As investors funnel billions into humanoid robots, a three-decade veteran MIT roboticist argues they’re squandering their funds.
Rodney Brooks—co-founder of Roomba maker iRobot—stated that the concept of humanoid robots as all-purpose helpers is “pure fantasy thinking,” partly due to robots’ struggles with coordination.
“Even with hundreds of millions (or potentially billions) of dollars from venture capitalists and big tech firms funding their training, today’s humanoid robots won’t develop dexterity,” Brooks noted in a .
Touch is among the human body’s most intricate systems. A human hand has 17,000 low-threshold mechanoreceptors for detecting light touches, with these receptors growing denser near the fingertips. These hand receptors react to countless stimuli—such as pressure and vibration—by syncing with 15 distinct neuron families. This all forms a complex system that humans aim to replicate in robots.
Although AI has been trained on and image-processing data, “we lack a similar body of touch data,” Brooks explained, noting he disagrees with how both Musk’s and AI-robotics firm Figure are training their humanoids—with , assuming this will lead to major dexterity improvements.
“If the big tech companies and VCs pouring money into large-scale humanoid training allocated just 20% of that sum to university researchers instead, I believe they’d reach their goals faster,” Brooks stated.
Notably, Brooks’ former company iRobot in December after pulled out of a 2024 plan to acquire it. Its value dropped from $3.56 billion in 2021 to roughly $140 million last year. The firm will be bought by its primary Chinese manufacturer and lender, though it has stated the restructuring won’t impact its current products.
Musk, for his part, told the World Economic Forum in Davos last month that Tesla will begin selling its Optimus robots to the public by late 2027. He has previously claimed Optimus already performs tasks autonomously in Tesla’s factories. Meanwhile, Figure secured a in September 2025 following its latest billion-dollar funding round. But according to Brooks, all this investment amounts to a very costly training program for humanoid robots that won’t look exactly like humans.
Brooks predicts that successful robots in 15 years will bear no resemblance to humans—featuring wheels, multiple arms, and possibly five-fingered hands, though they’ll still be labeled “humanoid robots.” However, today’s attempts will mostly be consigned to history.
“A great deal of money will vanish, spent on trying to extract any level of performance from today’s humanoid robots. But those robots will be long gone and mostly conveniently forgotten,” he added.
A version of this story first appeared on on September 29, 2025.
More on robotics:
- Uber’s CEO forecasts most rides could be within two decades
- A man accidentally accessed thousands of robot vacuums, revealing the AI cyber threat risk
- Elon Musk cautions the U.S. is ‘’ unless AI and robotics rescue the economy from crippling debt