Humanoid robots are coming. recently?

This is it Step backa weekly newsletter covering one essential story from the world of technology. For more on robotics and artificial intelligence, follow Robert Hart. Step back It arrives in our subscribers’ inboxes at 8 a.m. ET. Subscribe to Step back here.

We’re taking winter break! Stepback will return on January 11, 2026. Meanwhile, you can catch up on previous issues here.

I have a weakness for videos that fail on the bot. I watch them repeatedly, and laugh to myself, as a form of therapy. Maybe I’m a sadist, maybe I need to get out more – you can judge me later – but they catch me every time. So naturally I was attached to Clip of Tesla’s Optimus robot falling like a downed tree At the company’s Autonomy Visualized event in Miami that went viral on social media this week.

According to the footage, Elon Musk’s humanoid robot was dispensing water bottles from behind a table before knocking over a bunch of them, flailing its arms upward, and collapsing backwards like a doll whose strings have been cut. Watch closely and you’ll see two things: a small column of water as his arm crushes a bottle in mid-fall (it made me laugh), and a movement eerily similar to someone removing a VR headset.

This wouldn’t be the first time Tesla has done this Forged independent part For its autonomous robot, which Musk says is key to the company’s future. The early demonstration was just that Dancer wearing a tight bodysuit To show what Tesla’s robot, now Optimus, does. He could He is. Subsequent demonstrations were revealed (rather clearly by the sound of it) Humans are convincingoperating robots remotely using what looks like a VR headset, which we know as Tesla Uses in development.

It was humans Obsessed with robots for centuries; From ancient tales of stone golems and artificial machines to… Modern science fiction, self-driving cars, and RoombasWe were fascinated by the idea of ​​animating inanimate objects with something life-like. Much of the current hype around humanoids can be traced directly to Musk, so it’s reasonable to be skeptical when he and others promise they’ll revolutionize the world. The musk he has Pledged to build an ‘army of robots’ Of a million people, it has a well-deserved reputation as a weird and unreliable forecaster, and robots have weathered more than a few waves of hype in their history. In the past, technology has always lagged behind the enthusiasm of those eager to bring AI into the real world, but today, we’re told that the technology is finally ready to go.

So what does preparing for delivery in 2025 actually look like?

There’s definitely a gold rush of sorts for humanoid robots right now. Every major tech company has it on their roadmap in some capacity, and the like Nvidia, dead, Softbank, Google, Amazon, Microsoft, Inteland Tesla (duh) are both putting a lot of weight — and money — behind them as the next big frontier in technology. And they’re not alone: ​​there’s a growing constellation of competitors who want to get in on the action, including Apptronik, Boston Dynamics, Figer AI, and 1X.

China wants to enter too. Beijing has decided that embodied AI — which also includes non-human creatures such as drones, quadrupeds, and other autonomous machines — is the key to future economic growth. In its trademark style, it has moved to position itself as a world leader in robotics through massive investments, government directives, and government subsidies. Everyone from the tech giants Ant group And Baidu to startups such as Unitree and AgiBot are accumulating.

If you follow everyone’s demos in China, you’d be forgiven for thinking that the future of humanoid robots has already arrived. This summer, the robots competed in dance, fighting and track and field events initially World Robot Games In China. A similar event was held in Greece – the International Human Olympiad — in the birthplace of the Olympics. Human battles are more popular than I imagined and they seem to happen absolutely everywhere, from Organized competitions and Underground fight clubs to Sparring with CEOs.

Companies are also keen to bring robots out of factories and into homes. These kinds of human-centered spaces are why advocates say humanoid robots are worth pursuing rather than other robot body types that might be easier to produce. Figure company said that its new robot Figure 03 can do household chores, Video release A robot washes dishes and folds clothes. 1X For the first time Neo, which it claims is safe and “the world’s first consumer-ready humanoid robot.” There too videos New Shakily completes basic household tasks. I’m not sure I can afford it, but if you’re interested, the 1X sells units for $20,000, with US deliveries starting next year.

But actual use cases remain minimal, and great demos are nothing like working products. For all the progress – and there has been impressive progress – demos are being organised, written or even played remotely. For example, while it is technically correct to say that Ant Group’s R1 was Cooking for a crowd At a trade show this year, it moved at a very glacial pace Meryl Streep impressed Miranda Priestly. Likewise, the idea of ​​getting the 1X’s Neo seems a lot less appealing once you realize you also have to stick with it Having random people teleport into your house To control it remotely. Likewise, robotic sporting events are so fun to watch because the contestants are rambunctious, uncertain, and unpredictable, not because they are skilled athletes.

All of this raises an obvious question: If the technology clearly doesn’t fit into the hype, why have so many people gotten so excited about humanoid robots? Well, for the first time in a long time, it’s not crazy to think that technology has started to catch up.

Historically, it has been extremely difficult to get robots to perform even mundane tasks that humans do without thinking, such as walking or picking up a glass of water. Hardware – but mostly software – was a limiting factor, and machines were confined to highly structured environments and very narrow tasks. Advances in artificial intelligence are changing that and the development of turbocharged robots.

The extraordinary advances in AI image and text generation from companies like OpenAI and Google are largely due to the large language models, or LLMs, that power chatbots. These work by using complex networks to recognize and reproduce patterns in the vast amounts of data they are trained on. Larger AI models are trained on everything they can pull from the Internet, and the resulting systems are able to operate in a generic way, rather than following strict, hand-coded rules. For robots to function in the real world, they need this kind of generalization, and roboticists are trying to use the same approach to give machines the flexible understanding of the physical world they need to survive in it.

The challenge here is data. The type of materials used to train LLM holders were widely available online, but obtaining the data needed to train robots was not so easy. Robots need detailed examples of things moving in the real world, something that does not exist on a large scale. Companies are working hard to change that, and there is a major effort to create the kind of data needed to train robot models at scale. On the surface, this sounds ridiculous: Tesla has workers Wear cameras and sensors To teach Optimus to act like a human – but it works. It’s also why companies like 1X are deploying less autonomous robots into people’s homes, as they do Gives them the opportunity To collect the data needed to create autonomous robots by performing complex tasks remotely, such as loading the dishwasher or cleaning something difficult.

Human beings have become more affordable too – Especially in China – As hardware costs decline and economies of scale begin to be achieved. While prices vary widely – entry-level units can cost as little as $1,400. Chinese model Bomi To about $13,500-$20,000 for models from unitary Or 1X, while industrial models can cost more than homes – they are approaching prices that some consumers can afford, meaning more robots are being deployed in the real world. This creates a feedback loop, as companies get more data to work with, and build better models and better robots that more people might want, assuming they’re willing to welcome them into their homes.

But even with all this progress, it may still be just noise. In November, China’s leading economic planning agency Be warned that the robotic bubble may be brewingwhich contrasts with the huge number of companies, the volume of investment, and the absence of viable use cases. The truth is, they’re not independent yet, so other than hobbyists and researchers, who in their right mind would buy one? If I wanted to hire someone to clean my house, I could hire a cleaner without the high upfront cost and save my money until someone produces a robot that can actually do the job I bought it for.

Until companies stop hiding behind catchy promotional videos and running so-called autonomous robots, it will be difficult to know where we stand. Maybe the robots are finally here, or maybe I have more fail videos coming my way. I guess time will tell, but I’ll make the popcorn.