The robot (an autonomous vehicle) brought a robot (a humanoid) which brings everything as a service (it can bring specialized robots with it).
Need your garden weeded? Your windows or solar system cleaned? Your floor vacuumed? Your groceries ordered? Your laundry folded? Your home secured? All that, and more, will be offered to you as a service.
It’s why Figure AI’s founder, that just landed $675 million to build its humanoid robot, says the total addressable market is “the largest on the planet.”
Long before it shows up on your doorstep, though, the humanoid robots will take over many tasks inside factories, offices, stores, and other businesses. Why there first? Well, workers will put up with a weird movement or two and the business will see a business need to fork over the $100k or more to buy the first ones. Most consumers can’t afford more than $200 a month for a robot. Not to mention payments for a $100,000 robot! So consumers will need to wait until manufacturing scales up to a point where the price can be reduced to less than $25,000 per robot.
Even in the factory it will bring a new kind of service. Say they are being used in an auto repair shop, we could see a “SnapOn Service.” Your humanoid robot could bring you any tool you need from SnapOn. The most common ones kept in a tool chest in the shop, like they are today, but less common ones, a robot ride away.
If a rare car comes in for, say, a new transmission, your robot could prepare the shop for the new job by driving to a distribution center and picking up some rarer tools. Most jobs are like this, if you know a radical change is coming you can prepare for it ahead of time like if a music festival is coming with 150,000 attendees like Coachella does. There Uber prepared for the festival by building a parking lot ahead of time to deal with the 3,000 vehicles it moved to the Palm Springs area to serve the festival goers. A rare vehicle coming into your repair shop is the same kind of change: an unexpected, and rare, event.
Now that Apple has killed its autonomous vehicle project, we are taking a fresh look at the robotics industry and we see that Tesla is uniquely positioned to succeed here. Why?
Tesla
Consumers trust and love Tesla. Its customers are fanatical about the company. It makes batteries, electric motors, and AI computer hardware and AI software itself. At scale! Few other companies can match it in manufacturing expertise. No one else in the humanoid robot business can make vehicle-sized things at scale (in the millions) the way Tesla can. Nor do they have the financial ability to support a sizable R&D effort for years. That’s an exclusive club with members like Apple, Google, Meta and Nvidia. If you have $100 Billion in your pocket, you can play! But none other than Tesla has the distribution advantages of an existing car fleet. And they definitely don’t have the data advantage that fleet brings. And only Tesla and Apple have the brands it would take to bring a robot into peoples’ homes worldwide.
Apple is one of those few and would have been the only other company that could compete, but they found out the hard way consumer electronics isn’t the same as heavy manufacturing.
Apple, while they kill it when it comes to personal computing and phones, has fallen behind when it comes to its crown jewel: Siri. Most folks wouldn’t consider Siri to be Apple’s crown jewel. Most would say the iPhone. We say different. Siri is what links all Apple’s products together. But it’s been neglected! Siri was cutting edge once. It isn’t now. And Apple hasn’t made any major stride in consumer AI for 5-10 years. No one looks to them for what is to come in AI. That will change at WWDC come June. A new multi-modal and multi-model Siri arrives in June. Apple may get some ground back from ChatGPT, but it’s not enough to match Musk in the race to build the best AI robotics.
When we host or join X’s audio spaces we hear this over and over again: that consumers would welcome a Tesla robot into their homes, while they don’t feel that warmly to other potential companies. Which brings up another reason to believe in Tesla’s robot: Elon’s other companies. X, for example (formerly known as Twitter) is about to turn on a sizeable update to its AI, Grok.
If we did have a humanoid robot, it will be Grok that you talk to, even if it’s called something else. Plus, once in your home, the robot will need to continue the conversation with you, which includes topics consumers care about like news, sports, entertainment.
But it is the voice data that Elon Musk has from X Spaces that will take X’s Grok to be competitive with OpenAI’s Whisper API, which does voice to text and back again. Whisper will be hard to beat. We tested it in many noisy places including a loud dance floor at a wedding, and it’s the best at understanding humans that we’ve seen so far.
The winner in AI is usually the one with the most data. Tesla, along with X, has the most data of human audio conversations. It will lead to a robot that will understand humans better than others, which means acceptance in more homes.
It is Tesla’s coming Robotaxi network that we see as a huge advantage over others. Why? Because, as we laid out in our book, “The Infinite Retina,” having a humanless-vehicle network will dramatically lower the price to move these robots around town, which means a few customers might be able to share a humanoid robot since the price will be high at first.
But we see the Robotaxi network itself as the way that Optimus, Tesla’s humanoid, will be introduced to the world. One day you will be picked up by a Tesla to take you to the airport. In the driver’s seat is an Optimus, which can get out, help you with your bags, or your wheelchair/strollers, and then help the car drive you to the airport.
Help drive?
Yes, Optimus will have eyes that are better than most of the cameras in its fleet as of early 2024, and can provide redundancy, which will greatly increase safety. It can also look into spots where the existing cameras have a tough time seeing, like around a blind turn.
Tesla is training its robot to drive, we learned from a Tesla employee, which makes sense, right, since it’s based on the same digital twin the car is driving on.
Many customers will try a Tesla the first time this way, getting a ride at an airport, or home from a bar, like they do with Uber/Lyft today. But the robot will behave like a human, can high five you, and talk with you.
These customers who meet Optimus will see how competent it is at helping them with luggage, or driving a car. It will increase their love of, and trust of, the Tesla brand. To the point they will look into getting one for their home, or be much more accepting of it bringing a stack of pizzas to your front door.
Most people today can’t see inviting such a contraption into their homes. Worries about privacy, safety, security frequently are mentioned by consumers we talk to in doing our research. But Tesla owners already say they would invite Optimus into their homes. That’s a way to introduce millions of people to a humanoid robot.
This ability to “distribute” the humanoid robot to businesses and then homes, and having the trust to be accepted inside, even bragged about to others, is a large moat we don’t see others getting across.
See, once the humanoid robot gets accepted inside the home, maybe to set the table at first for your pizza party, but later to do things like dishes, clothes, general repairs and upkeep, security, and entertainment, the first robot inside the home brings other of these specialized robots.
Housekeeper robot of the future
Already there are specialized robots that vacuum, clean windows, blow snow, pick weeds, and more but these come with new troubles for humans. The window washing robots, for instance, need to be manually moved from one window to the next. Many homes have dozens of pieces of glass, so that’s quite a chore. One that a humanoid robot would be happy to do for a fee.
Window washing as a service just arrived in that home.
Because the humanoid is a generalized robot it should be able to turn literally everything else in the home into a service:
Groceries as a service.
Laundry as a service.
Meals as a service. (And a lot cheaper and more consistent than the ones where a human delivers).
Gardening as a service.
Security as a service.
It gets better for Tesla. As it is asked to perform laundry as a service it will quickly know what your favorite brand of detergent is for washing your clothes. What if the bottle of Tide is empty? Well, you asked the robot to wash your clothes and fold them for, say, $100 a month. That means the robot gets to replace your Tide with a brand it prefers. Or, gets to bulk buy Tide which changes the economics for Procter and Gamble. And not in a good way.
- Amazon, which bought Zoox, and has millions of robots inside its warehouses already.
- Figure, which just raised $675 million for its humanoid.
- Waymo/Google. Now expanding into more of California’s population centers.
We consider competitors as:
- Have enough cash flow to have a sustained R&D effort to complete not just a humanoid robot, but the infrastructure to enable a robot to be directed into people’s homes and keep track of tasks/services provided.
- Have a fleet of vehicles or a plausible ability to get such within three years (millions of vehicles).
- Have consumer-ready brand/trust to be welcomed into the home.
To get a better understanding of the everything-as-a-service market, we visited Electric Sheep, a Silicon Valley-based company that is making a variety of robots to do landscaping work. It bought a landscaping company to get the customers and employees to train the robots. Unlike other grass-cutting robots, theirs are generalizable. That means they don’t need a wire around the lawn to show the robot where it can and can’t go: it learns that by “seeing” the lawn, just like a human.
Humans are still needed to get the robot off the truck and to do other tasks that their robots can’t yet do (which are, for now, mostly focused on cutting grass, even though there are many other things that their landscaping company does).
Electric Sheep already is doing “lawn mowing as a service” and is prepared for when humanoids get good enough to do the other jobs. Founder, Naganand Murty told us that he sees a humanoid robot that’s good enough for his tasks to be a decade away. He decided to make a two-wheeled robot that can push around specialized tools, which they introduced last week, named “Verdie.” He says the AI job of training such a robot is a lot easier than building a humanoid and can do lawn care tasks better than a humanoid will be able to do for years.
Others, like Eric Jang, founder of 1X Technologies, another humanoid robot manufacturer, said on a post on X that all AI roads lead to robotics. and that the industry could get there a lot faster than that.
It is the ability to bring other robots to your businesses or homes (like Electric Sheep’s Verdie) that unlocks the huge total addressable market and brings us all into everything as a service that has both us, as well as many investors, excited.
Just Three Things
According to Scoble and Cronin, the top three relevant happenings last week
Humanoid robotics startup Figure receives $675M in Series B funding, valued at $2.6B
Figure received funding from Microsoft, OpenAI Startup Fund, NVIDIA, Bezos Expeditions, Parkway Venture Capital, Intel Capital, Align Ventures, and ARK Invest. Its humanoid robot, called Figure 01, has already been employed by BMW for an automotive manufacturing facility and is designed to mitigate workforce shortages and manage tasks that are too dangerous for human workers. Figure is facing competition from Tesla with their Optimus robot and Boston Dynamics. Ars Technica
Smart glasses from Apple? A possibility
According to an article from Bloomberg’s Mark Gurman, Apple is mulling whether or not to offer smart glasses like those from Meta and Amazon. It’s also considering designing a fitness ring and AirPods with cameras. We think that Apple producing smart glasses will serve as a bridge to their Apple glasses that may be offered around 2028. Bloomberg
Google announces Genie which can transform a single image into a virtual game
Google has developed something extraordinary with Genie. It’s in research preview mode and illustrates playable worlds could be generated from a single image, sketch or text prompt. Although it’s wonderful, it’s still quite crude looking. It’ll take more time to develop into a more elegant and easily orchestrated tool. Tom’s Guide.
