By Becky Peterson
The future of Tesla is an army of humanoid robots that Elon Musk says could eliminate poverty and the need for work. He has told investors the robots could generate "infinite" revenue for Tesla and have potential to be "the biggest product of all time."
Musk has bet the company and his personal fortune on this vision of the world in which Optimus, as it is known, works in factories, handles domestic chores, performs surgeries and travels to Mars to help humans colonize the planet. Though today each robot is made by hand, Musk has proposed manufacturing millions of robots a year.
Optimus still has a lot to learn about the world before it is capable of replacing its human creators in the type of full-scale societal shift that Musk has in mind. In public appearances, the robot is often remotely operated by human engineers. On the engineering side, it has proven difficult for Tesla to create a hand for the bot with both the sensitivity and dexterity of a human. Inside Musk's companies, some employees have questioned the usefulness of the bots for routine business operations like manufacturing.
Musk is motivated to prove the skeptics wrong. His new compensation package gives him 10 years to make Tesla a $8.5 trillion company and sell at least one million bots to customers, among other product and financial goals. Success could mean earning Musk a $1 trillion pay package, and expanding Tesla far beyond the electric vehicle industry where it made its mark.
"The car is to Tesla what the book was to Amazon," Adam Jonas, an analyst with Morgan Stanley, said this summer. "Tesla used cars as a laboratory to get good at other things."
On Friday, Tesla reported its vehicle sales fell 16% in the fourth quarter and dropped 9% for all of 2025, leaving it behind China's BYD for the year. Tesla's share price, which tumbled in early 2025 as the company's EV sales slumped, had rebounded in recent months amid optimism in Musk's pivot to robotaxis and humanoid robots.
Optimus is still under development, but the bot has become a familiar sight at company events and for many employees. Inside Tesla's Palo Alto, Calif., engineering headquarters, robots routinely circle the inside perimeter of offices gathering information on how to navigate a room alongside humans.
In Tesla's labs, the nearly 6-foot-tall machine practices rote tasks like sorting Legos by color, folding laundry and using a drill to screw a fastener, former employees said. In October, the bot made its red carpet debut at the "Tron: Ares" premiere in Hollywood, performing a choreographed fight sequence with actor Jared Leto.
Tesla is one of several companies pushing the frontiers of robotics in the hopes of cornering a nascent market. A crowd of Silicon Valley startups like 1X and Figure, other manufacturers like Hyundai's Boston Dynamics, and Chinese robotics companies are eager to sell their robots that can fold laundry or manufacture vehicles.
Today, there are limits on how much robots can do. Many factories, including Tesla's, rely on robotic arms to do heavy lifting or dangerous tasks like moving hot metal. Those robots are largely stationary and programmed to do specific tasks. That leaves humans with jobs that require flexibility and precision, like installing cables or seats into cars moving across an assembly line.
Humanoids have an obvious appeal: bipedal, with flexible joints, robots like Optimus are designed to function more easily in spaces meant for humans.
Roboticists, however, have struggled to design robots with enough dexterity, sensitivity and adaptability to move around freely, said Ken Goldberg, a roboticist at the University of California, Berkeley.
"I've heard Elon Musk say hands are the hard part. It's true, but it's not only the hand -- it's the control, the ability to see the environment, to perceive it and then compensate for all this uncertainty. That's the research frontier," Goldberg said. "Getting these robots to do something useful is the problem."
Some Tesla analysts have struggled to price Tesla's opportunity with humanoids given how new the industry is and exclude it from their financial models. Even Tesla bull ARK Invest, which expects Tesla's share price to climb to $2,600 from around $400 today, left Optimus out of its model for 2029 because it doesn't expect the product to be commercially successful until later on.
"We believe initial versions of the robot will likely have a limited set of performable tasks," Tasha Keeney, a director at ARK Invest, said in an email. "Given Tesla's competitive advantages in embodied AI and manufacturing scale, we expect the company to be a formidable competitor in the space."
Morgan Stanley's Jonas, who now covers the robotics industry, predicts that by 2050, humanoids will bring in $7.5 trillion in annual revenue across the industry globally. Capturing even a fraction of that market could supersize Tesla's revenues, which came in at $98 billion in 2024.
At first it seemed like a joke. Musk unveiled Tesla's bot concept at an event in 2021 with a human dancing on stage dressed in a robot costume.
"It's intended to be friendly, of course, and navigate through a world built for humans and eliminate dangerous repetitive and boring tasks," he said at the time. When Musk returned to the stage a year later, he demoed a prototype called Bumblebee, with visible wires and actuators.
Behind the scenes, Tesla engineers were working out of a kitchenette on campus. Soon after, the expanding group of Optimus engineers moved to a basement, then a large parking lot in another building, one former employee said. Tesla struggled to find the right parts to build its robots, and had to make certain components like its actuators, which power the bot's movements, from scratch.
The big idea was to take Tesla's learnings from its self-driving technology, which uses software and cameras to autonomously drive automobiles. Musk told colleagues that its cars were just robots on wheels.
The humanoids would need to learn how to move around indoor spaces and avoid safety risks like tripping and falling on top of a nearby human or pet. To solve this problem, Tesla hired human data collectors to wear cameras and backpacks, and walk around collecting training data. Tesla had people collecting data in several shifts, running 24/7.
Another solution was to collect data using Optimus itself. The company set up bots to circle the inside perimeter of its offices learning how to navigate indoors. Sometimes the bot would fall over, after which an engineer would wheel over a robot hoist and pick the bot back up.
In October 2024, at a Warner Bros. sound stage in Burbank, Calif., Musk demonstrated his vision for cities replete with autonomous vehicles and robots.
In a disco-ball-decorated rotunda, five Optimus bots performed a dance routine to Haddaway's "What Is Love." Elsewhere on the lot, the bots served drinks while outfitted in cowboy hats and bow ties.
Behind the scenes, Tesla engineers worked overtime to troubleshoot technical issues, according to people on the ground. While robots in the rotunda were programmed to dance, other robots at the event were teleoperated by engineers who wore body suits and virtual-reality headsets. They guided the bots' interactions with guests, including while serving drinks behind the bar.
Each robot on the ground required constant monitoring from several engineers: one in a suit teleoperating its movements, one with a laptop, and others standing nearby to keep track of the bot's physical performance.
Inside Tesla's lab, Optimus proved pretty good at learning simple tasks. In May, the company shared a video that appeared to show Optimus performing various jobs in response to verbal orders from an engineer, such as putting trash in the bin, cleaning up crumbs, vacuuming, and moving a Model X part from a box. All of the activities were "learned direction from human videos," according to the company.
Despite this progress, inside the company, some manufacturing engineers said they questioned whether Optimus would actually be useful in factories. While the bot proved capable at monotonous tasks like sorting objects, the former engineers said they thought most factory jobs are better off being done by robots with shapes designed for the specific task.
The big question, according to Goldberg, the Berkeley roboticist, is how to give robots the dexterity of humans and the ability to understand their environment well enough to complete useful but sensitive tasks, such as clearing a dinner table. "Even a child could clear a dinner table," said Goldberg, who is also chief scientist at Ambi Robotics and Jacobi Robotics.
Some of Tesla's competitors have concluded that legs are the problem. Evan Beard is chief executive of Standard Bots, which sells manufacturing robots on wheels. Beard said that wheels make the bots more stable, and therefore safer to work around, and easier to power down if something goes wrong.
"With a humanoid, if you cut the power, it's inherently unstable so it can fall on someone," Beard said. "For a factory, a warehouse or agriculture," he said, "legs are often inferior to wheels."
Tesla has backed away from its initial Optimus timeline of putting a commercial version to work into its own factories by the end of the year. The company is currently working on its third generation of the robot.
In Tesla marketing materials, Optimus has a role as a domestic worker watering plants, unpacking groceries and handling other household tasks, giving its owners time to hang out with their families.
"Who wouldn't want their own personal C-3PO/R2-D2?" Musk said in November, referencing the droid characters in "Star Wars" movies. "This is why I say humanoid robots will be the biggest product ever. Because everyone is gonna want one, or more than one."
Write to Becky Peterson at becky.peterson@wsj.com
(END) Dow Jones Newswires
January 02, 2026 21:00 ET (02:00 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
