Optimus Might Go On Sale As Early As Next Year, But It Won't Save Tesla
Who is falling for this?
Do you remember Optimus, also known as Tesla Bot? It was Musk’s answer to Boston Dynamics humanoid robots, except less capable. That isn’t hyperbole. Boston Dynamic’s robots can do assault courses and backflips. Meanwhile, Optimus can’t even fold a shirt without human help. Despite this, Musk has claimed that the potential demand for Optimus is in the tens of billions of units, as it could massively disrupt the jobs market by replacing flesh-and-blood humans. So, when Musk recently announced that this dystopian robot could go on sale to the public as early as next year, it created a stir, to say the least! So, could this be the shot-in-the-arm that Tesla needs? Or is Musk grasping at straws?
Let’s start with what Optimus is and what Musk and others think it is capable of.
Optimus is a humanoid robot powered by the same computational hardware that Tesla’s FSD runs on. The AI behind the robot is also very similar to FSD, incorporating the same computer-vision system. Musk claims this enables Optimus to undertake repetitive or dangerous tasks that previously would have taken humans to do. Think factory line operator, janitor, lumberjack, or ironworker. But, because Optimus is based on hardware that is already being produced at scale, Musk claims Tesla could sell it for as little as $20,000 per unit!
This is why Optimus has such a vast potential. Rather than paying, say, an ironworker $50,000 a year and having to deal with their holidays, need to sleep, and constant injury rate, just buy an Optimus for $20,000 and have it do the same job 24/7. As such, Musk has publicly stated he thinks Tesla could sell 20 billion Optimus robots! Or 2.5 for every human alive. He isn’t alone; Morgan Stanley predicted that Optimus could disrupt 30% of the global labour market. Needless to say, this could send Tesla’s profits and share price to the moon. In fact, Goldman Sachs predicts that Optimus alone could add $150 billion to Tesla’s stock price in the next fifteen years.
So, you can see why Musk’s promise that this robot could be on sale next year is a huge deal. That is, if all of these promises weren’t absolute fantasy. You see, in reality, robotics and AI simply can’t do this.
Firstly, humanoid robots make no sense. They are chronically inefficient. Take the idea that Optimus will do cleaning work. Automatic vacuums, dishwashers and self-cleaning windows are much more efficient in terms of price, materials and energy usage than Optimus. Yes, a single Optimus could do all these jobs, but even all these robots combined will still be more cost, energy and material efficient than Optimus. Partly because you still need to buy a vacuum for Optimus to hoover, but also because of how complex robots like Optimus are. This inefficiency issue is true for pretty much every single job you could get a humanoid robot to do. Moreover, robots that focus on specific and constrained tasks, like the automatic vacuum, always deliver way better and significantly more consistent results than general-purpose robots like Optimus. This is why car factories use articulating arm robots, not humanoid robots.
But, form factor issues aside, Optimus simply can’t perform the tasks Musk is claiming.
Firstly, driving a car is a far more constrained task than, say, a humanoid robot undertaking ironwork or lumber jacking. There are fewer variables, more apparent decisions, and easier-to-execute tasks. As such, it is significantly easier to create an AI for self-driving cars than for a competent humanoid robot. And you may have noticed that Teslas are still miles away from being self-driving. Current data suggest Tesla’s FSD is 10 times less safe than a human driver, and that is with the FSD being supervised by a human! As such, it’s no surprise multiple governmental agencies are investigating Tesla, along with many class action lawsuits. So, if Tesla still can’t crack self-driving cars, how can they create a functional, competent humanoid robot?
But even far simpler and highly constrained AI/robotics can’t replace human workers. I know I keep talking about it, but Amazon’s “just walk out” grocery stores were a perfect example of this. An AI would track customers around the store, and aided by shelf sensors, it would tot up what items the customer has taken. This way, rather than going to a checkout, the customer can simply walk out with the goods, and their Amazon account/card will be charged for what they have taken by the AI. This is a simple, constrained AI with no complex robotics, very few variables, incredibly obvious rules, and exceptionally easy-to-execute tasks. Yet, the AI got it so wrong so often that Amazon had to hire a small army of remote workers to monitor and correct it. The cost of hiring these AI nannies became so expensive that Amazon ultimately dropped the “just walk out” concept.
Now, Amazon arguably has better AI talent than Tesla, as many AI companies rely on their services. So if they couldn’t get this to work, how can Tesla get Optimus to work?!
Musk could use brute force and train Optimus’s AI on an excruciatingly large dataset to overcome this issue and make it capable and reliable enough. However, the computational power and energy cost to undertake this (we are talking over the hundred billion dollar mark) will make Optimus far too expensive to be viable. There is also a question as to how Tesla could acquire such data. It’s not like ChatGPT, which can easily scrape high-quality data from the web, or FSD, which can use data from the millions of Teslas driving around the world. Creating this monster dataset is likely utterly unfeasible. To top it off, simply feeding the AI more data is not guaranteed to solve this problem. As I have covered before, we are starting to hit diminishing returns regarding AI training. As such, it is possible that AI could never be good enough to enable a usable humanoid robot!
So, no, Optimus going on sale next year won’t be the shot-in-the-arm Tesla needs. It will never be capable enough to make a meaningful dent in the job market, and if it becomes capable enough, it will be more expensive than simply hiring a human. So why are Musk, Goldman Sachs and Morgan Stanley predicting it will? Simple, they are all massive shareholders in Tesla, and its stocks are sinking faster than Musk’s reputation. This is a blatant smoke-and-mirrors attempt to bolster the stock and stem their losses. How pathetic.
Thanks for reading! Content like this doesn’t happen without your support. So, if you want to see more like this, don’t forget to Subscribe and follow me on BlueSky or X and help get the word out by hitting the share button below.
Sources: Electrek, Will Lockett, Will Lockett, Manchester University, Amazon, Planet Earth & Beyond, Planet Earth & Beyond, Planet Earth & Beyond