AI-powered humanoid robots could take over large sections of factory work within the next five to 10 years, transforming the manufacturing industry, predicts Arm CEO Rene Haas.
One of the key forces pushing humanoid robots into factories is their advantage over the robotic arms and other automation machinery in use today, Haas said. Traditional factory robots are purpose-built machines designed for a single task, with both hardware and software optimized for that specific function. General purpose humanoid robots by contrast, combined with increasingly sophisticated “physical AI” that helps navigate the real world, will be able to take on different jobs on the fly with quick modifications to their instructions.
“I think in the next five years, you’re going to see large sections of factory work replaced by robots—and part of the reason for that is that these physical AI robots can be reprogrammed into different tasks,” Haas said at Fortune Brainstorm AI in San Francisco on Monday.
“One of the issues you’d had with factory robots in the past is that if it was a pick and place machine for a factory, they’re just optimized for one task—the software was for one task, the hardware is for one task. Now, if you design a general-purpose humanoid that the software is all AI and it learns by doing, it’s going to completely replace a large set of factory workers,” he said.
What happens to those workers and the broader job market as AI and robots proliferate in businesses is a growing concern among many policymakers and industry observers, with ideas ranging from worker re-skilling to universal basic income among the options under debate.
Haas did not specifically address the jobs issue, but suggested that widespread physical AI adoption could reshape global manufacturing dynamics, potentially helping to level the global competitive playing field by automating a large amount of factory work. “Physical AI will be a great enabler,” he said.
Haas also pointed to Waymo’s autonomous vehicles as an early indicator of physical AI’s potential.
He said the next generation of autonomous systems may require even less hardware. While current self-driving cars are fitted with radar and cameras surveying their surroundings, future iterations using more advanced AI models could operate with fewer sensors—relying on artificial intelligence rather than exhaustive data collection to make decisions.
The semiconductor supply chain has ‘many single points of failure’
Arm, which does not manufacture or sell its own chips, designs and licenses the architecture used in processors made by companies including Qualcomm and Apple. Chips based on Arm’s designs are used in everything from smartphones and refrigerators to cars and servers, and most people use between 50 to 100 Arm chips on their person or in their homes, Haas said.
That widespread use and market share is a testament to the energy efficiency and performance that have made Arm’s chip design so popular. But it also raises risks to the semiconductor supply chain.
Asked about this vulnerability, Haas acknowledged the extreme market concentration within the industry, and noted that several large companies each control vital parts of the semiconductor supply chain: “The semiconductor supply chain has many single points of failure. There’s TSMC, which is in a very obviously interesting part of the world geopolitically. There is also a very sophisticated device that has to go into these fabs that comes from one company on the planet … called ASML.”
In the past few years, the COVID-19 pandemic exposed some of these supply-chain fragilities when chip shortages left consumers unable to get key fobs for new cars for weeks. That crisis, Haas said, was “just a function of the semiconductor supply chain that has many single points of failure.”
Haas said the entire industry is “learning to live with“ the concentration of risk.
Read more from Fortune Brainstorm AI:
