Elon Musk has been promising the world a humanoid robot called Optimus for more than a year, but the two prototypes unveiled last week did not exactly dazzle with agility. The company’s most advanced robot—made with all Tesla components and close to production-ready, according to Musk—waved unsteadily before being shoved across the stage by three human helpers. “This means a future of abundance, a future where there is no poverty, where you can have whatever you want,” Musk said of the machine, which was mounted on a stand and cannot yet walk on its own. “It really is a fundamental transformation of civilization.” A second humanoid robot, described by Musk as for “rough development” and made from a mixture of Tesla and off-the-shelf parts, was able to walk forward—very unsteadily. The robots’ underwhelming debut and the contrast between Musk’s lofty rhetoric and the reality of Tesla’s humanoids reflect how, despite recent progress in artificial intelligence, it remains incredibly difficult for machines to operate in a sophisticated way in the messy, unpredictable real world. Robot industry veterans watched Musk’s grandiose sales pitch and still-clunky prototypes with a mix of interest and eye-rolling. “The idea that this will be doing anything useful in five years is laughable,” says Melonee Wise, CEO of Fetch Robotics, which makes robots that operate in warehouses alongside human workers. “There are a lot of obvious gaps.” Wise noticed that the joints on the Tesla-made robot look fairly rudimentary, appearing to be rigid and inflexible. The robot’s hands, driven by cables, seemed basic compared to those previously developed by roboticists. And the claim that Tesla was using Autopilot, its driving assistance system, to control the robot was perplexing, Wise said, given how different walking is from steering a vehicle. The problems that Musk says Tesla will soon conquer are the same problems that many roboticists have been working on for decades, Wise says. Yet Musk spoke casually of making an autonomous machine as physically capable as a person. “Do you know how hard that problem is?” Wise says. Musk first announced that Tesla would build a humanoid robot in August 2021, during a weird event that featured a human wearing a robot suit dancing around the stage. The project appears to be at least partly a recruiting pitch. Musk said ahead of last week’s event that it was intended to help recruit robotics and AI engineers—workers who might also contribute to Tesla’s work on autonomous driving. Musk rarely fails to win attention, but his robot livestream didn’t come close to rivaling viral videos from other companies working on humanoid robots. Boston Dynamics, now part of Hyundai, has got its humanoid Atlas doing parkour and dancing on a regular basis. Days ahead of Tesla’s event, the startup Agility released video of its two-legged robot bounding around a running track with an ostrich-like gait, covering 100 meters faster than any previous bipedal robot. Those robots are built on many years of research, but their physical prowess is also surprisingly limited. Legged robots’ viral moments take place in carefully limited situations, and often under remote human control. Confusion about what can be realistically expected of robots now and in the near future is understandable—perhaps even for billionaire technology magnates. Musk has previously admitted that deploying too much automation in some factories caused Tesla to miss production targets. Progress in AI presents something of a paradox when it comes to robotics. Computers can now do many things that were once impossible thanks to recent advances in AI, such as playing complex video games, modeling proteins, reliably transcribing speech, and generating artistic images from a text prompt. This has created expectations of a revolution in robotics, too. And yet going from the virtual world to the real one poses myriad challenges. AI can teach a robot to manipulate an object in simulation, for example, but once it tries to do it in the real world, or if the object or the setting changes, it can easily fail. Despite not having much progress to show yet, Tesla’s event did showcase the company’s commitment to working on advanced humanoid robots. Tesla researchers quickly ran through ongoing work in many areas that are key to building better robots, including actuators, perception, navigation, and simulation, where control strategies can be honed before deployment on a physical robot. Courtesy of Tesla Some robot experts watching saw a project that appeared to be quickly getting up to speed. “There’s nothing fundamentally groundbreaking, but they are doing cool stuff,” says Stefanie Tellex, an assistant professor at Brown University. Henrik Christensen, who researches robotics and AI at UC San Diego, calls Tesla’s homegrown humanoid “a good initial design,” but adds that the company hasn’t shown evidence it can perform basic navigation, grasping, or manipulation. Jessy Grizzle, a professor at the University of Michigan’s robotics lab who works on legged robots, said that although still early, Tesla’s project appeared to be progressing well. “To go from a man in a suit to real hardware in 13 months is pretty incredible,” he says. Grizzle says Tesla’s car-making experience and expertise in areas such as batteries and electric motors may help it advance robotic hardware. Musk claimed during the event that the robot would eventually cost around $20,000—an astonishing figure given the project’s ambition and significantly cheaper than any Tesla vehicle—but offered no timeframe for its launch.