Elon Musk wants you to feel “Optimustic” about a future full of robots that look like humans.
The Tesla chief executive recently released a video of himself dancing alongside Optimus, a robot prototype being developed by the US$1.3-trillion company.
But not everyone is convinced robots should look like humans.
In January, the Wall Street Journal reported that some engineers working on Tesla’s Optimus have questioned the usefulness of designing human-shaped robots.
“With a humanoid, if you cut the power, it’s inherently unstable so it can fall on someone,” Evan Beard, the CEO of American robotic arm company Standard Bots, told The Wall Street Journal. “For a factory, a warehouse or agriculture … legs are often inferior to wheels.”
Sources say the rationale for creating humanoid robots can have more to do with marketing aims than factory solutions.
“Nobody, other than to make marketing videos, is generating anything of value with their humanoids,” Brandon Dehart, manager of the RoboHub robotics lab at the University of Waterloo, told Canadian Affairs. “They are all marketing.”
Legs are a liability
For decades, industrial robots have not been designed to look like humans.
Most robots used in Canadian factories today are fixed or modular systems that perform a single task repeatedly, such as a mechanical arm that lifts parts, welds a seam or sorts items.
Even automated guided vehicles and autonomous mobile robots typically follow fixed paths, reflecting designers’ prioritization of stability, speed and safety.
More recently, some manufacturers have started using collaborative robots, or “cobots,” which can operate safely alongside human workers. But even these robots are built to fit into a factory’s layout rather than navigate it.
“Most of our manufacturing workers aren’t walking around all day,” said Beard, in an interview with Canadian Affairs. “We design the factory so [workers] have their incoming parts within arm’s reach.”
Tesla envisions its Optimus robots performing “unsafe, repetitive or boring tasks,” the company’s mission statement says. The company has marketed its robots for factories and logistics, but on a Jan. 28 earnings call, Musk said Optimus is still in R&D and “not in usage in our factories in a material way.”
The robots are targeted to be available by late 2027 and expected to cost US$20,000–$30,000.
Beard and others say Tesla’s decision to put legs on its robots adds unnecessary risk, cost and inefficiency.
“More than half of the cost is going to be the legs,” said Beard, about the Optimus robots. “If you don’t need mobility … it doesn’t really make sense.”
Beard has seen this firsthand. Standard Bots built a humanoid prototype before abandoning it.
“With a humanoid, every part throughout the arm has to be unique,” Beard said. “[That means] lower volume, it’s more complex, it’s harder to assemble, you need more spare parts to repair it, and because you’re trying to match the human form factor, it’s slower and can lift less weight.”
“The customer’s [return on investment] is worse … more expensive … lower speed, with less weight,” he added. “It’s almost worse on every metric.”
Modular, AI-driven robotic arms on wheeled bases are often more practical.
“If you have a flat floor and you’re indoors, a wheeled system with an arm or two, and sensors up high … is always going to make more sense,” said Dehart, of the University of Waterloo.
“In a hospital, you can usually get rid of the legs because stairs don’t matter.”
The guiding principle, Dehart says, is matching a robot’s form to its environment and task. “The environment isn’t being modified to suit the robot,” he said.
This month, at the World Economic Forum in Davos, Musk suggested Optimus could also be used within homes.
“Who wouldn’t want a robot to — assuming it’s very safe — watch over your kids, take care of your pets. If you have elderly parents — a lot of friends of mine have said that for elderly parents, it’s very difficult to take care of them,” he said.
But sources said humanoids would also face serious safety challenges in this context.
“For humanoids to work at a home, you almost have to pass … the Home Alone eval[uation],” Beard said, referring to the booby-trap gauntlet of toy cars, icy steps and swinging paint cans in the Macaulay Culkin holiday classic.
“That’s really hard, from a software perspective, to make them safe enough where, even if it’s going up the stairs, if two eight-year-olds jump on its neck, it doesn’t flip back and crush them with 200 pounds of metal.”
Tesla did not respond to multiple requests for comment.
Why humanoids?
Tesla is not the only company marketing humanoid robots for a wide range of applications.
Boston Dynamics, a global robotics company now majority owned by Hyundai, positions its Atlas humanoid robots as an industrial robot capable of autonomously learning tasks, handling materials and integrating itself within factory workflows.
In response to questions, Boston Dynamics’ marketing agency directed Canadian Affairs to an interview featuring Aya Durbin, the company’s product lead for humanoid applications.
Durbin says Atlas robots are humanoid rather than task-specific so they can replace humans in factories. “[Atlas is] a robot that could drop into any existing workplace and meet the functional expectations you naturally have of a person,” she said.
In Canada, Vancouver-based Sanctuary AI also promotes its general-purpose, humanoid robot Phoenix for sectors including automotive, manufacturing and logistics.
Isabella Oidem, Sanctuary AI’s communications representative, pointed Canadian Affairs to a blog post stating Phoenix’s humanoid design aims to make it “physically capable of doing most of the world’s work,” since “the world is … designed for people.”
The post emphasizes human-like hands and bipedal mobility but does not specify which industrial or logistics tasks actually benefit from a humanoid form.
Beard and Dehart say humanoid designs continue to be developed because they are intuitive, cinematic and easy to understand.
“It’s more likely [robotics companies] are pursuing [humanoids] because they saw that investors got excited about it,” said Beard. “Often you see technology chase these waves of investment, and I think it’s at a disservice.”
Dehart says humanoids also serve to showcase a company’s capabilities. “[A humanoid] is the perfect platform to demonstrate your ability to build any of the subsystems [of a robot],” he said.
And even Beard notes humanoid robots may have useful applications in military or security contexts, where uneven terrain and obstacles may favour legs over wheels.
“You can’t beat legs in a cluttered environment,” he said, pointing to scenarios such as navigating forests, climbing over debris or recovering from falls.
“But a lot of what we’re talking about now is a factory and a warehouse and a field for agriculture, and these environments are not cluttered like that.”
Even for these applications, Dehart is skeptical.
“If you want to carry stuff through the woods, a little tank or a quadruped is always going to be better,” he said.
Form follows function
Rejecting humanoid robots, sources say, does not mean rejecting innovation. It means designing machines around tasks, tools and environments, not anatomy.
“Once a robot gets good at doing something, we don’t call it a robot anymore,” said Dehart, pointing to examples like autonomous vehicles or industrial machines that fade into the background once they work reliably.
Beard said the real disagreement in robotics is not about whether automation will improve lives, but whether companies are deploying capital efficiently.
“Are we collectively spending our venture capital dollars in the most efficient way to get there?”
That question matters for Canada, where robotics adoption lags global leaders.
Canada ranks 15th worldwide in industrial robot adoption, according to the International Federation of Robotics, with most installations concentrated in automotive manufacturing.
“If you don’t include industrial manipulators that just move car parts around, we drop down,” Dehart said.
Both Beard and Dehart say Canada and the U.S. cannot afford hype-driven robotics development at a time when competitors like China are rapidly scaling practical automation.
“We need to get serious about adopting these tools if we want to be competitive,” said Beard.
