely transformed, enable a massive application economy, and obliterate a long range of product categories that we nowadays haven’t thought of in many years. We don’t really have physical calendars anymore, or calculators, or digital cameras. The iPhone came to be a general-purpose device for anything useful in our digital lives. An infrastructure for anyone creating a digital application.
In the coming 20 years, my bet is that we’re going to see the same thing also taking place in the physical world, both in terms how we transport ourselves in society and in respect to any type of physical labor. Or in other words, both the car and general-purpose robots, could become “iPhones” (platforms) on which a next-generation manufacturing startups can place their applications and aspirations.
For robots in specific, they’ll be able to connect their physical limbs to specific applications, like a welder, vacuum cleaner, surgical equipment, or kitchen appliances, and perform almost an infinite number of tasks. Theoretically, it’ll be much more efficient to have a few general-purpose robots that apply a lot of applications to conduct work in factories, shops, homes, or on our streets – instead of having all products becoming robots.
Why buy many products to solve many problems, when you can buy one (or few) products to solve the same problems?
All throughout my career, I’ve learnt that in order to fully capture the future value of disruptive innovations, you need to ask the really hard questions and think in much bigger terms – than you think you should. This was true when analyzing the potential proliferation of the automobile, the invention of the transistor, the explosion of the internet, the emergence of e-commerce, and current adoption rate increase of electric vehicles.
In hindsight, you’ve (almost) always been rewarded if you asked, “what would happen if this new thing got proliferated to the fullest extent, and who might be some of the winners and losers in such case?” Going forward, this will no doubt be the case for thousands of future disruptive innovations as well.
And for general-purpose robots specifically, the magnitude of societal impact could be bigger than anything we’ve ever seen before. Prospectively, the entire human labor market, currently worth more than USD 40tr (according to Figure), could be up for grabs. As well as a significant share of the USD 16tr manufacturing market. These are such mind-blowingly big figures and shares of global GDP (at around USD 110tr), that if general-purpose robots only take one-fourth of this combined figure (due to massive human resistance and regulation, for example) – it would be a staggering USD 12tr. To put this in perspective, it would be double that of global e-commerce, and five times bigger than the global automobile market.
Reference to Figure (robot maker): “Today, manual labor compensation is the primary driver of goods and services prices, accounting for ~50% of global GDP (~$42 trillion/yr), but as these robots “join the workforce,” everywhere from factories to farmland, the cost of labor will decrease until it becomes equivalent to the price of renting a robot, facilitating a long-term, holistic reduction in costs. Over time, humans could leave the loop altogether as robots become capable of building other robots — driving prices down even more”.
10 BILLION ROBOTS IN A STEADY STATE?
There are plenty of projections out there in terms of number of general-purpose humanoid robots (and other shapes), spanning from a few hundred million units, up to 10 billion. And in terms of price per unit, which currently is impossible to project because we’ll most likely have an extremely wide array of different robots with significant price differences.
But if we assume a USD 10k per-unit price and a very long-term steady state of one robot per person globally, some 40-50 years from now, along with a replacement rate of 10 years (total lifespan in other words) – it could generate an annual output of roughly USD 10tr.
This is certainly not a projection of mine (only laying the groundworks for discussion) – but a grand size of the prospective market value of general-purpose robots, should be considered a very viable potential future. The reason for this, is that all companies on earth always want to lower costs and provide a better service than competitors. If having general-purpose robots in their factories leads them to accomplish these things, it should be considered a no-brainer, to bet on a substantial market for this product category.
So, we’ve established that if general-purpose robots are much more productive and capable in solving tasks compared to humans, there is a case to be made for this novel product becoming the biggest product category so far in history – at least outside of the real estate in which we live and shop, and the fixed infrastructure we base our industrialized economy on top of.
iPHONES OF THE PHYSICAL WORLD
And if we take this thesis to a deeper level, not only in terms of magnitude as a whole, but in terms of transformation for all that is manufacturing and consumption – we arrive at the notion of humanoid robots potentially becoming the “iPhones” of the physical world.
In my view, these robots will be purchased and placed in basically any type of location and be asked to do an almost infinite number of tasks. They will vacuum our floors, cook our dinners, paint our walls, weld our equipment, and conduct surgery on our bodies.
However, I’m not saying that the makers of the robots will embed all of the physical applications mentioned. I’m actually saying the opposite, that it will be the current application makers and new startups – that predominantly become the application layer.
In other words, we’ll have the general-purpose infrastructure layer, and we’ll have the application layer. Now, this might not feel like all too transformative compared to what we have today. Because we already have these two layers. But the thing is, the current structure for a lot of manufacturing sectors and vectors, is based on very traditional supply-chains, business models, distributors, and singular problem solving.
To a very large majority, we have physical products (across the board, with everything from trucks and construction equipment, to vacuum cleaners and lawn mowers) that are made by companies that makes these products for the specific problem they’re solving for their customers. And this is super rational. Companies should be focused on the specific problem they’re solving. This has made some companies very big in their respective fields, meaning substantial market shares and global competitive advantages. But things could change…
INCREASED COMPETITION AS A CONSEQUENCE
Currently, it is extremely difficult for budding entrepreneurs to build manufacturing companies and set out to compete with major corporations that boast both supply chains and distributive capacity, that simply are too big and complex to shake. They’re both the backends and the frontends of their respective fields.
Consider many of the dominant media companies of the 1990s. There were mighty TV-networks, newspapers, music record labels, and video rental companies. They were both the backends and the frontends. However, once information converted from physical parts to digital bits, it upended their strongholds on their respective markets.
During this disruption, and frankly the cause of it, were US big tech companies that developed digital backends (cloud infrastructure) for anyone to build their frontend on top of (digital applications). Prior dominant brands got to feel the burn of newcomers not having to build their backends. Companies like Youtube, Facebook, Instagram, Twitter, and so forth, enabled any person to build their own media company (content creators, in other words).
My thesis is that many of the physical product categories that we today take for granted and have come attached to for several decades, might start feeling the burn of new backend manufacturing companies stepping onto the manufacturing scene and enabling physical application companies to easier build their products and reach their customers. They’ll become physical “Manufacturing clouds” for anyone to employ.
For example, developing and manufacturing a product application won't need to have an electric motor within (because it’s inside the general-purpose humanoid robot it’s attached to), and it can piggyback on the robot’s distributive capacity. Theoretically, this would make it a lot easier to build and scale a manufacturing company.
IMAGINING A FUTURE STATE IN 2035
Imagine a day in the life of your future self. The year is 2035 and you own your very own general-purpose humanoid robot. This robot was shipped to you with a plethora of accessories, just like when buying a car today. You chose to add vacuum cleaning, general tools, kitchenware, as well as standard features which includes picking up, placing, closing, the physical surrounding. But compared to today, the brands for the products acting as the application layer – are no longer the brands we’ve grown accustomed to in present day.
Instead, there’s been an explosion of entrepreneurship in the manufacturing sector, enabled by the emergence of general-purpose robots – both on the frontend and backend of things. Robots have made it so much easier to start a manufacturing company anywhere on earth because they’ve broken the cost-arbitrage between low-wage and high-wage countries.
And furthermore (we’re still in the year 2035), autonomous mobility has exploded and enabled the start of a (partial) shift from ownership to usership of physical products. AI has started to enable this business model shift, which later could lead to a massive shift from offshoring, to reshoring. At this point in time, we’ll have seen a few hundreds of major factories moving back to high-wage countries – to be as close as possible to where the consumption takes place. The core reason for this is that general-purpose robots has made it cheaper than human labor costs in currently low-wage countries.
AI ENABLES A TROJAN HORSE FOR CIRCULARITY
Essentially, when AI embeds itself into the physical fabric of our global economy, it holds the potential to disrupt a lot of traditions; including the past few decades of offshoring, traditional assembly-light manufacturing, many single-utility product categories – and the start of replacing the linear economy with a circular one instead.
In my view, autonomous mobility, general-purpose robots, and AI-invented renewable materials, are the three massive components that will enable AI to lead to a Trojan horse for reversing our global pollutive footprint. Because it enables a globally scaled incentive of designing and making physical products so that they last as long as possible, are refurbished, and are renewed to the very same quality and product at the end of life.
The core reason for this, is that when physical products become robots, meaning that they can conduct a utility by themselves for a customer, and then move to another customer by themselves and conduct the same utility – it enables the shift from ownership to usership. From buying units, to buying access instead. Think Software-as-a-Service, but in the physical world. Effectively, this shift promotes durability and renewability, because those are the most economic things to prioritize. This change in incentives, is the Trojan horse, for actual sustainability.
In my humble view, after having researched the topic extensively, many different sectors, and historical disruptions, for quite many years – the solution to pollution will not be politics. Instead, it will be a new equation governing our global economy. It’ll be a new, globally scaled, change in incentives to make products durable and renewable. Or in other words, circularity instead of linearity. Because, as Albert Einstein said: “Politics is for the present, but an equation is for eternity”. He’s been correct in an astronomically long list of things so far, so I wouldn’t go against this foundational analysis of human behavior
MARKET ACTORS WITHIN GENERAL-PURPOSE ROBOTS
But let’s now get back to how general-purpose humanoid robots could upend the current backend and frontend manufacturing complex. And specifically, why there should be a strategy as an investor to really start paying attention to this space. Because pricing of the future always moves much earlier and faster than when the future arrives. And this goes both on the up- and downside…
In the next few years, we’ll able to read thousands of new research reports , startup pitch decks, articles, quarterly reports, and other types of analysis – boxing in the potential of general-purpose robots. These reports will span everything from actuators, motors, batteries, AI-models, torque sensors, to body materials (among many more areas) – trying to figure out where the biggest upsides will be if investing into this space. This will surface some of the most intriguing investment opportunities out there, and of course increase their valuations.
Currently, there are upwards of 50 serious contenders, making general-purpose humanoid looking robots. Most of these stem from the US and China, predominantly because these countries usually are the most innovative and boast the by far strongest capacity when it comes to investments into AI chips and models, but there are a few from other countries as well.
In this research piece, you’ll be able to read a list of some of them, and a longer list of smaller sub suppliers.
SELECTED MARKET ACTORS
Below is a selection of companies present in the general-purpose humanoid robot space. These companies include Figure, Boston Dynamics (owned by Hyundai), Tesla, Agility Robots, Apptronik, Unitree Robotics, Sanctuary AI, UBTech, and 1X Technologies.
MAIN THESES (SUMMARY)
1) TRILLION-DOLLAR MARKET; BIGGEST PRODUCT CATEGORY EVER
When general-purpose robots are much more capable than humans in performing both general and specialized tasks, the roughly USD 40 trillion global labor market will meet a new type of mechanic competition. Most of, if not all, manufacturing, retail, restaurant, logistical (etc.) companies on earth will want to employ robots that are cheaper and more productive than humans. Not binarily, but a partial conversion. As such, there is a case to be made for a multiple trillion-dollar industry forming in the next few decades.
2) DISRUPTION OF MANY SINGLE-UTILITY PRODUCTS
When robots are extremely capable of conducting a wide array of tasks, it will enable viable strategic roadmaps for makers of general-purpose robots to build a new platform-type product category. Similar to how the iPhone set off a digital application economy, robots could set off a physical application economy (products connected to robots). My bet is that building a platform and generating take-rate sales from applications, will be sought after.
3) ENABLING THE SHIFT FROM OWNERSHIP TO USERSHIP
When most products in society are able to become robots, meaning that they can move and conduct tasks by themselves, it enables customers to buy the utility (service) of the product – not solely the product unit itself. For many product categories, this could rearrange their respective markets substantially, and usher in novel usership platforms.
4) RESHORING: FACTORIES CLOSE TO CONSUMPTION
When selling physical products as a service, the economics of manufacturing physical products completely flips. Servicing, refreshing, upgrading, and refurbishing, become more relevant. When general-purpose robots reach economic critical mass, vs. human labor costs in low-wage countries, it could (together with autonomous mobility and renewable materials) unleash a globally scaled reshoring effect of manufacturing capacity.
5) CIRCULAR ECONOMIC INCENTIVE
Today, the core economic incentive on a global basis is to sell as many products as possible, with an optimized lifespan (not maximized), and not design it for full renewability. This is the core linear economic incentive that governs our global economy. However, when selling a product as service instead (vs. units), the most economic thing to do could change in favor of eternal revenue recognition (durability) and renewability. Politics will not yield the fundamental change we need on a global basis, a new circular equation (incentive) will.
6) REPUBLICAN US LEADERSHIP = PROTECTIONISM = AUTOMATION
In the next four years, the coming Republican leadership in the US will most likely raise tariffs on imports for a variety of product categories and countries. US companies will bare these tariffs and most likely strengthen company management discussions on higher investments into automation in general, and protectionism on an aggregate. General-purpose robots could, as a result of the change in leadership, gain traction faster.
7) MANUFACTURING STARTUP WAVE
When general-purpose robots are capable enough (things will take time), they will make it easier for a next generation of manufacturing startups. Lowered cost-base, speed, and overall stronger supply chains, could be beneficial factors of raised automation.
8) MANUFACTURING CLOUDS
If we extrapolate general-purpose robots (together with autonomous mobility and renewable materials) it poses the potential for physical “Manufacturing clouds”, that similar to digital clouds – enables anyone to create a manufacturing company simply by opening an account, designing a product, and letting the Manufacturing cloud make it.
This marks the end of the writing part of this research letter, however below you'll find a couple of rather extensive lists of market actors in the general-purpose space, along with notable events and product visuals.
Thank you and until next time,
Christopher Lyrhem
Chief Future Officer
Sircular
SELECTED MARKET ACTORS
Below is a selection of companies present in the general-purpose humanoid space, along with a handful of relevant meta data such as a description of the company, country of origin, when it was founded, example of product name(s), some of the investors of the company, what sectors they are targeting, and their respective websites.
Do note that these are just a selection of all the aspiring general-purpose humanoid (and other shapes) robot companies out there. The intriguing thing with this space is that we’re witnessing the dawn of one of the biggest product categories ever, including the formation of its’ value chain, company formations, partnership, etc.
Source: OpenAI Search, Google, respective companies' websites (subject to mistake)
SELECTED SUB SUPPLIERS (small companies)
Below is a selection of companies present in the general-purpose humanoid space, but as sub suppliers, and in particular smaller companies. The table includes a brief description of each company, their origin, notable news items, and selected investors.
Source: OpenAI Search, Google, respective companies' websites (subject to mistake)
ADDITIONAL ROBOT VISUALS
Below is a selection of compelling visuals, depicting various general-purpose robots, and their makers’ visions of the future. These images span work in factories, warehouses, tunnels, kitchens, and laundry rooms, etc.
NOTEABLE EVENTS DURING 2024
In 2024, the humanoid robotics industry experienced significant advancements, including substantial investments, technological breakthroughs, and strategic collaborations. Below is a compilation of notable events taking place in 2024 so far.
Figure AI: In February 2024, Figure AI secured USD 675 million in Series B funding, elevating its valuation to USD 2.6 billion. This round attracted prominent investors, including Microsoft, OpenAI Startup Fund, Nvidia, and Jeff Bezos. The company unveiled Figure 02, a humanoid robot designed for manual labor tasks in manufacturing and logistics sectors. Trials at BMW's South Carolina factory demonstrated Figure 02's capabilities in automobile assembly tasks.
Tesla: The company announced plans to commence sales of the Optimus humanoid robot in 2026, following internal testing within Tesla's facilities. Optimus is envisioned to perform repetitive or hazardous tasks, aligning with Tesla's broader automation goals.
Boston Dynamics: The company introduced a fully electric version of its Atlas robot, showcasing enhanced agility and autonomy. A demonstration video highlighted Atlas performing tasks such as moving engine covers and navigating complex environments without human intervention.
Agility Robotics: Agility Robotics' humanoid robot, Digit, gained traction in manufacturing and logistics. Notably, GXO Logistics employed Digit at USD 30 per hour.
Unitree Robotics: The Chinese company expanded its product line with the H1 humanoid robot, designed for dynamic movements and human-robot interaction. Unitree's robots are increasingly utilized in research, logistics, and entertainment sectors.
1X (formerly Halodi Robotics): Rebranded as 1X, the Norwegian company continued to develop its EVE humanoid robot, focusing on applications in security, healthcare, and service industries. The company attracted investments from OpenAI and Tiger Global Management.
Physical Intelligence: This San Francisco-based startup raised USD 400 million in early-stage funding from investors like Jeff Bezos, OpenAI, Thrive Capital, and Lux Capital, achieving a valuation of USD 2 billion. The company aims to develop universal AI software for robots, enabling them to perform diverse household tasks.
Apptronik: Apptronik introduced Apollo, a versatile humanoid robot designed for collaborative tasks alongside humans in dynamic environments. The company secured investments from Capital Factory and Grit Ventures to accelerate development.
Xiaomi: In August 2024, Xiaomi unveiled the SU7, a humanoid robot designed for household assistance and companionship, marking the company's foray into the field of humanoid robotics.
XPeng Robotics: At the 1024 Tech Day event in October 2024, XPeng introduced its second-generation humanoid robot, featuring over 60 joints in its hands for enhanced dexterity, and announced plans for a flying car project, highlighting the company's commitment to integrating automotive and robotics technologies.
CloudMinds Robotics: In September 2024, CloudMinds launched the XR-2, an upgraded version of its humanoid service robot, equipped with advanced AI capabilities for applications in healthcare and hospitality sectors.
UBTech Robotics: UBTech introduced Walker X, an advanced humanoid robot designed for tasks such as visual quality inspections and assembly, targeting industrial applications.
Sanctuary AI: The Canadian company announced successful trials of its general-purpose humanoid robots in various work environments, demonstrating their versatility and human-like intelligence.
OpenAI: OpenAI expanded its focus to robotics, collaborating with Figure AI to integrate advanced AI models into humanoid robots, enhancing their capabilities in natural language processing and decision-making.
Nvidia: Nvidia announced Project GR00T, a general-purpose foundation model for humanoid robots, aiming to enable them to learn skills and solve tasks using natural language and video input. This initiative reflects Nvidia's commitment to advancing AI in robotics.
Comments