Like a DJ wrapping up a particularly tight set, the Tesla Optimus robot reached to its ears and removed an invisible pair of headphones…and then keeled over, revealing, it seems, its teleoperation and, perhaps, peeling back the curtain on the fiction of many cutting-edge humanoid robotics. It also makes clear what the immediate future of robotics holds, and it’s neither as awesome nor as bad as you were expecting.
That incident, caught on video when numerous Tesla Optimus robots were set up as engaging bartenders who served drinks and chatted you up, but were all likely controlled by headset-wearing teleoperators, demonstrated what many people think about most of the current bleeding-edge humanoid robots like Optimus, 1X Neo Bot, and Figure 03 from Figure AI: the coolest stuff they do is mere fiction.
If there was any question that Optimus uses teleop for their robots. Here one clearly has a guy take the headset off and it falls over.Absolutely hilarious though. pic.twitter.com/4gYVohjY00December 8, 2025
Entering the uncanny valley
Broadly speaking, I don’t expect 2026 to be the year of humanoid robots. They won’t be ready for the home, in part because they’ll still lack the requisite skill to match or surpass what we can do on our own.
However, while I’ve seen robot expectations pumped up to unrealistic levels, others think we got some dose of reality,
“In many ways, 2025 was the year that expectations about humanoid robots started coming down to Earth,” Brian Heater, Managing Editor for A3 (Association for Advancing Automation), told me via text.
A longtime colleague and friend, Heater has been covering robotics for decades and is now part of the industry (he also has a newsletter and a podcast on the topic). So he’s developing an insider’s view on the future and development of consumer and manufacturing robots. There are still companies that are, he told me, “bullish about the [humanoid] form factor, but conversations have taken a more pragmatic shape with regard to efficacy and timelines.”
Heater is not the first to tell me that many robot companies are considering swapping wheels for legs. Obviously, it’s harder to teach a robot to walk (and not run into mirrors) than it is for them to wheel around. Over the years, I’ve seen many smaller consumer bots that balance on two wheels (the original Segway is one).
So the first prediction for 2026 is that while we’ll see more development on the humanoid robot side, there may be other companies that move faster, especially in the manufacturing space with robots that feature humanoid or at least dual-arm bodies on top and wheels on the bottom.
Humanoid robots will creep slowly into homes (Heater sees the deployment “at very limited scales”), but as I’ve written before, few consumers will be ready to shell out $20,000, even for a glimpse of our humanoid robot future.
Remote control
If anything, 2026 will mark the year of the home-teleoperated robot. Early reports on the 1X Neo Betas indicate that most of what it can currently do is heavily reliant on teleoperation. Tesla Optimus robots, which do not have a clear timeline for home deployment, are also, it seems, more Christian than Cyrano, requiring remote control to keep the fiction of ability going.
The question for 2026 humanoid robot customers is how comfortable they are with a slow-moving, incredibly expensive product that requires almost constant remote access by a third-party company to perform even the most basic tasks.
Heater explained that one of the chief obstacles standing between these robots and wider adoption is mobile manipulation/dexterity.
“There’s not nearly enough data to train these systems,” wrote Heater, “UC Berkeley’s Ken Goldberg calls this the ‘100,000 year data gap,’ referring to the millennia of information that have been used to train LLMs, versus what physical AI currently has to go off of.”
Let’s go to the videotape
Heading into 2026, we’ve been fed a steady diet of incredible robot capability videos, showing them running, dancing, flipping, and even performing Karate moves. It set unrealistic expectations for our near robot future.
Heater even has a term for this: “This is due, in part, to a phenomenon known as Moravec’s Paradox. Boiled down, it holds that some things that are relatively simple for humans are complex for AI and robots, and vice versa…Just because you see a robot do a backflip doesn’t mean it possesses the motor skills to tie a tie (it likely doesn’t).”
1X, Tesla, and Figure AI will continue making promises and videos depicting humanoid robots doing amazing things, but few consumers will buy them in 2026. Those that can afford them will very quickly escort them to a closet where they will sit, untouched in the darkness until archaeologists unearth them a millennium from now.
One company I do not expect to disappoint is Boston Dynamics. The humanoid robot pioneer has never sought to sell its bots to consumers and is far more transparent about its work; a recent video about why the all-electric robot stands up the way it does is especially illuminating. As they explain the video: “Humans kind of stand up without thinking about it, but robots really need to think.”
The ones who get it
In 2026, Boston Dynamics will unveil even more incredible robot athletic feats that will surely inspire robot fans and the industry, but it will be years before it sells Atlas to anyone except researchers.
Obviously, Boston Dynamics will also continue to update its Spot robot, which has shown up in factories and municipalities. That robot faces increasing competition from companies like Unitree.
2026 will see far greater use of robots of all kinds in manufacturing, fueled, in part, by improvements in safety. Heater told me, “Top firms in the space (and my current employers at A3) have been working on a new standard for these systems to allow them to more safely work alongside humans outside of a fenced-in environment.”
This means that instead of more videos of humanoid robots quietly working in closed-off spaces on repetitive tasks, we might finally see some robots working safely alongside humans who have no fear of harm, accidental or otherwise.
March of the mini bots
Leaving aside humanoids and factory bots, there will be a whole range of home, entertainment, and work robots, mostly building upon previous work, but with AI energizing innovation and capability.
Training simpler robots in a small set of tasks is easier than ever thanks to AI-supported visualization, which can help robots prepare for the unknown and even learn on the fly based on previous training and on-board models.
So, yes, robots in 2026 will not be as wild and exciting as you hoped, but it should set the stage for an important decade in robotics development.
As for that Tesla Optimus robot lying on its back, don’t worry. As soon as its operator put his headset back on, I’m sure Optimus returned to serving drinks, making small talk, and inspiring the next generation of roboticists.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.
https://cdn.mos.cms.futurecdn.net/BdRfN7dNnWpYhFmuSByJ7G-1920-80.jpg
Source link
lance.ulanoff@futurenet.com (Lance Ulanoff)
