From Tesla to Figure AI, billions of dollars are pouring into humanoid robotics. But behind the spectacular demonstrations, anthropomorphic machines are still far from fulfilling their promise.
They were the stars of the latest Consumer Electronics Show (CES). They? Humanoids, of course. At the major consumer electronics show, held from 6 to 9 January in Las Vegas, these human-shaped robots literally invaded the aisles of the world’s largest tech show. Among the most impressive was CLOiD. Developed by the Korean firm LG, this wheeled robot butler, designed for domestic tasks, demonstrated its ability to fold laundry, tidy up, fill and start washing machines, and even prepare breakfast. A little further on, the G1, from Chinese firm Unitree, tried its hand at a few boxing matches against humans during the show, while the X2, from Agibot, gave a Tai Chi demonstration. Do these prototypes foreshadow a future world where humanoids will mingle with humans? The prophets of Silicon Valley firmly believe so, arguing that robotics is currently experiencing its ‘moment’, just as there was the ‘iPhone moment’ in 2007 and the ‘ChatGPT moment’ in 2022.
The best way to convince yourself of this shift into the era of intelligent robots is to listen to Elon Musk’s technomessianic sermons. In May 2025, at the US-Saudi Investment Forum in Riyadh, the controversial boss declared that in the near future, “everyone will want their own personal robot. You can imagine it’s like owning your own C-3PO or your own R2-D2, but better”. And this humanoid that everyone will be clamouring for will, according to the billionaire, be none other than Optimus, the model developed by Tesla – “the most important product in history”, as claimed by Elon Musk. The entrepreneur elaborated on his thoughts at Tesla’s Annual General Meeting on 6 November, during which he claimed that Optimus would be “the best surgeon”, capable of “eliminating poverty” and “multiplying the economy tenfold, or even hundredfold”. In short, a robot that can do everything, ushering in a global golden age.
According to Musk, production of Optimus will begin “with a bit of luck” at the end of 2026. Tesla is currently building a production line with a capacity of one million units in Fremont. A second line, with 10 times the capacity, will follow. “The initial production rate of Optimus will be excruciatingly slow, but will eventually be incredibly fast,” Elon Musk warned at the end of January on X. Tesla is not the only company betting on humanoids – there are reportedly hundreds of projects around the world, including nearly 200 in China alone. Among the most advanced is Boston Dynamics, now owned by South Korean car manufacturer Hyundai.
The company unveiled the commercial version of its Atlas industrial robot at CES on 5 January and has begun production. All models to be manufactured in 2026 have already found buyers, and the company will only accept new customers from 2027 onwards. On our side of the Atlantic, the Norwegian-American company 1X launched pre-sales of its ‘domestic robot’ in the United States last October, with deliveries scheduled to begin in 2026. For $20,000 – or a monthly subscription of $499 – the humanoid NEO can now live in your home!
There is similar excitement in China, where manufacturer Unitree, for example, is already offering its G1 humanoid robot for sale at a competitive price of $13’500. As for XPeng, last November it unveiled Iron, which it plans to bring to market by the end of 2026. In a note published in June 2025 entitled ‘Is the world ready for 1 billion robots?’, UBS estimates that the number of humanoids will explode between now and 2050, rising from 2 million units in 2035 to 300 million in 2050. Meanwhile, analysts at Bank of America Global Research, in a study published in April 2025, predict a population of 3 billion humanoids in 2060 – or one robot for every 3.3 humans if the planet has 10 billion inhabitants by that date, as projected by the UN.
In light of all these advances and promises, capital is flooding in to support the promising humanoid robotics sector. California-based startup Figure AI, one of the most prominent in the sector, announced last September that it had raised more than $1 billion in Series C funding, valuing the company at more than $39 billion, to develop its general-purpose humanoid robot. Its Pennsylvania-based counterpart, Skild AI, raised $1.4 billion in January 2026, valuing the company at over $14 billion; and Norway’s 1X raised $24 million in 2023, then $100 million in January 2024. As for Tesla, the company that was worth more than $1 trillion, its valuation now depends more on robotics than on its automotive business, which is nevertheless its current source of revenue. Taking advantage of investor enthusiasm for the sector, Chinese companies Unitree and Agibot are expected to go public in 2026.
Science fiction
However, we should be wary of getting carried away. Most of the analysts we consulted remain cautious: “I think it would be very premature to invest in this sector right now,” warns Karen Kharmandarian, CIO & portfolio manager at Mirova. “Even if a few signs, such as Tesla’s Optimus, are emerging, we are still a long way from a functional humanoid robot in a real environment. It remains science fiction. It is difficult to imagine humanoids becoming widespread in the short term.” Pieter Busscher, portfolio manager at Robeco, points to some figures: “The industry could see very rapid growth in humanoid robot production, with a CAGR in the double or even triple digits, but this is because the current market is extremely limited.
Annual humanoid production is expected to reach one million units by 2030.” According to ABI Research, the global market for humanoid deliveries will skyrocket by 138% per year between now and 2030, but will only reach $6.5 billion by that date. The firm predicts that the take-off will come in 2027 – the year in which, according to its estimates, 150,000 humanoids are expected to be delivered worldwide. “The hype surrounding robotics is driven by the promises made about humanoids. This has pushed the valuation of certain companies to new heights, without there currently being sufficient value creation.” points out Nicola Tomatis, president of the Swiss Robotics Association. “Expectations are now extremely high, much higher than what current technologies can deliver. On the positive side, the hype is attracting significant funding for technology development. On the flip side, a speculative bubble has formed around robotics, and when expectations are disappointed, its bursting could hurt the entire sector.”
Unveiled with great fanfare at CES, the CLOiD robot butler, for example, is still a long way from being commercially available. It takes several minutes to take a product out of the fridge. That’s better than Aidol, the Russian humanoid robot – which collapsed after a few steps during its presentation in Moscow in November – but not enough to convince potential customers. As for Tesla, the company is regularly accused of overestimating the true performance of its robot, or even of ‘cheating’. In 2024, for example, Optimus impressed the crowd by walking through them at the We, Robot trade show. Alas, a few days later, it emerged that the robots were being remotely controlled by operators, something that Tesla had been careful not to mention to the audience. The same thing happened again in December 2025 with the publication of a video taken at an event in Miami, leading internet users to believe that Optimus was not as autonomous as it claimed to be, even though in this specific case, it is still unclear whether the machine was remotely operated or not.
A speculative bubble has formed around robotics, and when expectations are disappointed, its bursting could hurt the entire sector
The same was true at 1X Technologies: at the futuristic presentation of NEO in October 2025, some of the machine’s movements had to be remotely controlled by a human. Total autonomy is not yet a reality. Why is there such a gap between promises and reality? “While artificial intelligence has undergone phenomenal development in recent years, bringing AI out of a virtual universe and into the real world in humanoid form is proving extremely complex,” explains Karen Kharmandarian, CIO & portfolio manager at Mirova.
This view is shared by Amir Ben Ammar, mechanical engineer and co-founder of the French company AI Robotics: “Personally, I think humanoid robots will remain in the laboratory or in demonstrator form for a long time to come. Their development will require billions in investment, because these machines consume enormous amounts of energy and there are still many challenges to overcome before they can be functional, particularly in terms of interactions between all the parts as well as with the outside world.” This is especially true as mechatronics is progressing more slowly than AI.
“Technically, current robots, especially humanoids, are still far from matching human dexterity, mobility and endurance. Fine motor skills, tactile perception, actuator performance and battery life remain limiting factors in real, unstructured environments,” points out Matthias Röser, a partner at Bearing-Point. “Economically, cost is a major obstacle. Humanoid robots with complex functionalities remain very expensive, raising questions about their short-term return on investment and limiting their deployment beyond pilot projects and basic use cases.” Does this mean we should bury the humanoid dream? “Elon Musk is talking about 2026. But we know he’s always very optimistic about timelines. In 2025, we saw a lot of humanoids operating in labs or during demonstrations. I think it will be the same in 2026 and 2027,” says Humberto Nardiello, fund manager at DPAM. “Then prices will start to fall and we could see an explosion in their use, but not before 2028.”
