A large amount of capital has flowed into humanoid robot startups, with investors betting that humanoid robots will soon enter warehouses, factories, and even millions of households.
However, even the humanoid robot companies themselves believe that the technology has been overhyped.
Even the founders of companies whose robotic products have begun to find a foothold in the market believe that promoting the notion that 'the technology is already mature' carries risks.
Since the beginning of this year, the global humanoid robot market has shown rapid development, with related concepts receiving enthusiastic support in stock markets across multiple countries. A large amount of capital has flowed into humanoid robot startups, with investors betting that humanoid robots will soon enter warehouses, factories, and even millions of households.
However, there has been an ongoing debate about a “bubble” surrounding humanoid robots. Some experts warn that the technology is far from ready for scaled application, and even some humanoid robot companies themselves believe the technology has been overhyped.
Executives at many of these companies hope to temper external expectations. They state that despite recent advances in the field, the hype around humanoid robots has become excessive. Before moving from scientific experiments to replacing human labor, they face numerous significant technical challenges.
“What we’ve been exploring is not just how to build humanoid robots, but how to create ones that can be genuinely useful,” said Pras Velagapudi, Chief Technology Officer of Agility Robotics, a private American humanoid robot company.
Agility Robotics focuses on industrial automation scenarios, with the bipedal robot Digit as its core product, which has achieved commercial deployment ahead of others.
Agility currently has hundreds of Digit robots working for clients, including Amazon and automotive parts manufacturer Schaeffler. These robots handle tasks such as picking up items and transporting them within warehouses.
As robots like Digit begin to find specific demand scenarios, some analysts and tech executives are starting to optimistically predict that the wave of humanoid robots is imminent.
Velagapudi expressed skepticism about this. He stated that sending humanoid robots into warehouses or industrial settings to move boxes is one thing, but creating a “robotic housekeeper” far exceeds the current technological capabilities of the industry. Current robots are too unreliable to handle complex tasks.
Another issue is safety. Ani Kelkar, a partner at McKinsey, noted that a survey of executives revealed installation costs are the biggest reason companies are reluctant to deploy robots. Currently, for every $100 companies spend on deploying robots, only about $20 goes toward the actual machine cost, with the remainder allocated to equipment and systems designed to protect humans from harm.
In theory, humanoid robots do not require the complex safety mechanisms that industrial robotic arms weighing thousands of pounds and operating at high speeds necessitate. Tesla's Optimus robot stands approximately five feet eight inches tall and weighs 125 pounds.
However, Kelkar noted that there remains a significant gap between the technological vision and current practical capabilities. "When we see videos of robots doing laundry, we infer that there will be a robot butler capable of handling everything in our homes, which is an enormous leap," he remarked.
At a recent humanoid robotics summit held in Mountain View, California, Isaac Qureshi, CEO of the American startup Gatlin Robotics, wore a virtual reality headset to control an early prototype robot designed for cleaning office spaces. The event was touted as the largest global gathering focused on humanoid robotics.
Qureshi followed the robot as it scrubbed a brick wall. He stated that they would gradually teach the robot to perform more tasks, with toilet cleaning being one of the ultimate goals.
On stage at the summit, one founder after another from various humanoid robotics startups attempted to temper expectations for this technology.
Kaan Dogrusoz, CEO of the American general-purpose humanoid robotics venture Weave Robotics and a former Apple engineer, stated, "There is indeed remarkable technical work being done in this area, and a large pool of talented individuals are contributing, but these robots are still far from being clearly defined, mature products."
Dogrusoz explained that while the direction of humanoid robotics development is correct, the technical capabilities are insufficient to support such grand ambitions. He compared it to Apple's most famous product failure – the Newton handheld computer.
The Apple Newton was a portable personal digital assistant (PDA) introduced by Apple in 1993. It became a landmark attempt in early smart handheld devices in the consumer electronics sector but ultimately failed due to immature technology and ambiguous positioning, making it one of Apple's well-known 'failures.'
This product, launched during the PDA hype wave of the 1990s, was discontinued after just a few years due to commercial failure. However, only a decade later, handheld devices became ubiquitous with the advent of the iPhone.
Dogrusoz remarked, "Fully bipedal humanoid robots are the Newton of our era."
Weave manufactures robots that fold clothes, and some laundromats in San Francisco are already using these robots. However, even founders whose own robotic products have begun to find a market believe that promoting the idea of 'technology being fully mature' poses risks.
Several company executives stated that the range of positions truly suitable for humanoid robots is currently very limited, including performing simple repetitive tasks such as moving boxes.
Nicolaus Radford, CEO of Persona AI, an American industrial humanoid robotics startup, said during his keynote speech at the summit: 'I think we must take responsibility regarding the timelines and adoption rates we discuss.'
Persona is developing a welding robot for a shipbuilding company. Radford noted that such work, due to its high risk and recruitment challenges, is highly suitable for automation. However, he added that applications like robot butlers remain quite distant from the market.
In stark contrast to the optimists
The cautious stance of executives and engineers in the humanoid robotics industry contrasts sharply with the optimistic projections made by several tech giants.
For instance,$Tesla (TSLA.US)$CEO Elon Musk predicted that the demand for humanoid robots would be 'endless,' stating that Tesla aims to produce one million Optimus robots annually by 2030.
$NVIDIA (NVDA.US)$CEO Jensen Huang stated that he believes the world is at a tipping point for 'robotizing everything that moves.' 'Humanoid robots and the technologies enabling them are within reach,' Huang said on a podcast in January this year.
Kelkar, a partner at McKinsey, stated that dozens of robotics startups are attracting significant investment, with approximately $5 billion flowing into the humanoid robotics sector this year.
However, Rodney Brooks, a world-renowned robotics expert, computer scientist, and artificial intelligence researcher, has warned that the humanoid robotics bubble is bound to burst, and investing billions in humanoid robot startups is a waste of money.
On one hand, he pointed out that human hands are highly complex, containing approximately 17,000 specialized tactile receptors, which robots cannot match. On the other hand, he highlighted safety concerns, stating that once a robot falls, it can become very dangerous.
Brooks also predicted that within 15 years, successful 'human-like' robots would actually be equipped with wheels, multiple arms, and specialized sensors, and would abandon human-like appearances.
Editor/melody